Idealism and the Harmony of Thought and Reality 0198823630, 9780198823636

Citation preview

Idealism and the Harmony of Thought and Reality

Idealism and the Harmony of Thought and Reality T HOM A S HO F W E B E R

Great Clarendon Street, Oxford, OX2 6DP, United Kingdom Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries © Thomas Hofweber 2023 The moral rights of the author have been asserted First Edition published in 2023 Impression: 1 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by licence or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this work in any other form and you must impose this same condition on any acquirer Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016, United States of America British Library Cataloguing in Publication Data Data available Library of Congress Control Number: 2022945555 ISBN 978–0–19–882363–6 DOI: 10.1093/oso/9780198823636.001.0001 Printed and bound by CPI Group (UK) Ltd, Croydon, CR0 4YY Links to third party websites are provided by Oxford in good faith and for information only. Oxford disclaims any responsibility for the materials contained in any third party website referenced in this work.

für meine Mutter Christiane Hofweber und zum Andenken an meinen Vater Josef Hofweber

Contents Preface

ix

1. Idealism and our place in the world 1.1 The Big Question 1.2 Idealism: the main idea 1.3 Idealism: some constraints 1.4 Idealism: the main options

1 1 4 8 12

2. Forms of idealism we should reject 2.1 Phenomenalism 2.2 Conventionalism 2.3 The subjectivity thesis 2.4 Fragmentalism 2.5 Idealist theories of properties and truth

19 22 31 38 44 49

3. Conceptual idealism: the basic idea 3.1 Idealism via harmony 3.2 Harmony via internalism 3.3 How the argument is possible 3.4 But is it really idealism? 3.5 Filling in the picture

61 62 67 78 81 88

4. The internalist conception of facts 4.1 Fact-terms 4.2 Quantification 4.3 Context-sensitive quantifier inferences 4.4 Quantification and modality 4.5 The internalist position

91 93 96 100 106 108

5. Harmony and ineffability 5.1 From structural harmony to complete harmony? 5.2 The concept of the ineffable 5.3 Objects as a source of ineffability 5.4 The argument for complete harmony 5.5 The consequences of complete harmony 5.6 Constraining logical space

111 111 115 119 124 126 133

6. Inescapable concepts 6.1 The language-metaphysics gap 6.2 The idea of inescapable concepts 6.3 The significance of inescapable concepts

139 139 147 157

viii contents 6.4 The inescapability of the propositional 6.5 Conceptual idealism as a deep result

162 168

7. The immanent stance 7.1 Rational traps 7.2 Immanent rationality and the asymmetry thesis 7.3 The immanent and the transcendent stance 7.4 Eklund on ought 7.5 Immanent metaphysics

171 171 179 189 192 195

8. Conceptual idealism: the overall picture 8.1 The Big Question revisited 8.2 How the constraints are met 8.3 The metaphysics of conceptual idealism 8.4 The meta-metaphysics of conceptual idealism 8.5 Onward

203 203 208 211 215 222

Bibliography Index

227 235

Preface This book admittedly defends idealism, but it’s not as bad as that sounds. I know it sounds bad. Idealism seems to go against everything we have learned over the last couple of centuries about the world and the place of minds in it. Minds are puzzling, to be sure, but nowadays the real puzzle is taken to be how the mental fits into the material world, which is to say, how to naturalize the mental. There is no larger place for minds in metaphysics, and there especially is no place for anything like idealism. ‘Idealism’ is thus a bad word these days in large parts of the Englishspeaking philosophical scene. It is the kind of view that people defended in the history of philosophy, before we knew better, a view that is now merely of historical significance. And there undoubtedly is some justification for this negative attitude towards idealism. It is often associated with positions that are rightly rejected, like the view that there is no world outside of our minds. Furthermore, the writings of contemporary self-declared idealists can sometimes be rather hard to understand, and rely heavily on historical figures, who in turn can be hard to understand. No wonder idealism isn’t exactly popular any more. But idealism can do so much better than that. There are a number of intriguing, although often neglected, options for defending idealism that make perfect sense, don’t rely on historical figures, aren’t in conflict with what we have learned about the world, and are supported with intriguing arguments. All of them should be seen as real contenders in contemporary metaphysics. I will critically discuss several such options below, and add one more to the list that I believe is the correct one, which I will label ‘conceptual idealism’. Although I will argue that the other forms of idealism to be discussed face a series of problems that give us good reason to reject them in the end, they should nonetheless all be taken seriously. Conceptual idealism overcomes these problems, and it is not just a contender, but the version of idealism we should accept. To defend this view and highlight its significance is the main goal of this book. Conceptual idealism is an especially radical version of idealism. It is not just a form of idealism in general, which holds that minds are metaphysically central in the world, but a stronger position, which holds that our human minds in particular are metaphysically central in the world. The ultimate target of this book is thus the metaphysical status of human beings and our place in reality as a whole. Idealism alone sounds bad already, but a claim of the metaphysical centrality of human beings takes it a step further. I hope to make clear in the chapters to come that this is also not as bad as it sounds. It does not mean that our species is disconnected from other species, nor that earth is the center of the universe, nor that human

x preface beings are morally in a class of their own. But it does mean that human beings have a special place in metaphysics, not just as it concerns our local environment, but as it concerns reality as a whole. All that might sound bad enough, but it gets worse. I will argue that we can see the special metaphysical status of human beings in reality as a whole simply by reflecting on our own representations of reality. We can see that our human minds and reality are in a kind of harmony that supports our special metaphysical status simply by thinking about how we represent reality, with no regard to whether these representations are true. Simply by looking inwards, at our own concepts and our own language, without considering what reality is otherwise like, can we establish that a harmony between our minds and all of reality has to obtain. This way of arguing for idealism is an example of a particular approach to making progress in metaphysics, which I will develop below and label ‘immanent metaphysics’. Immanent metaphysics takes seriously that we start out with some questions we hope to answer, and that such questions are stated by us, using our concepts and our language. It contrasts with approaches that focus simply on giving a description of reality in some language that the world itself recommends, as if the world had already asked a question for us to answer. The book will conclude with some reflections on what we can hope to achieve with immanent metaphysics, how it relates to the rest of metaphysics, why it deserves to be called neo-Kantian, and how we should understand conceptual idealism as a metaphysical position. Let me briefly say how we get there by outlining the chapters to come. Chapter 1 starts with the question of the metaphysical place of human beings in the world, and how a certain strong form of idealism could answer it. Such an answer would add a further answer to the mix, besides the familiar theistic answer that we have a special place due to our connection to a divine being, and the familiar naturalistic answer that we have no special place at all. I outline some basic options for formulating and defending idealism, and some minimal constraints that any reasonable defense of idealism must meet. Chapter 2 considers a series of attempts to defend idealism, all of which I argue should be rejected. Nonetheless, they include several intriguing and neglected options, besides the more familiar phenomenalistic form of idealism. This chapter hopes to make clear just how varied, interesting, and promising idealism can be. It also hopes to make clear that any defense of idealism faces certain problems, which are serious enough to sink the so-far discussed versions of idealism. Any new proposal will need to do better to overcome these problems. Chapter 3 presents the basic idea behind conceptual idealism, the form of idealism I defend in this book. I propose that we can formulate idealism via the notion of a harmony between our minds and reality, and that we can establish that this harmony obtains with arguments from the philosophy of language. We will discuss how considerations from the philosophy of language could possibly support metaphysical positions, and what status our own questions have in

preface xi metaphysics. This chapter outlines the main ideas of this book, which are then developed and defended more carefully in the chapters to come. If you only have time to read one chapter, I’d say pick this one. Chapter 4 clarifies and defends the position in the philosophy of language which implies that harmony obtains, and with it that idealism holds. I will formulate and defend a position labeled ‘internalism’, which concerns how we talk about facts with that-clauses and how to understand quantification over facts. Chapter 5 strengthens the version of harmony that can be established via internalism, and with it the proposed version of idealism. This proceeds via a clarification of the notion of an ineffable fact. In particular, I will strengthen the argument from Chapter 3 in favor of structural harmony to an argument for the noticeably stronger and more radical thesis of complete harmony. Chapter 6 concerns the general problem of achieving results about the world via considerations of our own representations of the world, and presents an argument that this is indeed possible. What’s more, this chapter defends that not only can we answer questions about reality by thinking about our representations, in certain special cases we can even find out that we asked the right question to begin with. And we can do all that simply by thinking about our own representations alone, without antecedently knowing what reality is like. I will argue that just this is the case with the question about our place in reality and with the argument for conceptual idealism. A key idea here will be to consider a special class of our concepts: inescapable concepts, which are concepts that one cannot rationally replace with other ones. Chapter 7 continues the discussion of the possibility of finding out that one asked, and answered, the right question simply from considerations about our own representations alone. It dives deeper into the issue of what follows from the fact that one can’t rationally accept that there is a better question than the one that was already asked. We will consider whether it is coherent that rationality might trap us into having to think this, even if it is false. And we will discuss two general approaches to the status of an initial question in metaphysics, and how to improve from there. The chapter defends what I label the ‘immanent stance’, which takes our own starting point seriously, and contrasts it with the transcendent stance, which disregards our starting point, and focuses on the alleged demands from the world for a particular description. Chapter 8 finally puts everything together. It argues that conceptual idealism establishes the metaphysical centrality of human beings in reality, it shows how conceptual idealism overcomes the problems that refuted other versions of idealism, it outlines the metaphysical picture tied to conceptual idealism, it presents the strategy of making progress in metaphysics via immanent metaphysics, it elaborates on the comparison with Kant’s philosophy, and it looks ahead at what else one might hope to achieve this way. I hope that by this point idealism won’t sound all that bad any more.

xii preface I am indebted to many people for their help with this project. In addition to all those mentioned below, I profited from discussions with at least Ralf Bader, Jc Beall, Paul Boghossian, Tim Button, Dave Chalmers, Aaron Cotnoir, Paul Egré, Dave Estlund, Hartry Field, Michael Friedman, Anja Jauernig, Toni Koch, Matt Kotzen, Martin Lipman, Alan Nelson, Manish Oza, Kenny Pearce, Jeff Pelletier, Tobias Rosefeldt, Thomas Sattig, Kevin Scharp, Ulrich Schlösser, Nick Stang, Rob Trueman, Gabriel Uzquiano, and Helen Yetter-Chappell. My thanks to my since dispersed local idealism support group at UNC, which was very supportive during the time I came to terms with being an idealist, and very helpful in figuring out what form that idealism should take. It consisted of various idealists and idealism sympathizers, including Bob Adams, Krasi Filcheva, Bill Lycan, who co-taught a graduate seminar on contemporary versions of idealism with me in the spring of 2013, Rob Smithson, and Craig Warmke. I was fortunate to get very helpful comments on individual chapters from Matti Eklund, Markus Kohl, Marc Lange, Ram Neta, Mike Pelczar, Jim Pryor, Ted Sider, and Alex Worsnip. An earlier version of the manuscript was discussed in a graduate seminar at UNC in the fall of 2020. For critique and many good suggestions for improvement I am indebted to Cal Fawell, Genae Matthews, Conner Schultz, Alyse Spiehler, Ripley Stroud, Aaron Thieme, Zach Thornton, Nolan Whitaker, and Aurora Yu. Material from this book was presented at Berkeley, FU Berlin, Brown, Cambridge, Cornell, Düsseldorf, Jerusalem, Johns Hopkins, Madrid, Minnesota, Munich, Northwestern, NYU, Ohio State, Oxford, Rice, Simon Frasier, St. Andrews, Stockholm, Syracuse, Tel Hai, Toronto, Tübingen, UConn, UNC, Uppsala, and Zurich. Thank you to all those who came to the talks and especially those who stayed to participate in the many good discussions that followed. My thanks to two referees for OUP for their detailed and helpful comments on an earlier draft, and to Peter Momtchiloff for his support and guidance. This book was written using TexShop, BibDesk, and LATEX, all freely available and terrific. I am indebted to those who donated their time and skill to the development of these tools. I would also like to gratefully acknowledge a WN Reynolds Senior Faculty Research and Scholarly Leave during the spring of 2021, which gave me time to work on this book. Like every other contemporary book, much of this book was written during the Covid-19 pandemic, usually at home, in the kitchen or weather permitting on the porch, wishing for a home office. I greatly profited from my Sony WH1000XM3 noise canceling headphones, which made progress possible despite the best efforts to defeat any peace and quiet by our kids, cats, dog, and roosters. My thanks to Rebecca Walker for sharing the kitchen table with me while working on her own book.

preface xiii I dedicate this book to my parents, one living and one dead, in gratitude for all their love, the band-aids, and the breakfasts I didn’t eat, since I got up too late and had to run to catch the train to school instead. They weren’t thrilled by my going into philosophy and on top of that my moving to the United States to do so, but they were supportive nonetheless, then and during all the years that came before and after.

1 Idealism and our place in the world 1.1 The Big Question What is the place and significance of human beings in the world overall? Are we special or merely more of the same? Do human beings have a distinguished standing in reality overall? These are natural and pressing questions, in particular for us human beings. But as they are stated they are neither very precise nor do they get to the heart of the matter. Human beings might well be special in a number of ways, but that by itself doesn’t answer the question as it is intended. We might be the best at music, say, not just on earth, but anywhere. But even if no better music is ever made anywhere, that by itself does not make us special in the right way. By itself it wouldn’t be much different than being the largest volcano. The largest volcano is special, too, but again not in the right way. To put it metaphorically, the universe might not care about volcano size, and it similarly might not care about music. If we are special in a way that matters, then just being the best or worst at something isn’t enough, even though it does make us special in a particular way. The real issue is something else, something grander, something about how we fit into the world as a whole in large-scale ways. But how to state that less elusively is not so clear. To approach it metaphorically once more: when the overall story of reality is written, will we appear in the main text, or are we merely mentioned in a footnote, if at all? Being the best at music will at best put us into a footnote, unless, of course, music is really important for the larger overall story of reality. To appear in the main text, we would have to be important for reality overall. We would have to somehow be central for the grand or most general aspects of the world, not just for some local events or some particular features. To put a label on it that at least points in a certain direction, we would have to be metaphysically central to reality to be in the main text and to be significant in the intended way. This still is not especially precise, but it’s a start. Let us call the Big Question about our place in the world the question that we intend to ask when we ask about whether we are special while reflecting on the significance of human beings in reality. How to articulate the Big Question more explicitly is unclear. Metaphorical attempts can help point in a certain direction, but by themselves they are not enough. To make progress we can try to approach the question via its two most prominent answers. These answers seem to answer the question as it is intended, and this way the answers can shed light on the

Idealism and the Harmony of Thought and Reality. Thomas Hofweber, Oxford University Press. © Thomas Hofweber 2023. DOI: 10.1093/oso/9780198823636.003.0001

2 idealism and the harmony of thought and reality question itself. One of these two answers holds that we are not significant, the other one holds that we are. The first is the naturalistic answer. It holds that we are nothing but a complex and accidental arrangement of the same matter that is found everywhere else as well. That matter ever arranged itself in this way didn’t have to happen. It is at best a lucky bonus to reality, but not a central part of it. If matter hadn’t arranged itself the way it did when it formed us, then nothing of global significance would have been lost: the world would otherwise have been just as it is, except for a local difference on one small planet. But overall, from the point of view of the universe, nothing much would be different, and so we are not globally significant. Our influence and significance in the world is local, not global. There is no special place for us in the story of reality as a whole. The second answer to the Big Question is the theistic answer, which says that we are significant in the intended way. We have a special and distinguished place in the world, since we have a special relationship to God, one that no other creature has. We are central to the world, since we are in good part the reason why there is a material world in the first place. God created the material world for us and with us in mind. Thus via our relationship to God we have a central place in the material world, and with it in all of reality as well. Both of these answers to the Big Question are well known and widely debated. Naturally, much of the debate between these two answers centers on whether there is a God who has a special relationship to human beings in the first place. But besides the debate about God, there are also other arguments which are widely taken to support the answer that we are not special. Chief among them are the arguments against the geocentric worldview and that earth is not in some objectively distinguished spatial location, and the arguments from evolution that we are connected to other species. Here it is often said that Copernicus and Darwin undermined our belief in our significance in the universe. But those arguments miss their target, and are at best indirect, since the spatial location of human beings wasn’t really at issue when we wondered whether we are special. The issue is not whether we are spatially special, but whether we are special in other, more elusive ways. Being connected to God, and to the reason why there is a material world at all, certainly would make us special in the intended way. But without God this line of reasoning will go nowhere, and thus naturally the debate about our place is focused on the existence and nature of God. But there is also a third answer to the Big Question. This answer used to be more popular in the history of philosophy, but it has nowadays almost completely disappeared from serious consideration. On this answer we are central to reality, not via a connection to God, but more directly: we are central to reality, since there is an intimate relationship between our minds and reality itself. We are somehow centrally tied to reality, since our minds and reality are inextricably linked. This third answer to the Big Question is thus an idealist answer: it takes inspiration

idealism and our place in the world 3 from idealism, roughly the position that minds play a central metaphysical role in reality. Idealism in some form or other was widely defended in the history of philosophy, but it is nowadays almost universally considered a non-starter. Idealism is generally taken to face a series of unsurmountable problems: it is unclear how this intimate connection between minds and reality is supposed to be understood, it is unclear what good reason one might have for there being such a connection, and it seems to go contrary to what we by now know about how minds arose out of matter through a long process of evolution occurring in an already existing world. These are all very good points, and I think it is fair to say that the idealist answer does not look very promising. Nonetheless, it is this answer that I hope to defend in this book. We are indeed special in the world, since our human minds are centrally involved in all of reality in a way that other things are not. There is an intimate link between our minds and all of realty that makes clear that when the story of the world as a whole is written, this connection must be mentioned in the main text or something really important is left out. In this book I hope to make clear what this intimate connection between our minds and reality is, what reasons we have for it obtaining, and why all this is compatible with what we otherwise know about how minds arose in the material world. This position, if correct, would support that we are indeed special and central to all of reality. It would support a metaphysically special status of human beings in the world, although not via their spatial location, nor via their distinction from other species, nor via a connection to God, but via a direct, intimate connection between our minds and reality itself. Any defense of an idealist answer to the Big Question must rightly seem suspicious. It will appear to be more wishful thinking, derived from some delusion of grandeur for human beings, and expressing a deep desire for significance rather than a position that can be defended on its merits. I agree that this is the proper initial reaction to it, but, nonetheless, I will argue that the position stands on its merits and not wishful thinking. For what it’s worth, I arrived at the position I defend here slowly, defending positions in various debates in metaphysics and the philosophy of language over the years which on the face of it have little to do with idealism. But as time went on it dawned on me that these positions are related to idealism, that they indeed support a particular form of idealism, and with it a third, idealist answer to the Big Question about our place in reality. I hope to spell out and defend all of these claims in this book. But first we need to make progress on idealism itself and try to get clearer on what it involves and what kind of position it is. So far we have mostly approached it metaphorically, for example, as the view that holds that there is an intimate connection between our minds and reality itself. Before we can seriously consider defending idealism we thus need to get clearer on this view, as well as what would be required for a proper defense of idealism.

4 idealism and the harmony of thought and reality

1.2 Idealism: the main idea Idealism is, first and foremost, a grand vision about the place of minds in reality. That vision is a metaphysical one, about what role minds play in the metaphysics of the world. Broadly understood, idealism is the view that minds are metaphysically central to reality. This at first leaves many questions open, in particular what metaphysics is supposed to be and what counts as being central for it. As a first approximation, metaphysics tries to find out what reality is like in the most general ways. For minds to be central for what reality is like in the most general ways means at least that the correct metaphysics of the world will give minds a central and important role in its account of what the most general features of reality are. How important and central that role is is, of course, a matter of degree, but there will be clear thresholds for being sufficiently or insufficiently central for a particular position to count as idealism or not. But so far all this is naturally just a first approximation. The most paradigmatic idealist position is what I will call classical idealism. Classical idealism is a position about the relationship between minds and matter. Whereas materialism holds that minds somehow arise from matter, and dualism holds that minds and matter are independent of each other, classical idealism holds that matter somehow arises from minds. Classical idealism thus inverts the order of priority between minds and matter from materialism: matter is derivative on minds, not minds on matter. So understood, idealism is incompatible with materialism, and this is how idealism is often understood. But this incompatibility is special to classical idealism. Classical idealism clearly is idealism on our broad understanding of the latter term, since if matter somehow is derivative on minds, then minds will clearly be metaphysically central to reality. But minds might be metaphysically central in many other ways than giving rise to matter. It could be, just to mention some examples, that minds are not central for there being matter at all, but for which chunks of matter compose an object, or for what matter is like in various ways, or for what the ultimate purpose of matter is, and so on. Thus classical idealism is clearly idealism, but idealism should be understood more broadly than merely as classical idealism. And on this broader understanding it might well be compatible with materialism.1

1 A terminological side note: some idealists think the idealist position is that there is no matter, and that objects are thus made up from something other than matter. On this strict use of ‘matter’, matter itself is incompatible with idealism. On a more loose use of ‘matter’, matter itself is neutral between idealism and its opposition, and the question remains whether matter is mind-dependent or not. Famously, Berkeley denied the existence of matter, and thus used that notion in the strict sense, since he did not deny that tables exist and are made up of something or other. I will use the term ‘matter’ in the loose sense here, and thus accept that tables are made from matter, but this leaves open what matter is like, whether it is mind-dependent or mental, and with it whether idealism is true. This difference is merely terminological and of no substance, but worth clarifying.

idealism and our place in the world 5 Understanding idealism broadly, however, can quickly become too broad: minds might be metaphysically central in the world in ways that intuitively are not idealist ways, as the term is often taken to be. For example, if God created the world, then God’s mind is presumably metaphysically central to reality, and thus minds are central to reality. Taking this to be idealism seems to go too far, since God’s creation of the world alone doesn’t seem to be enough for idealism to be true. Idealism concerns how the world is, after all, not how it came to be. Here it might be tempting to simply exclude causal dependence from minds being metaphysically central and to focus on non-causal dependence instead. But this has two problems. First, it relies on the notion of non-causal dependence, which can seem just as elusive as that of centrality. This is especially vivid in proposals that hold that idealism is to be understood as the view that the non-mental is grounded in the mental.2 ‘Grounding’ in such a proposal likely needs to be understood as a placeholder for one of many yet to be spelled out senses of dependence, and thus simply pushes the issue towards these senses.3 But, secondly, this move does not avoid the issue of centrality. It is too strong for idealism to demand that all nonmental facts are grounded in mental facts, since someone who holds that all except mathematical facts are grounded in mental facts is clearly an idealist. Similarly, if one demands only that all concrete non-mental facts are grounded in the mental, then this is still too strong, since someone who holds that all non-mental facts except the laws of gravity, say, are grounded in the mental is still an idealist. And it is too weak to hold that some non-mental facts are grounded in mental facts, since everyone will hold that on any reasonable way of understanding grounding. What matters is how central the mental is to the non-mental metaphysically, be it via grounding or otherwise. Thus I find it best to start with a rather broad and fairly neutral conception of idealism, risking that it is too broad, and then narrowing it down from there. After all, it is not to be expected that the notion of idealism itself is perfectly precise, since it merely captures a certain metaphysical vision which includes a rather large group of positions, many of which at least have in common that they give a metaphysically central place to minds in the world.⁴ Thus at first we can understand idealism broadly: Broad idealism:

Minds are metaphysically central to reality.

2 See (Chalmers, 2019) and (Guyer and Horstmann, 2015) for proposals to understand the ‘idealism’ along those lines. 3 See (Hofweber, 2009), chapter 13 of (Hofweber, 2016b), (Wilson, 2014), and others for a criticism of giving grounding a central role in metaphysics. ⁴ The term ‘idealism’ is also used in various other ways in philosophy, which I will simply leave aside. Not only is it used for the position in political philosophy of sticking with one’s ideals, which is, of course, completely unrelated to our topic here, I have also heard it used for the metaphysical position that objects are classified by independently existing “ideas” or Platonic universals, which is also quite independent of our main topic here. We are simply concerned with the role of minds in reality.

6 idealism and the harmony of thought and reality Broad idealism concerns the centrality of minds in general, be they human minds, divine minds, or any other kinds of minds. Even if broad idealism were true, this would not thereby answer the Big Question about our place in the world. Idealism is only of direct relevance to the Big Question if it concerns the centrality of human minds, not just any minds whatsoever. It is a question about our place, not the place of minds more generally. Our main concern is the significance of human beings in particular, not just any beings with minds. Our focus thus will be on the issue whether our human minds are metaphysically central to reality. This strengthens broad idealism, since it requires the centrality of our minds, not just any minds, and we can thus call it strong idealism: Strong idealism:

Human minds are metaphysically central to reality.

If strong idealism were true, then this would answer the Big Question. If our minds are metaphysically central to reality, then we would be special and significant in the world overall. The overall story of the world would need to mention us in the main text, not just in a footnote. The third, idealist, answer to the Big Question will thus require strong idealism, not just broad idealism. The idealist position defended in this book will support exactly this strong version of idealism. Broad and strong idealisms are not opposites. Instead, they focus on two separate dimensions of idealism. Broad idealism concerns the centrality of minds in reality in general, not something narrower such as the relationship between minds and matter or the grounding of non-mental facts in mental facts. Strong idealism focuses on our human minds, not something weaker such as any minds. A narrower version of idealism could also be strong. For example, it might be that our minds in particular give rise to matter. In the following I will consider idealism in general as broad idealism, and then focus on whether a strong version of idealism so understood is true. Since our focus is the strong form of broad idealism, some of the worries about broad idealism being too broad will disappear. After all, the positions that are broad idealism in letter, but not idealism in spirit, like God’s mind creating the world, do not carry over to the strong version of broad idealism. No one suggests that our minds causally create the universe. Whether broad idealism is in the end still too broad, even when restricted to strong forms of idealism, has to be left open for now, but I hope it will be clear later that the examples of strong idealism discussed below are not merely idealist in letter but also in spirit. What reason might anyone have for being an idealist and, in particular, for being a strong idealist? As we will see soon, there are a number of motivations for idealism that come from various different philosophical angles. But first there is the problem of making sense more precisely of what the idealist position comes down to. How are we supposed to be central to reality? Intuitively the picture is often driven by rather metaphorical statements: that there is no ready-made world,

idealism and our place in the world 7 that although the world is in some sense independent of us, in another sense it is dependent on our minds, that the world is constructed by our minds, that the world is constrained by our minds, and so on. In a nutshell, somehow there is a significant contribution from our minds towards making up reality as it is, and reality thus depends on our contribution for being how it is. The problem is to make sense of this. How is it to be understood that reality depends on us, and in what sense of dependence is that supposed to be? How can we make sense of us making a contribution to what reality in general is like? And how can we possibly accept anything like this, given that we know that reality was there already before we ever entered the scene? We know that human beings weren’t around 100 million years ago, but reality was already there, more or less as it is now. So how can we be responsible for making a contribution to what reality in general is like? To say that the world is empirically real, but transcendentally ideal, or something similar, only puts a label on the issue, but doesn’t resolve the difficulty. It is in light of such difficulties that idealism has completely fallen out of favor these days. Admittedly, idealism does seem like a terrible idea on the face of it, in particular the strong version of idealism. It is self-aggrandizing, and rings of the good old times when human beings thought of themselves as the center of the universe and the only creatures of any significance, like a species-level form of egomania. It seems to fly in the face of what we have learned about ourselves and other creatures over the last few hundred years, and thus it faces the legitimate charge of being antiquated and only of historical interest. Because of such associations, ‘idealism’ is a bad word in contemporary philosophy. Peter Simons states in print what many others would assent to more privately: “ . . . there are few philosophical views I find more repugnant than idealism. . . . ” (Simons, 2013, 305)

Idealism is thus nowadays mostly considered a position of only historical interest: something believed centuries ago before we found out about our proper place in the world. These days almost all discussion of idealism is in the context of the history of philosophy. This is made vivid by the contrast of the treatment that idealism and realism get in the Stanford Encyclopedia of Philosophy: the survey article on idealism is written by two philosophers who work mostly in the history of philosophy, Paul Guyer and Rolf-Peter Horstmann. It surveys the history of idealism from Descartes to Carnap, with hardly any discussion of contemporary idealism.⁵ On the other hand, the article on realism is written by a philosopher who works mostly not in the history of philosophy, Alexander Miller, and has little purely historical material in it.⁶ Realism is not unusual here; most of the other

⁵ See (Guyer and Horstmann, 2015).

⁶ See (Miller, 2014).

8 idealism and the harmony of thought and reality great -isms which have their own article in the Stanford Encyclopedia of Philosophy are not mostly historical. Idealism is a real exception in this regard. But this historical conception of idealism strikes me as a mistake, for more than one reason. Most of all, it strikes me as a mistake since I have come to believe that a version of idealism is indeed true. But besides that, it is a mistake because there are in fact numerous prima facie defensible versions of idealism, several of which rely on more recent advances in metaphysics and thus they make the most sense in a contemporary setting. I will discuss several such options shortly, all of which I will in the end reject, but all of which should be on the table as real contenders. Thus overall I hope to make the case that idealism should be put back on the menu: not only is it true and significant, but there are also several options available that are well worthy of serious consideration, even though they are in the end mistaken, as I will argue in the next chapter.

1.3 Idealism: some constraints Not every version of idealism is worth taking seriously. Idealism more than other positions in philosophy attracts those who are drawn to esoteric, metaphorical, and suggestive ways of doing philosophy. Before we can begin to discuss idealism as a non-historical position, it will be important to put down some ground rules and to articulate some minimal constraints that any defense of idealism needs to meet. Simply put, those constraints are that one must formulate the idealist position in an explicit and accessible way, that one must support it with some good reasons in its favor, and that one must make clear that the proposed form of idealism is compatible with what we otherwise reasonably take ourselves to know about the world. These constraints will be important in our discussion below, since I will reject various versions of idealism on the grounds that they do not meet one or another of these constraints. The constraints are not unique to a defense of idealism: any defense of any large-scale philosophical view should meet them as well. But these constraints are especially pressing for a defense of idealism, since idealism is the kind of view that invites violating one or the other of them. Actual defenses of idealism, both in print as well as in conversation, often do violate them. Towards the end I will show how the version of idealism defended in this book does indeed meet all of these constraints. But let’s first make the constraints themselves more explicit. First and foremost, the idealist position itself must make sense. Idealism does make sense in the abstract. That minds are metaphysically central in reality is a clear enough statement by itself, but it completely leaves open how minds might be central in reality. What about minds makes them central in reality, and in what way are they central? A particular idealist position should make this clear and spell it out explicitly so that we can evaluate the proposal properly. Here it is not enough

idealism and our place in the world 9 to say that the world is transcendentally ideal or that the world is mind-dependent; what needs to be added is to spell out what ‘transcendentally ideal’ is supposed to mean and in what sense the world is supposed to be mind-dependent. Of course, philosophers who have used these terms do attempt to spell them out. I don’t think I am making a very controversial demand here, only the very reasonable one that one must explicitly articulate the idealist position. For example, one can’t simply rely on novel, primitive terminology in the formulation of idealism. If I were to propose that the world is transcendentally ideal, that this is compatible with it being empirically real, and that the notions of being transcendentally ideal and being empirically real can’t be spelled out any further, but must be accepted as primitive metaphysical terms, then this is clearly unacceptable. Not only would it remain mysterious how one could have an argument in favor of idealism so understood, it remains mysterious what the position itself even is.⁷ Obviously, metaphysics might rely on primitive terms, and not every notion can be spelled out. Notions like ‘property’ or ‘object’ might be primitive. But to rely on “transcendentally ideal” or “mind-dependent” as primitive should seem highly suspicious. The first constraint that any defense of idealism should meet is thus to state the position explicitly. And that is to say, not to rely on novel, primitive, and unexplained terms at the crucial moment that is supposed to illuminate how our minds contribute to what reality is like. Thus our first constraint is: The Explicitness Constraint: The idealist position must be stated explicitly, without novel, unexplained notions at crucial points. How explicit such a statement of the idealist position should be can differ depending on which notions one finds acceptable and which ones one finds dubious. Here, I would imagine, a logical positivist might have much stricter standards than a contemporary neo-Aristotelian metaphysician, who has no worries about such notions as essence, grounding, metaphysical explanation, and so on. My own preferred standard is somewhat in-between those two extremes. One illustrative and controversial case here concerns the notion of grounding. Understanding idealism as something like the position that the non-mental is grounded in the mental is a promising start, but is it enough by itself, or does one need to say more about how and in what sense the non-mental is grounded in the mental? Does it by itself merely say that the mental is prior to the non-mental in some metaphysically significant way, with it being left as a promissory note to say more precisely in which sense of ‘prior’ the mental is prior? Or is it to be taken as the complete, and fully specific account of what idealism comes down to? My own view is that relying

⁷ Kant relies on this distinction in (Kant, 1781), where he tries to spell it out, with unclear success. There are many attempts to make Kant’s idealism intelligible as contemporary metaphysics, including (Langton, 1998), (Bader, 2010), and (Stang, 2017).

10 idealism and the harmony of thought and reality on grounding by itself is not enough, but that is subject to debate. Still, I hope to make clear later on that the positive proposal and defense of idealism given in this book meets even the strictest standards for explicitness. The idealist position can be stated in completely ordinary terms, relying on no distinctly metaphysical primitive notions at all, but only on ordinary terms in their ordinary use. Pulling this off is one of the main results of this book. Meeting the explicitness constraint is hard, and a notorious problem for idealist positions. We will see several examples of idealist positions that fail to meet this constraint below. Those positions are suggestive and spelled out in metaphorical ways, but that is not good enough. What we need is a fully explicit version of what the idealist position is. A special and important case of the explicitness constraint concerns the notion of dependence. It is very tempting to formulate idealism as the position that the world is mind-dependent. Which aspects of the world are mind-dependent can, of course, vary with different positions: which things exist, which properties these things have, that there are objects with properties in the first place, etc. are all candidates for being declared mind-dependent in some sense. The hard part is to say in what sense this dependence is supposed to hold. There are at least two ways of understanding dependence which are for our purposes uncontroversial but also unhelpful: causal dependence and counterfactual dependence. Simply put, A causally depends on B if B caused A, and A counterfactually depends on B if A wouldn’t be there if B weren’t there. How causal and counterfactual dependence are to be spelled out better and how they relate to each other is widely debated in the literature on causation.⁸ Those details don’t matter for us now, since basically no one claims that the world causally or counterfactually depends on our human minds. However our minds contribute to reality, it is not by bringing it about causally. And it is hard to make sense of how reality could counterfactually depend on our minds, in particular in light of the fact that the world was there before us, and we human beings developed in the world over time. It is essentially agreed upon that causal and counterfactual dependence are not the right notions of dependence when it comes to the idealist’s claim that the world is mind-dependent. The question thus remains in what sense of dependence the world is supposed to be mind-dependent. It is not enough to simply introduce a new notion of dependence as a primitive and declare the world to be mind-dependent in that novel sense. At best this can be a label for a notion of dependence that is to be spelled out, something like a promissory note that one will make this explicit later. Thus the explicitness constraint requires that if one relies on the notion of minddependence in one’s statement of idealism, then one needs to say explicitly in what sense of dependence reality depends on minds. This problem I will call:

⁸ See, for example, (Menzies and Beebee, 2020).

idealism and our place in the world 11 The Dependence Problem: Any sense of ‘dependence’ in which the world is claimed to be mind-dependent must be made explicit. The idealist position defended in this book can be formulated with a novel notion of mind-dependence, but that notion can be made perfectly explicit, as I will show below. Meeting the explicitness constraint is hard, but it might be even harder to present good reasons that support the idealist position that one has explicitly formulated. What kinds of reasons might one have for holding that the world is mind-dependent in a proposed sense of dependence? Obviously this will depend on what sense of dependence one proposes, but the worry is that one will at best come up with some fairly weak reasons: maybe it would make for a simpler theory of how objects relate to their properties, or a theory with fewer metaphysical primitives, or a slightly sparser ontology, and so on. I doubt that any reasons like this would be very powerful for defending a position like the centrality of human minds in the world, although in the end it will depend on the particular case. Still, the problem remains what kind of reasons one could possibly have for an idealist position, and without such reasons the whole project is futile. Although it is hard enough to simply formulate the idealist position in a coherent way, what is required in addition are good reasons in favor of that position. This is our second minimal constraint: The Support Constraint: position as formulated.

There must be good reasons that support the idealist

And to show that this constraint is met one must be able to specify what these reasons are. In a sense, this goes without saying, but it is worth making explicit, since it is tempting to focus too much on the explicitness constraint and to simply come up with a coherent version of idealism, which is hard enough, and then give some slim reasons for why this version might be true. We will need to do better. The paradigmatic, but not only, way to give such reasons that support idealism is to have an argument in favor of idealism. To have an argument that does not rely on idealist premises for the idealist conclusion is the gold standard for such support, but other options are on the table as well. Maybe philosophical progress does not require such arguments and some other motivations for idealism might work as well. But some support with sufficient force in favor of the idealist position as stated must be given. How one might support the thesis of the centrality of human beings in all of reality is not at all clear, but I hope to do so by giving an argument for that conclusion in this book. Finally, we should put some limits on what kind of idealist position we should take seriously. It won’t do to propose that the world started with human beings, who created it somehow, since we know that there was a world before us. We have

12 idealism and the harmony of thought and reality found out all kinds of things about reality and our place in it, and it is very unlikely that someone will have a philosophical argument that shows that these things are false after all. Thus any reasonable version of idealism will be compatible with what we generally take ourselves to know to be true. This does not mean that it needs to be compatible with common sense. Common sense can be false, and we have good reason to think that it is false in many cases. But it should be compatible with what we know from the results of inquiry more generally. We don’t have to show this in all detail, but it should include a number of key facts: that the world did not start with us, that the world goes further than we have seen, and so on. It is not that it is somehow impossible for these things to turn out to be false. It is just so unlikely that there is any reason from philosophy to reject them that it is reasonable to instead focus only on idealist positions that are compatible with what we rightly take ourselves to know about the world. Let us make that explicit as a third constraint that any reasonable version of idealism should meet: The Compatibility Constraint: Any version of idealism must be compatible with the general facts about the world that we reasonably take ourselves to know. What we reasonably take ourselves to know is up for debate, of course, but of relevance for us will only be various general facts that are universally considered soundly established. It could be that the true form of idealism does not meet these constraints. Maybe the true view cannot be explicitly stated by us, and maybe there is no reason in its favor that we can give, and maybe what we take ourselves to know is in fact false, and so we falsely rule out the true form of idealism, since it is incompatible with what we take ourselves to know. I can’t rule this out, but if so, then we would be in a terrible position to investigate idealism. In the following we will focus on versions of idealism that we can properly assess. I will argue that several options to defend idealism are to be rejected, since they do not meet one or another of these constraints. However, the positive proposal I will make below does meet all three of these constraints and solves the dependence problem as well. Only such a version of idealism has a shot at giving us a third, idealist answer to the Big Question. To do this we cannot simply rely on metaphors and intuitions about our significance. We need to explicitly state the idealist position, then support it with an argument, and finally show how the position so understood is compatible with what we reasonably take ourselves to know about the world. How that might work is not easy to see, but it might help to think about some basic options one has.

1.4 Idealism: the main options Idealism is understood broadly here, as the view that minds are metaphysically central to reality. To see how that might be it is helpful to think about each of

idealism and our place in the world 13 the key parts in this characterization of idealism: [minds]1 being [metaphysically central]2 to [reality]3 . We will look at them in turn. First, minds are composed of various different mental components that relate to each other somehow to form a mind, at least a mind like ours. An idealist might hold that one or another of these aspects is what makes minds metaphysically central to reality, and depending on which one is supposed to be central, this will likely lead to a very different form of idealism. Minds like ours contain at least consciousness, self-consciousness, emotions, conceptual thoughts, perceptions, feelings, and intentionality. To hold that consciousness is metaphysically central to reality will likely lead to a rather different position than to hold that emotions are. Among the different aspects of minds, I think it is fair to say that consciousness, self-consciousness, perception, and conceptual thought have been the most prominent ones to motivate idealist positions, while emotions or feelings have been far less popular. I am not sure what this shows, but the position defended here in the chapters to come will also side with the tradition and hold that conceptual thought is what makes us metaphysically central to reality. However, consciousness plays no direct role in the defense of idealism. For now we can note that if minds are central to reality, then there is a real issue about which aspects of minds make them central, and that different proposals about this will likely lead to very different forms of idealism. Second, our issue concerns the metaphysical centrality of minds in the world, not the centrality of minds in other ways. Similarly, strong idealism concerns whether or our human minds are metaphysically special, not necessarily whether they are special in other ways. In particular, we need to distinguish our issue from questions about the ethical or normative significance of human beings. The metaphysical and ethical significance of human beings are largely independent issues, barring some argument to the contrary. We might be metaphysically special, but the way that is so has no ethical significance. Or we might be ethically significant in a way other creatures are not, even though we are metaphysically unimportant. Our concern simply is the metaphysical significance of human beings, not their ethical significance. Metaphysical significance should be understood as contrastive, but not exclusive. When we say that we are significant, then this is to be understood as that we are significant in a way that other things are not. If everything is equally significant, then this would not support idealism. So, if we are made of matter and matter is metaphysically significant, then in a sense we are significant, derivatively on the significance of the matter we are made from. But for strong idealism to be true we need to be distinctly special and more special than other things. We need to be special in a sense that other things are not. But this shouldn’t be taken too far. It should not be required that we are uniquely special in order for strong idealism to be true. It could be that other creatures, which are not biologically human, are special in just the way we are. Maybe there is some similarity between their minds and ours that makes them just as special

14 idealism and the harmony of thought and reality as us. Maybe the human mind is special in a way that other things are not, but not because it is tied to human beings, but because it is a mind of a certain kind which has a central metaphysical place in the world. Besides some unclarity about what being central amounts to, there is a whole set of issues connected to what it is to be metaphysically central. Much has been written about how precisely to understand metaphysics and its relationship to other parts of inquiry, but this is not the place to get into that debate. We will simply have to work with an approximation, where metaphysical centrality concerns something like being central for the large-scale, general, and significant aspects of the world. It would be hopeless to first try to clarify the aim and purpose of metaphysics, and then to understand metaphysical centrality in light of that. A vague, working understanding of this notion will be enough for us now. Being central for the happenings on the surface of the earth does not count, but being responsible for the difference between objects and properties would certainly count. There are numerous ways how minds might be metaphysically central to reality, and these will again lead to numerous different forms of idealism. One option is that matter is derivative on minds, which could be in one of several ways: minds give rise to matter in some sense, or minds determine which parts of matter form an object, or minds determine what objects are like and what properties they have, either in general or only for a limited range of properties. But besides focusing on matter and how it might be tied to minds, there are also other ways in which minds could be metaphysically central. Minds might play some metaphysical role such that them playing that role is key for something of metaphysical importance. This could be something very fundamental, like there being objects in space and time, or there being a division between objects and properties, and so on. Or it could be that although minds do not make it the case that there are objects or what they are like, but they limit what there can be and what it can be like. Thus it could be that although minds do not construct the world, they constrain the world. The relationship between construction and constraint will reappear prominently below. At a first stab one might think that if our minds construct the world, somehow, then they also constrain it, since the limits of our constructions would be a constraint on the world. But maybe our minds constrain the world without constructing it. Maybe the world somehow has to fit our minds, and anything that would go beyond that is for some reason ruled out. How that could be is not at all clear, of course, but if so, then our minds would be metaphysically central in a very different way than if they construct the world. Thus overall there is a large range of options how minds might be metaphysically central, options that go much further than those envisioned by classical idealism, which holds that minds give rise to matter. The third part of our understanding of idealism involves the concept of reality. ‘Reality’ can be understood in numerous ways, and on some of them the issue

idealism and our place in the world 15 whether we are central to reality trivializes, while on other ones it leads to a substantial question. Minds, and our human minds in particular, will be central to reality understood as the world of appearances, or as the world as it is for us. On the most natural way of understanding reality as the world of appearances, it is reality as it appears to us to be, which in turn is just reality as it seems to us to be. So understood we might well be central to it, but this is not a special notion of reality at all, but rather simply plays on the distinction between how things are and how they appear to be. What is at issue for us is our place in reality as it is, not just in how it appears to us to be. Thus this notion of reality can safely be put aside, unless, of course, it is understood differently. And there might well be another notion of reality in the neighborhood here, which is closer to our topic: the distinction between reality as it is and reality as it is in itself. This distinction suggests that there is one way to think of reality on which it is the result of a contribution from us and a contribution from something else, both of which get combined to form reality as it is. And then there is also another way of thinking of reality, as it is in itself, which subtracts our contribution to reality as it is: reality without our contribution to it. If those are two equally legitimate notions of reality, then it would make sense to say that we are central to reality, understood in one of two ways. And that might well motivate idealism. But by itself this merely sets the target for what a defense of idealism has to work out: how should we understand the idea that we make a contribution to reality in general? How can we make sense of a distinction between reality as it is and reality as it is in itself? So far this distinction does not yet make sense; it is merely the vague articulation of an idealist vision. At this point in our investigation we can only rely on distinctions that already make sense; the rest needs to be earned and worked out. And if there is a coherent distinction between reality as it is and reality as it is in itself, we have not yet earned it. Of course, various philosophers have used a distinction like this,⁹ and they and their followers likely will claim that we at this point have earned versions of such a distinction. But I have to confess that despite putting some effort into trying, I am unable to articulate any of these attempts in a way that I would find satisfactory. That could be my fault, of course, but I think I am not alone in finding this distinction suggestive, but not yet earned, and this is how I will treat it for now. There is, however, one key distinction in how we can think of reality that does make sense and that can be an initial guide to some options we have in pursuing the idealist vision in a more precise way. It is the distinction between reality as the totality of facts and reality as the totality of things, or in other words, the distinction between reality as all that is the case and reality as all that is. Although Wittgenstein famously declared that “the world is the totality of facts, not of things” ⁹ Most famous among them is in particular Kant, [1781], who, of course, also used ‘world of appearances’ in a different way than ‘world as it seems to us to be’.

16 idealism and the harmony of thought and reality (Wittgenstein, 1921, 1.1), it is not required that one takes sides here. It makes sense to think of all the facts as well as all the things, and which one is properly labeled ‘the world’ or ‘reality’ does not need to lead to a substantial disagreement. Instead we can accept two notions of reality, or of the world, one as the totality of things and one as the totality of facts. And although some philosophers would like to reserve the term ‘reality’ just for something related to facts, and call the totality of things not ‘reality’ but ‘what is real’, (Fine, 2009), this again should best be seen as a proposal about terminology, not as a substantial disagreement. The distinction between reality as the totality of facts and reality as the totality of things is intuitive and clear. It is not like the distinction between reality as it is and as it is in itself, which merely hints at something that needs to be earned, but isn’t something that can be relied upon at the start. With this distinction at hand we can now consider two strategies for a defense of idealism: one that holds that minds are central for reality, understood as the totality of things, and one that holds that minds are central for reality, understood as the totality of facts. Since the totality of things is the ontology of the world, at least on one common use of that term,1⁰ we can call ontological idealism the view that minds are central for reality understood as the totality of things. And we can call alethic idealism the view that minds are central to reality understood as the totality of facts. For both alethic and ontological idealism we can distinguish the weak and the strong version: minds in general, or our human minds in particular, are central to reality so understood. Although the distinction between alethic and ontological idealism might be legitimate, it might also seem not to be helpful, since there is a close connection between the things and the facts. First, there is a close connection between facts about what exists and what exists. For every thing that exists there is a corresponding fact that this thing exists. And for every fact that a particular thing exists there is that thing. Besides facts about existence there are, of course, also other facts, but at least it seems that the totality of facts determines the totality of things, since it includes facts about which things exist. Second, there is a question about whether facts themselves exist, and thus whether facts are a kind of a thing. Uncontroversially, facts obtain. But is the fact that snow is white also something which exists? If so, then the totality of things includes the totality of facts, and thus all the things determine all the facts, since the facts are just part of the things. Whether this is indeed so will be an important question later on. In any case, it is clear that the totality of facts and the totality of things are closely tied to each other. Since it was hard to see how minds might be central to reality, it doesn’t seem to help much to focus on reality as either the totality of things or the totality of facts in making clearer how minds might be central to reality. After all, how could they

1⁰ Some philosophers prefer to reserve the term ‘ontology’ only for the fundamental things, not necessarily all the things. See (Schaffer, 2009). This is again merely a disagreement of terminology, and does not really affect our main issue here.

idealism and our place in the world 17 be central for one, but not the other? But besides these initial worries, I hope to make clear later on that this distinction is crucial for seeing how idealism can be true and how it is compatible with a particular form of realism. Overall then there are a number of ways in which one could try to formulate idealism, connected to which aspects of minds are taken to be metaphysically central to reality understood one way or another. Depending on how this goes, one could pursue different strategies for defending idealism so formulated. In this book I will defend the view that human conceptual thought is metaphysically central to reality, understood as the totality of facts, since it constrains, but doesn’t construct, reality so understood. The main goal of this book is to properly formulate and defend this position and its consequences, especially how it leads to an idealist answer to the Big Question. I hope to show that the proposed version of idealism does not have the problems that have plagued more traditional versions of idealism. There is a series of problems and obstacles that idealist positions face, which have ruled them out from being serious contenders for being the right metaphysical view according to the assessment of many philosophers. To properly appreciate the positive proposal made in this book, it is crucial to see why and how this position does not suffer the same fate. I thus want to first discuss several, more or less well-known, versions of idealism and why they fail in the next chapter. After that I will present and defend the version of idealism I believe to be correct, and why it does not succumb to the same problems.

2 Forms of idealism we should reject Although idealism has been defended in various forms throughout the history of philosophy, it is almost universally rejected these days. There are good and bad reasons for this widespread rejection of idealism. The good ones are based on the fact that the positions commonly associated with the label ‘idealism’ are indeed highly problematic and, I believe with the majority, false. The bad reasons for rejecting idealism are tied to generalizing this to all versions of idealism broadly understood. Idealism is a much more diverse and interesting position than how it is commonly taken to be. Often idealism is simply understood as classical idealism, the view that matter is somehow derivative on minds. Classical idealism, in particular as a form of strong idealism, is hard to defend, since it quite clearly runs into the constraints discussed on page 8. Strong classical idealism is not compatible with what we know to be true about our place in the world, since the material world was there before us. It is thus hard to makes sense of how matter might arise from our minds. Unless possibly if ‘arise’ is understood in a non-causal way. But then it is unclear how it meets the explicitness constraint, at least until more is said. There are certainly other options available, and we will see several of them in this chapter. Still, classical idealism looks rather unpromising. But idealism does not only concern a priority between minds and matter; it should be seen as a wider range of views about the metaphysical place of minds in the world. Idealism concerns the metaphysical centrality of minds in the world, not merely the priority of minds over matter. And this wider conception of idealism gives rise to a series of options for how one can be an idealist that are often neglected and not as well known as they should be. In this chapter we will discuss several such options for being an idealist that go beyond classical idealism. Although they generally are better than classical idealism, they nonetheless face their own serious problems and should in the end be rejected as well. In the following I hope to present several non-classical versions of idealism and the reasons for why we should reject them. But it won’t all be negative for long. Once we see why various forms of idealism run into problems we will be ready to consider the positive proposal I will defend in the next chapter, and why it does not face those problems. I do not intend to survey all the options one in principle has for defending idealism, nor can I hope to survey all versions of idealism that have been presented over the history of philosophy, nor even that are defended by someone today. I will instead focus on five approaches that strike me as specially deserving of Idealism and the Harmony of Thought and Reality. Thomas Hofweber, Oxford University Press. © Thomas Hofweber 2023. DOI: 10.1093/oso/9780198823636.003.0002

20 idealism and the harmony of thought and reality attention, and that in a natural way move further and further away from classical idealism. There are many other options as well, but in the following critical survey of different versions of idealism I picked only from those that meet the following criteria. First, I will focus exclusively on contemporary versions of idealism and simply leave historical idealists aside. The reason for this is not that I think the history of philosophy is uninteresting and nothing can be learned form it; far from it. The reasons are simply the following: first of all, the proper discussion of even just one of the famous historical idealists would go much beyond what I have space for. This book is not a survey of idealism as it has been defended over the history of philosophy, but an attempt to defend a new version of idealism. The following brief survey is not for completeness sake, but to set the stage and to highlight some problems that need to be overcome in any defense of idealism. Furthermore, I am the wrong person to make a contribution to the understanding of the historical idealists. Many good books have been written about various historical idealist philosophers and movements, including at least (Beiser, 2002), (Förster, 2012), (Mander, 2011), (Pinkard, 2002), and many others. I do not have the expertise to contribute to this body of work, although I very much appreciate reading it. I hope to approach the topic from the other side, so to speak, by aiming to defend a version of idealism, largely without aiming to compare and contrast it with historically prominent forms of idealism. I hope that this will be of interest also to those who work primarily in the history of philosophy, just as work in the history of philosophy is of interest to those who work outside of it. Second, I will leave aside motivations for idealism that come primarily from the philosophy of physics, that crucially rely on God, and that downplay the role of human minds. In the case of the first, there is a tradition to support a version of idealism by connecting it to the philosophy of quantum mechanics, in particular the role of minds in solutions to the measurement problem. Early discussions of quantum mechanics often have an idealist element in them, because they claim that minds play a crucial role in the world, since they are responsible for the collapse of the wave function when measurement occurs. Thus how the material world evolves over time is tied to minds, and therefore minds are central in the material world.1 However, such an interpretation of quantum mechanics has long lost its status as a main contender, and all of the most widely discussed contemporary versions of quantum mechanics—spontaneous collapse theories, Bohmian mechanics, and the Everett interpretation—do not have this feature of giving minds a special place in the physical world. To pursue this further is easily a book-length project on its own, and again one that shouldn’t be written by me. Since it does not look promising to motivate idealism along those lines, and since

1 For an example of such an understanding of quantum mechanics, see (Faye, 2019).

forms of idealism we should reject 21 many others have argued this case,2 I will simply leave it aside here and focus on other motivations for idealism instead. Since our overall goal is to determine the place of human beings in the world and to assess the viability of an idealist answer to the Big Question, I will also leave aside motivations for idealism that rely on a special role of the mind of God in their support of idealism, as well as versions that postulate a special place for human beings via their connection to God. To illustrate the former option, consider what we can call minimal theistic idealism. It holds that the world is just as the standard naturalistic picture has it, with matter distributed in spacetime, obeying the laws of physics, humans evolving out of matter by chance, etc., but there is one more addition: matter is an aspect of the mind of God. Thus matter is derivative on the mind of God, and so broad idealism is true. This version of idealism would not get us a new answer to the Big Question, but it is quite clearly a version of idealism. There is, of course, a question what argument one might possibly have for this form of idealism, but I will not try to support or refute it here. Instead, I hope to make progress on a more radical version of idealism that gives the human mind directly a central metaphysical role in the world. Our focus will be strong idealism, and with it the possibility of a third answer to the Big Question. Minimal theistic idealism will not get us there. For a similar reasons I will also largely leave aside motivations for panpsychic versions of idealism, which hold that all matter is mental or has some protomental features. Although panpsychism has enjoyed a bit of a revival in recent years, mostly via the motivation that some aspects of the mental, in particular consciousness, cannot be explained in purely non-mental, materialist terms. Since dualist approaches to the problem of explaining how the mental and the nonmental interact with each other are problematic, this suggests pursuing a panpsychic route, where all matter has mental, or at least proto-mental, features. But not only does this not seem to solve the problem that motivates the view, since one still must explain how the proto-mental gives rise to full on consciousness,3 it also does not lead to a form of strong idealism and with it to a third answer to the Big Question. Our focus will instead be how our human minds in particular might be metaphysically central to the world. Third, I will hold all proposed versions of idealism to the constraints that any reasonable version of idealism should meet, as discussed above on page 8: they need to be formulated in an explicit and accessible way, they need to be supported with good reasons, and they need to be compatible with what we generally take ourselves to know about the world. As mentioned above, a form of idealism could

2 See, for example, (Maudlin, 2019) and (Lewis, 2016) for more on both the substance and the literature. 3 For a discussion of panpsychism and its relationship to idealism, see (Goff et al., 2020) and (Chalmers, 2019).

22 idealism and the harmony of thought and reality be true without meeting these constraints, but it is unreasonable to pursue such options of idealism. I will focus only on views that have a shot at meeting the constraints. Even within these boundaries, there is much to discuss. This chapter will focus on five versions of idealism, and why we should reject them. I hope to work out a series of problems that any version of idealism faces, and which are not resolved in these five versions to come. This will be the largely negative, critical part of this book. Anyone who prefers to directly move to my positive proposal about which version of idealism is indeed correct can, of course, skip this chapter, and move straight to the positive proposal in Chapter 3. But to properly appreciate why and how the positive proposal succeeds, it is important to first see why other proposals fail.

2.1 Phenomenalism The most popular form of idealism these days is phenomenalism. Most recent idealists are phenomenalists in some form or other, with particular well-developed accounts given by John Foster in (Foster, 2008) and Michael Pelczar in (Pelczar, 2015). Nowadays it is not uncommon to think of phenomenalism as more or less synonymous with idealism. But this is a mistake. Idealism should be seen as a much broader group of views, one that includes phenomenalism but also many other quite different positions. And phenomenalism is not the best of those views, it seems to me, and it has its popularity among idealists unjustly. In this section I would like to briefly review the main idea behind phenomenalism, how it can be developed in different ways, and why it seems to me that this form of idealism is mistaken. One way to see the motivation for phenomenalism is to consider a representational theory of perception. According to such a theory, when I see a chair, then I most immediately relate to something mental—a mental chair image—which in turn represents a material chair. What I am directly and consciously aware of is, first and foremost, the mental image of the chair. I relate only indirectly, via that image, to something external of me: the material chair. This view contrasts with a direct realist position, which holds that I am aware of the material chair in an unmediated way. The direct realist view thus holds that the immediate object of my conscious perceptual experience is the material chair itself, not some mental image of it. Both views have prima facie problems. The direct realist view needs to make clear how something external to me, something outside of my mind, could be immediate object of my conscious experience, which is inside my mind. The representational theory needs to make clear how this representational relationship is to be understood. In particular, how can it be epistemologically significant so that we come to know about the external, material world in perception?

forms of idealism we should reject 23 Both of these problems can prima facie be overcome if we, in a sense, move the material chair inside the mind. If the material chair is identified with, or somehow constructed from, the mental image, then this solves the problem how we can be directly related to the chair. And it promises to solve the problem how perception can lead to knowledge about the chair itself. Bridging the gap between the internal mental world and the external material world by moving the material world inside the mind is one of the key motivations for traditional phenomenalism. Phenomenalism thus holds that there is something mental closely tied to perceptual experience—the phenomena—and that the material world is somehow constructed from these phenomena. What these phenomena are more precisely we will have to largely leave at an intuitive level: the mental image of a chair, or the color patch in the visual field, or the firmness sensation associated with it, and so on. The material object, on the phenomenalist’s understanding of it, is somehow derivative on, or constructed from, these phenomena. How that is supposed to go more precisely is left open by this so far, and different ways of doing it give rise to different versions of phenomenalism. The simplest way of thinking of it is that a material object is just a bundle of phenomena: the bundle of the color patch, the firmness sensation, the shape sensation, and so on. So understood, phenomenalism turns into a version of ontological idealism: material objects are identical to some constructions of mental things, in particular bundles of phenomena. But this version of idealism has some immediate problems, prominent among them are at least the following. Unobserved objects: If objects are just bundles of phenomena, then there are no objects unless there are the corresponding phenomena. But sometimes objects seem to be unobserved, but nonetheless there. This is not merely part of common sense, for what it’s worth, but in fact required to be so by what we know to be true. Not only is it required by the fact that there was a world long before human beings, but furthermore many scientific explanations of ordinary events depend on working whether or not someone observes various antecedent events. None of this would work if observation were required for the existence of an object, and thus prima facie this simple form of phenomenalism violates the compatibility constraint, i.e. being compatible with what we know to be true. Explaining regularities: Some phenomena always go together. But what explains why my sensations of yellow and of banana shape are often followed by that of a banana taste? The materialist seems to have a clear answer: there is a mind-independent object, the banana, which causally affects different senses in different ways, and which thus explains why there are certain regularities among my phenomena. But how can the phenomenalist explain that there are these regularities? Intersubjectivity: The phenomena I directly relate to belong to my mind. The phenomena you relate to belong to yours. But does that mean that there is no shared world? If so, how come that our different, private worlds are so well

24 idealism and the harmony of thought and reality coordinated? When we play table tennis, then whenever I see the ball going away from me, you see it coming towards you, and the other way round. But why would such a coordination of your phenomena with mine occur, unless, of course, there is a shared mind-independent material world which causes both? These are all serious problems, and there doesn’t seem to be an obviously good way out of them. For example, concerning the problem of explaining regularities, it is certainly no good to hold that nothing explains these regularities, since there are perfectly good materialist explanations of them available, which clearly speaks in favor of materialism. To claim instead that there are phenomena-level laws, which explain these regularities, is a bit better. But it is still terrible, since one needs a separate basic law for each kind of regularity—a banana law, an apple law, and so on—whereas the materialist alternative can explain such regularities as derivative on much fewer basic laws that govern simpler parts of matter. Materialism clearly wins by comparison here. One attempt to get out of all of these problems is in essence inspired by the archphenomenalist George Berkeley in (Berkeley, 1710): reliance on God.⁴ On this line, all objects are bundles of phenomena, but not all objects are bundles of human phenomena. Objects unobserved by human beings can still exist as bundles of divine phenomena. Furthermore, maybe there is some much simpler explanation of why divine phenomena exhibit certain regularities, ones that are comparable to materialist scientific explanations? And maybe God’s phenomena can be relied upon also to explain why there is a coordination of the phenomena in our human minds. But even if we grant the existence of God and that God’s mind contains phenomena that can be properly bundled, this solution is more problematic than it might appear. First, phenomena are generally seen as being internal to a particular mind: they belong to only one mind. Bundles of phenomena consequently belong also to just one mind, and so objects, understood as bundles of phenomena, are in a sense private: I have mine, you have yours. But then it doesn’t help to rely on God’s mind to explain why the cheese, a bundle of my phenomena, got half eaten overnight, because a mouse ate it, even though I never saw the mouse. There surely can be a mouse, understood as a bundle of phenomena in God’s mind, but without more being said, it can’t eat my cheese, since the mouse and the cheese belong to different worlds, so to speak: one is a bundle of God’s phenomena, living only in God’s mind, the other is a bundle of my phenomena, living in my mind. So far this doesn’t solve our problems, but maybe we can do better. There are two ways ⁴ How Berkeley’s own philosophy is to be understood more precisely is subject to scholarly debate, and beyond the scope of this book. For scholarly work on Berkeley related to his idealism, see (Winkler, 1994) or (Dickler, 2011). In the following, we will simply take inspiration from issues that arise with interpreting Berkeley’s philosophy, without the ambition of offering such an interpretation. And a reminder of the terminological side note from page 4 above: I use the term ‘matter’ in the loose sense, where matter is just whatever regular objects are made off and thus the existence of matter is compatible with idealism, while Berkeley is generally taken to use it in a strict sense, where the idealist denies the existence of matter.

forms of idealism we should reject 25 around this problem that suggest themselves: first, God’s mind causes phenomena in my mind to be as they are and, second, phenomena can belong to more than just one mind. If the phenomena in God’s mind are causally responsible for the phenomena in my mind, then this might be a start to explaining regularities in the phenomena in my mind, but it in essence gives the phenomenalist game away. When I say that the mouse ate the cheese, even though I didn’t see a mouse, only the half-eaten cheese, then I must be talking about bundles of phenomena in God’s mind for this to make sense. There is no mouse bundle in my mind, and if the cheese bundle in my mind is what I am talking about with ‘the cheese’, then it makes no sense to say that the mouse ate it: the mouse lives in God’s mind, the cheese in mine. But if the objects I talk about are all in God’s mind and these objects are the cause of phenomena in my mind, then we can get around this problem. However, we then in essence get a representational theory of perception: what I immediately relate to in perception is not the object itself, which is in God’s mind, but something caused by these objects, which is in my mind. Phenomenalism so understood is not too dissimilar to what I called minimal theistic idealism above on page 21: the view that the material world is just as a materialist thinks it is, with one addition: matter is an aspect of the mind of God. There is nothing distinctly phenomenalistic about this proposal any more, it simply collapses into a fairly standard view, augmented with the additional claim that matter is an aspect of the divine mind. As discussed above, this view has little going for it, and it is also outside of the scope of this book. If phenomenalism has something to offer, it should not simply collapse into minimal theistic idealism. If, on the other hand, the phenomena in my mind are not caused by the phenomena in God’s mind, but are identical to the phenomena in God’s mind, then we face a different set of problems. God’s mind would then overlap not only with my mind, but with all other minds as well. Thus in our table tennis example above, God’s mind would not just contain the phenomena of the ball coming towards one’s own perspective but also of the ball moving away from it, as well as the phenomena of the ball moving to the right, as a spectator might see it, and as it moving to the left, as another spectator might see it. It is a bit hard to make sense of how a single mind can at the same time have all these different phenomena in it, as God’s mind would have to have. Would it be that God sees the world from every perspective at the same time? Can we make sense of a single mind having all perspectives at once? But maybe that isn’t even necessary, since it might not be crucial that these phenomena all occupy God’s mind. Maybe the phenomena don’t have to belong to a single mind at all. Maybe we can make sense of these phenomena simply being there, as the stuff that makes up the world, whether or not they are all collectively experienced by a divine or any other mind. On such a picture, the world simply consists of a whole bunch of phenomena, and some of them come together

26 idealism and the harmony of thought and reality somehow to form minds like ours. Those are unified in some coherent way, like our minds, while other phenomena come together to form other minds, like that of my table tennis partner, which is internally unified just like mine, and which is coordinated with my mind in a way as indicated above: the ball moving towards me is coordinated with the ball moving away from them, and so on.⁵ But then we face the same problem about the explanation of these regularities and coordinations as we did before: what explains that there is a coordination across minds so that when certain phenomena bundle up to form a mind, other, coordinated phenomena bundle up to form a different mind, as in the table tennis example? And what explains that when certain phenomena belong to one bundle at one time, then certain other phenomena belong to this bundle at another time, as in the banana example? It seems the phenomenalist has few resources available to explain this. They could try to claim that there are basic laws governing the phenomena within and across bundles. But if so, then these laws must be numerous and rather specific. After all, the phenomena do not arise from something more basic, but are themselves the basic aspects of the world. There must be one law for the table tennis ball, another one for the banana, and so on. The materialist has a much better and more satisfactory explanation of these coherences and coordinations, and as such they clearly have the upper hand here. In fact, the explanatory advantages of materialism over this form of phenomenalism are so great that phenomenalism so understood seems only a viable position if materialism can simply be ruled out as untenable. Some phenomenalists have certainly argued this way: either mind-independent matter is incoherent,⁶ or minds arising out of matter is impossible,⁷ or some other aspect of the materialist picture is simply to be ruled out by philosophical considerations.⁸ I don’t want to discuss these arguments here, but I do confess that I find them unconvincing. And without simply ruling out materialism, this version of phenomenalism does not look very promising. But phenomenalism can do better than this. So far phenomenalism was understood as a position that takes material objects to be constructed somehow out of phenomena. It takes objects to be bundles of phenomena, or some similar construct, be they in my mind, or in God’s mind, or just there, as the basic parts of the world. The first problem with this approach was to deal with unobserved objects, objects for which there are no phenomena that can be used to construct them with. The reply we considered above was to rely on God’s mind, or more generally, on phenomena being there, but outside of our mind. But there is also another option, and this gives rise to

⁵ ⁶ ⁷ ⁸

For a position like this, see (Yetter-Chappell, 2018). Most famously, see (Berkeley, 1710). See (Robinson, 2020) for a survey of various arguments for that conclusion. For a congenial approach, see (Smithson, 2021).

forms of idealism we should reject 27 the main alternative way of formulating phenomenalism. Instead of constructing material objects from phenomena that are there, one could try to connect them to phenomena that would be there if observation were to occur. Objects are then tied to hypothetical phenomena, not actually occurring phenomena. We can thus call categorical phenomenalism the view that objects are constructed from actually occurring phenomena, be they in our minds or elsewhere. This contrasts with hypothetical phenomenalism, which associates objects with phenomena that would occur if the objects were to be observed. Categorical phenomenalism looks unpromising, as we saw just above. But there are also several different ways to be a hypothetical phenomenalist, some of which look much better. Somewhat popular during the first half of the twentieth century was what we can call analytical hypothetical phenomenalism. It holds that there is a close connection in meaning between sentences about what is the case and sentences about what would be observed if someone looked. For example, ‘There were rocks 10 million years ago’ is claimed to be equivalent in meaning to ‘If someone would have looked 10 million years ago, they would have seen rocks’. That equivalence reveals that ordinary talk of material objects is simply disguised talk about potential experiences, and in light of this we can see, via the analysis of our talk about objects, that it is ultimately merely about potential experiences, not about independent material objects.⁹ But I have a hard time taking this proposal seriously. If the idea is that the analysis of language will reveal that talk of objects evaporates into talk of potential experiences, then such an analysis should withstand linguistic scrutiny. But there is no way such a proposal will survive as a proposal about natural language. If there is a connection at all between sentences about material objects and counterfactual conditionals concerning possible experiences, then this connection does not obtain at the level of language, as something that the meanings of words in natural language would determine. If there is such a connection at all, it must have a different source, not one in language itself. The analysis of language is no route towards phenomenalism. A more promising approach towards this relationship is less linguistic and more metaphysical. Let us call reductive hypothetical phenomenalism the view that somehow the material world is reduced to or identified with constructs of hypothetical phenomena. Here one could hold that the fact that there were these rocks is identical to the fact that if someone would have looked, they would have seen them. This avoids considerations about meaning, but it still is a real stretch to defend this, even as a claim about fact identity rather than sameness of meaning. After all, the corresponding facts are about different things. More promising is the route originally suggested by John Stuart Mill in (Mill, 1865)

⁹ See (Ayer, 1946) or (Lewis, 1946) for versions of this approach, and (Chisholm, 1948) for criticism.

28 idealism and the harmony of thought and reality and worked out in some detail by contemporary hypothetical phenomenalist Michael Pelczar: identify material objects with potentials for experience.1⁰ On this proposal, the apple is identical with the potential for apple experiences. Thus the apple is not a bundle of experiences, but identical to the potential for certain kinds of experiences itself. Material objects, on this proposal, are thus things of a special kind: they are potentials, and potentials for experience in particular. We can call this view potentials phenomenalism. On orthodox materialist views objects have potentials to produce experiences, and there is an association of some kind between certain objects having certain properties and that object having the potential for producing certain experiences: (1) All things being equal: Fido is a dog iff Fido has the potential to produce doglike experiences. But potentials phenomenalism goes further by not merely acknowledging a connection like this, which basically everyone can accept in some form or other, but by identifying the object with the potential itself: Fido is identical to the potential to produce certain experiences. Thus while almost everyone can accept that objects have the potential for experiences, the potentials phenomenalist holds that they are the potential for experience. This version of phenomenalism avoids some of the problems of categorical phenomenalism. For example, it has no problem with unobserved objects, since potentials for experience can be there without being actualized, that is, without producing experiences. Potentials phenomenalism does better than the other versions of phenomenalism considered so far, but it can’t be right either. First of all, potentials phenomenalism must at least prima facie seem metaphysically dubious, since it has the material world as consisting of free floating potentials: potentials which are just there, while their being there is not explained by, or grounded in, anything else. In particular, on the natural way of understanding the position, there are no “categorical” features of the world which explain why a certain potential is present. The potentials are simply there, without anything explaining why they are there and why it is just those potentials that are there.11 Whether or not it is in the end justified to demand that any potential be derivative on something categorical is not completely clear, and several philosophers have

1⁰ See, in particular, (Pelczar, 2015) and (Pelczar, 2019). 11 An interestingly different argument for idealism connected to this very issue is given by Robert Adams in (Adams, 2007). Adams argues, simply put, that causal powers and potentials must be grounded in categorical properties, but the only categorical properties we know of are mental ones. Thus we have reason to believe that mental features ground all other ones. I personally don’t quite follow the reasoning in Adams’s article for why the categorical features should be expected to be mental ones. Still, Adams’s route to idealism is an intriguing and original one, although we won’t be able to investigate it further here.

forms of idealism we should reject 29 denied it.12 How serious of a problem free-floating potentials are is controversial, but they surely are prima facie problematic.13 All this might be bad enough, but the real problem with the view lies elsewhere. The real problem with potentials phenomenalism is this: if the object just is the potential for experience and these potentials are fundamental, then there can be no explanation of why the object has the potential for experience. But in many cases we know exactly why a given object has such a potential. Thus potentials phenomenalism isn’t compatible with what we know to be true. After all, what we know does not only include which states of affairs obtain, but often also why some states of affairs obtain. That is to say, we know of causal and explanatory relationships. But potentials phenomenalism isn’t compatible with certain kinds of these explanations, and thus we have good reason to reject it. Let me develop this point first more generally and then with an example. If objects are potentials for experience and if nothing grounds this potential, then certain kinds of explanations of why this potential is there are impossible. As is common, we can distinguish causal from constitutive explanation. To illustrate, the causal explanation of why the wall is white is because Sue painted it white. The constitutive explanation of why the wall is white is that it reflects light in this and that way.1⁴ If there is a constitutive explanation of the wall being white, then it isn’t fundamental that the wall is white: the grounds cited in the constitutive explanation are more fundamental than the wall being white. However, there being a causal explanation of the wall being white is not in conflict with it being fundamental that the wall is white, only a constitutive explanation of this fact would be. Fundamental things can figure in causal relations, but they can’t have constitutive explanations of why they are there; they are supposed to be fundamental after all. Thus potentials phenomenalism is incompatible with constitutive explanations of certain potentials for experience. However, we do know of many 12 See, for example, (Mumford, 2006) and (Vetter, 2015). 13 In (Pelczar, 2019), Michael Pelczar hopes to defend potentials phenomenalism from such objections by arguing that “not only is it metaphysically possible for a potential to exist without any categorical basis, we also have reason to believe, or at least not to disbelieve, that many of the potentials that actually exist have no categorical basis.” His example to illustrate this are the probabilities of radioactive decay. Pelczar claims that the fact that radon atoms have about a 50% chance of decaying in four days is not grounded in anything categorical, and that there are categorically indistinguishable worlds from ours where radon has a 90% chance of decaying in four days. I dare say that this is mistaken. That radon has the half life it has is explained by the structure of the nucleus of radon and the various forces that are active at the nuclear level. This fact thus has a perfectly good explanation and with it likely a categorical basis. What has no explanation is why this particular radon atom decayed now. But that radon itself has a certain half-life is perfectly explainable on the basis of what likely are categorical features, and thus it is not a free-floating potential. 1⁴ Some philosophers use the term ‘causal explanation’ more broadly to include both of these explanations of why the wall is white. For example, Marc Lange in (Lange, 2016) uses ‘causal explanation’ for any explanation that concerns the causal aspects of something. He does this mostly to have a very strict notion of non-causal explanation, which is then used to show that even with such a strict notion there are non-causal explanations. I will use the term ‘causal explanation’ in the more common narrower sense of citing the cause of a particular event or state of affairs.

30 idealism and the harmony of thought and reality such explanations. Let me illustrate this with an example: the sweetness of maple syrup. We know why the maple syrup has the potential to give rise to sweetness sensations. It contains sugar, which affects the taste buds in a certain way to produce the experience of sweetness. This is why the maple syrup has the potential for certain sensations. This is not a causal explanation, but a constitutive explanation. If maple syrup were a fundamental potential for experience, then such a constitutive explanation of why it has that potential should be impossible. Potentials phenomenalism is thus not compatible with what we know to be the case. It is not enough to defend potentials phenomenalism from such an objection by holding that the maple syrup has the potential for tasting sweet, since it contains something which is the potential for tasting sweet, namely sugar. First off, maple syrup contains all kinds of things, but those things don’t affect its potential for phenomena: it also contains some salt, but doesn’t taste salty. Simply pointing to parts is not much of an explanation. The proper explanation of why it tastes sweet, but not salty, is not tied to its having some potentials, but not others, as parts, but rather a constitutive explanation of the chemical make-up of maple syrup: what molecules make up the syrup in what ratio and what effects they have. A non-causal explanation of why the syrup has a certain potential for experience is not something a potentials phenomenalist can accept. The potential for experience was, after all, supposed to be the fundamental building block of the material world, and thus can’t have constitutive explanations. Potentials phenomenalism is not merely supposed to be a minor modification of the standard materialist picture, where matter is identified with potentials for experience, but the rest is left alone. Instead, it has the ambition to be a completely different outlook onto the world, one that is closely tied to the problem of explaining how the mental can arise in a material world. The new outlook is supposed to be tied to material objects themselves being potentials for experience, and it is supposed to be compatible with there being no way to explain how minds can arise from mind-independent matter.1⁵ But it is hard to see how this ambition can be fulfilled. Leaving aside the above objection, it is not clear how material objects being potentials for experiences makes progress on how minds relate to matter. The problem here is simply that minds do not arise from the potential for experience directly, but only when that potential is actualized and thus experience happens. But then the question remains why the potential for experience, which is the material object, only gets actualized when it interacts with creatures like us, but not in general when one material object interacts with another one. What is it about us and our brains that actualizes a potential for experiences and thus leads to experiences? That the object is the potential for experience does not answer this question. What would answer it is a detailed account of how experience arises from

1⁵ See (Pelczar, 2019) for an endorsement of this goal for phenomenalism.

forms of idealism we should reject 31 the interaction of brains with matter, but such an account isn’t given by mentioning that the material objects is a potential for experience. To propose that doesn’t help explain why experience arises unless it can be explained why the potential gets actualized by us. And to do that one needs to solve the mind-body problem the hard way, just as the materialist needs to do. All of these issues obviously deserve much further discussion. The discussion above can only be an outline of why it seems to me that phenomenalism is mistaken. Any viable form of idealism must be compatible with what we know to be true, and neither categorical nor hypothetical phenomenalism meet this requirement. But there are also many other ways one could be an idealist, and I hope to highlight several other ones in the remainder of this chapter. Phenomenalism is generally motivated from considerations tied to perception, consciousness, or some alleged problem connected to mind-independent matter. Other versions of idealism are motivated from very different angles, and lead to very different positions. Our next candidate for idealism—conventionalism—sees neither mindindependent matter as problematic, nor is it motivated from considerations tied to perception or consciousness. Nonetheless, it, too, would support the metaphysical centrality of minds in the world.

2.2 Conventionalism Conventionalism is an umbrella term for a class of positions in philosophy that certain philosophically relevant facts are due to us in some usually contingent way. The view can be proposed for logic, objects, essences, and lots of other things, and it is quite clearly true for some limited cases. Although it is unclear whether conventionalism is true for being an object in general, it is quite clearly true for certain particular kinds of things: a dollar bill, a referee, and so on. However conventionalism is to be understood more precisely, these limited cases should turn out to be cases where conventionalism applies, since it is contingently due to us what a dollar bill is and who is a referee. But those cases do not motivate idealism or any other larger metaphysical view. They merely point out that some aspect of our local, social world are due to us, which is uncontroversial. To associate conventionalism with idealism, it would have to apply on a grander scale. In this chapter I would like to discuss one form of conventionalism— conventionalism about material objects—as an example of an attempt to have conventionalism apply to something grand and not merely a local part of our social world. This kind of conventionalism comes in at least two versions. One is concerned with composition: when do certain chunks of matter make up an object? The other one is concerned with persistence: when do objects remain the same over time? The former is intended as an answer to the “special composition question” of (van Inwagen, 1990): under what conditions do material things

32 idealism and the harmony of thought and reality compose an object that has these things as parts. To illustrate: the pre-theoretically most plausible answer is that sometimes, but only sometimes, do things compose a larger thing that has them as parts. Let’s assume for the sake of the argument for now that the basic building blocks of matter are simply atoms. On the pretheoretically most plausible view not all collections of atoms make up an object, but some do. The atoms that are part of the rock do, but a random collection of some rock atoms, some leave atoms, etc., do not. The question is: which ones do make up an object, and what justifies that answer? This is a widely discussed problem,1⁶ and the general options for an answer are: all, some, or no collections of atoms compose an object. In the metaphysical debate about composition the two extreme views—composition always or never occurs—are rather popular. The intuitively more plausible middle view is seen as hard to defend, since it is difficult to find a good reason for why just certain collections, but not others, compose an object. Anyone who does not take one of the extreme answers will have to make clear not just for which collections of atoms composition occurs, but they also need to justify why the cutoff is significant for composition to occur. If one just looks at features of the collections of simples themselves, then this has proven to be hard to do.1⁷ A conventionalist about composition can be seen as aiming to answer this challenge. On the conventionalist proposal it is due to our contingent conventions that certain collections of atoms compose objects, while others do not. In particular, those that correspond to the “ordinary objects” of everyday experience—tables, rocks, trees, etc.—do indeed compose such an object, and these objects thus exist. It is due to us and our use of concepts and language that some groups of atoms make up a further thing, while other ones do not. It is thus due to us and our conventions which objects exist. Obviously, by a ‘convention’ one can’t simply mean some explicit agreement. Instead the term ‘convention’ is used broadly in this context, to indicate some aspects of our collective mental activities that make this happen: our use of concepts, or our contingent psychological make-up that groups things together mentally, and so on. This form of conventionalism about composition will be our main example of conventionalism in this section. A second form of conventionalism about ordinary objects concerns not composition directly, but persistence over time. For all regular objects there are some conditions under which they would continue to exist in a changed form, and some conditions under which they would cease to exist altogether. The tree continues to exist after losing a leave, but it ceases to exist after being turned into sawdust. Or to put it differently: there are some features of a tree that are essential to it, and without which it can’t continue to exist, and others which are accidental,

1⁶ See (Korman, 2016) for an overview of this debate. 1⁷ Nonetheless, there are numerous attempts to do it, for example (Korman, 2015) and (Thomasson, 2007). I won’t aim to evaluate them here, but focus on the conventionalist alternative instead.

forms of idealism we should reject 33 and which may change while the tree continues to exist. The conventionalist about persistence holds that under what conditions which objects persist, and thus what their essences are, is due to us and our conventions.1⁸ These two forms of conventionalism can be seen as two independent issues, but see (Sidelle, 2010) for an argument that they are closely connected. Be that as it may, our main focus will be conventionalism about composition. In (Sidelle, 2010, 109) Alan Sidelle notes that neither of these conventionalist views is “. . . in any interesting way idealist—the ‘material’ of [. . . ] individuals is taken to be wholly mind-independent . . . ”. And on a common, more narrow way of understanding ‘idealism’, this is certainly correct. On that way, idealism requires that matter itself is mental, and this does not need to be so for conventionalism: the atoms can be there without any contribution from minds. Nonetheless, conventionalist views about objects are idealist on a broader understanding of the term. In particular, our minds would make a crucial contribution to what exists, since they would be responsible for which atoms compose an object, and thus for which composite objects exist. This would vindicate strong ontological idealism in our above broad sense, since our human minds would be metaphysically central for reality understood as all that exists. Conventionalism is especially illustrative for how our minds might be central to reality, even if matter itself is not mental. It makes clear how minds might affect reality to vindicate idealism in the broad sense, but in a completely different way than phenomenalism aimed to do. Consciousness and perception play little role in it, while purely metaphysical considerations about composition and persistence take central place. The question remains whether conventionalism is indeed defensible, in particular whether we can make sense of our minds being responsible for composition occurring and for certain objects existing. And here things don’t look too good. Who precisely is a conventionalist, and if so in what sense, is obviously controversial, with the most famous candidates being Rudolf Carnap and Nelson Goodman.1⁹ But I would like to focus my discussion on the more recent and more straightforward version of conventionalism defended by Iris Einheuser in (Einheuser, 2003) and (Einheuser, 2006).2⁰ On Einheuser’s view, the world of objects as we know it is the result of two things: a substratum and a carving of the substratum. The substratum is what the 1⁸ See (Sidelle, 1989) for this kind of position. 1⁹ See (Carnap, 1956) and (Goodman, 1978). Conventionalism should be distinguished from linguistic determinism, also known as the Sapir-Whorf hypothesis. Conventionalism, in one version, holds that our language or our thought shapes the world. Linguistic determinism holds that our language shapes our thought and how we think about the world. Thus the Sapir-Whorf hypothesis is not one about the language-world relationship, but about the language-thought relationship. Although there are many who accept the Sapir-Whorf hypothesis, they are thereby not conventionalists in our sense. 2⁰ A related view is the ‘mereological idealism’ defended by Kenneth Pearce in (Pearce, 2017).

34 idealism and the harmony of thought and reality world itself contributes: the raw material, which might simply be the distribution of atoms in spacetime, or it might be atomless matter somehow distributed. The carving of the substratum into a world of objects is our doing, due to our conventions. It determines which collections of atoms or chunks of matter compose an object. The world can then be associated with a pair of a substratum and a carving of that substratum: < s, cs >. We could have carved the substratum in different ways, leading to a different final result. But we couldn’t just have carved any things out of any substratum: the substratum must support a particular carving, there must be the right raw materials there to be carved a certain way. Since the final result of what the world is like and what objects there are depends on how we carve the substratum, this view is a form of strong ontological idealism: our minds are metaphysically central for what exists. It is because of us and our minds that the world contains the objects it contains. Is such a view defensible? In particular, does it meet our three constraints of explicitness, support, and compatibility? Here the explicitness and the compatibility constraints work against each other. There is a tension between how to formulate the view as it is intended and whether it is compatible with what we know. In particular, it is hard to make sense of it being due to us and our conventions that the world is the way it is, while at the same time keeping the view compatible with what we know to be true. One can either formulate the view such that it is compatible with what we know, but then it becomes hard to see how our minds are metaphysically central, and in what sense the world depends on our conventions. Or one can formulate it in such a way that the world depends on our conventions, but then it is not compatible with what we know. There is a real tension between having an explicit formulation of the view as it is intended and meeting the compatibility constraint for the view so formulated. We can label this general problem the explicitness-compatibility tension, which is not restricted to conventionalism, but applies to other forms of idealism as well. Although much of Einheuser’s discussion is focused on conditionals, to be discussed shortly, the basic idea of her view can be illustrated with how it can make sense of the truth of statements like (2) There were rocks before there were humans. In general, statements are evaluated at substratum-carving pairs, with < s@ , c@ > being the pair of the actual substratum with the actual carving we imposed on it. Evaluated at this pair, (2) is, of course, true. < s@ , c@ > just is the world with the objects in them as we normally take them to be, including very old rocks. That is easy enough, but how should we then understand that the carving, and with it the rocks, depends on us? By itself the pair < s@ , c@ > is just a representation of the world, in terms of what atoms there are, as the first element, and which objects they form, as the second element. That so far says nothing about us playing any role in this, despite using the letter ‘c’ for ‘carving’. Everyone can accept that the world as

forms of idealism we should reject 35 we normally think of it can be represented by such a pair. The conventionalist also needs to spell out how the carving depends on us, and thus how it is a true form of conventionalism. And this Einheuser aims to do with conditionals. In particular, she argues that a subjunctive conditional like (3) If our conventions had been suitably different, then there would have been no mountains. can be understood in at least two different ways; one of which is false, while the other one is true. Since one and the same sentence can be used and understood in more than one way, we can say that such a sentence has more than one reading. One of these readings, according to Einheuser, involves the counter-substratum conditional. Its truth conditions concern what happens when we keep the carving c@ fixed, but change the substratum s@ . In particular, (3) is true just in case if we change s@ in the minimal way to s′ such that < s′ , c@ > makes the antecedent of the conditional true, then < s′ , c@ > also makes the consequent true. On this reading (3) is false. The smallest change of the substratum that makes the antecedent true changes the matter that makes up human bodies a little bit, but leaves the rest alone. Since the carving remains the same, there would still be mountains where there are mountains now. The second reading is the counter-conventional conditional. Here we fix the substratum, but change the carving. In particular, on this reading (3) is true just in case if we keep the substratum s@ the same, but change the carving from c@ in a minimal way to c′ so that < s@ , c′ > makes the antecedent true, then < s@ , c′ > also makes the consequent true. And so understood, Einheuser holds, the conditional is true. If the actual substratum were carved up sufficiently differently, then there would be no mountains. We can thus distinguish two kinds of dependence: substratum-dependence and carving-dependence. And we can spell them out explicitly with the respective reading of the counterfactual conditional. Mountains substratum-depend on us just in case the conditional (3) is true in its counter-substratum reading, and mountains carving-depend on us just in case the counter-conventional reading of (3) is true. Mountains then do not substratumdepend on us, but they do carving-depend on us. And that formulates and supports conventionalism, or so Einheuser. To properly evaluate Einheuser’s proposal, it is important to be clear about what it says. There are two central parts to it: The first is a formal semantic machinery, namely thinking of the world as a pair < s, c > and how to use such representations in the semantics of conditionals. The second is a claim about subjunctive conditionals having two different readings. The formal machinery illustrates how the two different readings are supposed to work, but by itself this machinery is philosophically completely neutral. Anyone can hold that the world can be represented by a pair < a, o > where the first member is all the

36 idealism and the harmony of thought and reality atoms and the second member is all the objects composed by those atoms. Such a representations will be incomplete, of course; there is more to the world than what atoms there are and what objects they compose. But it can uncontroversially be an accurate, although incomplete, representation of the world. Nothing about conventionalism follows simply from employing this machinery, although something about conventionalism is clearly suggested by calling the second member of the pair a ‘carving’ and using the corresponding letter ‘c’. Furthermore, it can also be granted for present purposes that if subjunctive conditionals do indeed have these two readings, then they can be modeled using the proposed formal machinery. Obviously, whether truth conditions like the ones specified by Einheuser correctly capture subjunctive conditionals in general is controversial, but this is not the controversy that matter for us here. The real issue for us is whether conditionals indeed do have these two readings. And here things don’t look too good for the conventionalist. To bring out the key aspect of Einheuser’s proposal, it is important to distinguish her key claim—that subjunctive conditionals have a counter-conventional and a counter-substratum reading—from the uncontroversial claim that subjunctive conditionals can have antecedents that concern either conventions or the substratum. The conditional (3) makes perfect sense, and so does the one that makes a claim what would have been different had matter been arranged differently. Einheuser’s claim goes beyond that in that she holds that the subjective conditional has two readings, i.e. that the very same sentence (3) can be used to make two different assertions, tied to different versions of the subjunctive conditional. It is this claim that is key for her proposal, and that is the truly controversial part of it. Conditionals are generally accepted to have a variety of readings. Most prominently, there are indicative and subjunctive conditionals, which are generally differentiated by the indicative or subjunctive mood in the antecedent: ‘if I am . . . then I will . . . ’ as opposed to ‘if I were . . . then I would . . . ’. That there is a difference in the conditional tied to this difference in mood is widely accepted. Furthermore, it is widely defended that conditionals are context sensitive, and thus the truth conditions can vary with context. But both of these claims are quite different than Einheuser’s proposal of different readings of (3). Here it is the very same sentence that has different readings. These readings are thus not connected to the mood of the antecedent. And these readings are not taken to be derivative on context sensitivity. They are supposed to be two distinct, but so far neglected, readings of the subjunctive conditional itself. But why would we think that there are these two readings? There are in essence two main ways to argue that certain sentences have more than one reading: one is to give examples to bring out these readings, the other is to motivate the presence of these readings from general considerations about the needs we have in communication. It is hard to see how either one motivates that there are these two readings in conditionals. I doubt many would recognize

forms of idealism we should reject 37 two such readings of (3). There surely can be disagreement about the truth of (3), but I doubt many ordinary speakers would see two readings in this sentence, in particular including one that makes it true. That ordinary speakers recognize these readings certainly isn’t required for them to be there. But if they did, that would be evidence for them being there, and without such recognition there is little evidence from that source. There is also little reason to think that there are these readings from general considerations about the need we have in communication. In particular, the counter-conventional reading would be a distinct reading of the subjunctive conditional that arises only in very limited context: subjunctive conditionals with antecedents that involve conventions. The counter-conventional reading of the subjective conditional thus has a very limited role in communication; it only figures in assertions about what would be different if our conventions were different. It is hard to see why the conditional itself has a distinct reading just for those cases. Not only is talk about such counter-conventional situations rare, it is also only of special philosophical significance, and so does not play a broader role in communication. And what’s more, the counter-conventional reading is only of any real use if conventionalism is true. If conventionalism is false, then the counter-conventional reading is essentially always false. Thus there would be little point of having both readings. But conventionalism is a substantial metaphysical claim. Why would our language have a distinct reading of conditionals that in essence is only of use if conventionalism is true, and that is only of use to talk about conventionalism and the dependence of the world on our conventions? But for conventionalism as spelled out by Einheuser to make sense there must be such a counter-conventional reading of the conditional: it must be part of our shared language that the conditional has these two readings. And from what we have seen so far, there is little reason to think that this is the case. Einheuser’s proposed account of conventionalism is promising, since she aims to spell out how the world depends on our conventions with a notion of dependence that is based on an alleged and so-far neglected reading of the subjunctive conditional. But I argued that we should be skeptical about natural language conditionals having such a distinct counter-conventional reading. Without the counter-conventional reading of the conditional being a part of our shared natural language, it must be seen as a proposed addition to it. And with that we must think of the notion of convention-dependence as an addition as well. But so understood it is not a satisfactory way to make explicit what the conventionalist view comes down to. It would essentially turn into a proposal that there is a novel sense of dependence according to which what objects there are depends, in this proposed sense, on us. Dependence of objects on us is then taken to be compatible with there having been rocks long before humans, since although rocks do not depend on us in the familiar senses of dependence that we can express in our shared natural language, they do depend on us in the novel sense of dependence. All this is not

38 idealism and the harmony of thought and reality enough to meet the explicitness constraint. To meet it, one needs to spell out the notion of dependence in generally accessible terms, not via a stipulated notion of dependence. This novel notion of dependence might make the dependence of rocks on us compatible with what we generally know, but this is mostly because it is stipulated to do so, not because it explains it. And that is not enough to solve the dependence problem discussed above on page 11. To do this we need make the notion of dependence explicit, and not merely add an unexplained novel notion. On the other hand, with notions of dependence that we can clearly articulate with our shared conceptual repertoire, it uniformly seems to be false that objects in general depend on us and our conventions. This is the key tension for conventionalism: how can we, on the one hand, make sense of a notion of dependence according to which the existence of rocks depends on us and our conventions, but, on the other hand, rocks have been there long before humans. Einheuser’s proposal does not solve this problem, I have argued, and neither does any of the other conventionalist proposals that I know of. Thus I see little hope for conventionalism about material objects being correct, or even a clearly articulated position. If our minds are central in reality, it has to be in some other way.

2.3 The subjectivity thesis Less well-known than phenomenalism and conventionalism, but just as interesting, is a position defended by Anton Friedrich Koch, which he calls the subjectivity thesis.21 The subjectivity thesis states that subjects are required in a world of material objects. That is, a world containing material objects, but no minds, is impossible. For there to be material objects at all, minds have to be there as well and thus a mindless material world is impossible. We will discuss this thesis in this section, including especially Koch’s argument for it. But even before seeing the details, we can already note how the subjectivity thesis leads to a different kind of idealism than the positions discussed so far. It would not necessarily lead to idealism in a more narrow sense, where matter itself arises from the mental. The subjectivity thesis can be true even though matter does not arise from minds. It merely requires that there are minds as long as there are material objects. These minds might well be derivate on matter, as long as they are there. And they might as well only exist at a later time, as long as they exist at all. In fact, we have been continuously moving away from the narrow conception of idealism exemplified by classical idealism. Phenomenalism is closest to classical idealism and holds that matter itself is mental or at least directly tied to minds. Conventionalism goes further in that for it the basic building blocks of matter, the

21 See (Koch, 1990), (Koch, 2006a), (Koch, 2006b), and (Koch, 2010).

forms of idealism we should reject 39 substratum, is not necessarily mental, but which chunks of matter form an object, and thus which objects there are, is directly due to our minds and our conventions. The subjectivity thesis takes it even further: neither matter itself nor which objects there are has to be due to minds, but nonetheless, there being minds is required for there being material objects at all. If that were so, then minds would play a metaphysically central role in reality: they make the existence of material objects possible. Intriguingly, Koch hopes to establish this with a purely metaphysical argument. I would like to present this argument in the following in my own words, differing slightly from Koch’s own presentation in terminology and setup.22 After presenting the argument for the subjectivity thesis, I will make clear why I think that this argument is mistaken. The subjectivity thesis holds that (ST) Subjects necessarily exist in a spatiotemporal world.23 A subject here is someone with a mind who is part of the spatiotemporal world and relates to it in perception and thought. We are subjects in this sense, but other creatures can be as well. If (ST) is true, then a spatiotemporal world without minds is impossible. Subjects, and thus minds, must exist for there to be such a world. Since material objects exist in spacetime, subjects need to exist for there to be material objects, and thus minds are metaphysically central in reality. (ST) is therefore a form of idealism in our broad sense. The question remains how one might possibly think that the existence of minds could be required for the existence of material objects. On Koch’s argument the basic idea is this: subjects are central for the grounds for facts about identity and difference. On this line of thought, facts about identity—that a is identical to b—and facts about difference—that a is different from b—are not brute facts. Such facts obtain because certain other facts obtain. And these other facts, which are the grounds for identity and difference, involve subjects, for a reason to be spelled out shortly. What facts could ground identity or difference? At first the natural candidate for such facts are facts involving properties: a is different from b because a has a property that b lacks, and a is identical to b because a and b share all of their properties. If this is on the right track, then the strategy of an argument could be that it is properties tied to subjects that in the end must do this grounding. Now, this setup is clearly not uncontroversial. There are at least two major concerns with it. First, why should we think that there need to be grounds for 22 Koch’s own presentation of the argument can be found in his works cited above. For a more detailed presentation of the version given here, see (Hofweber, 2015). 23 That is, a spatiotemporal world without subjects is impossible, not that subjects can only exist in a spatiotemporal world. Koch also holds the latter thesis, which he calls the personality thesis, see (Koch, 2010). We will only discuss the subjectivity thesis here.

40 idealism and the harmony of thought and reality a and b being different? Why does there need to be something like a metaphysical explanation of their difference? On the most natural way of asking why a and b are different, one is asking for an explanation of their qualitative difference: why do they have different properties, not why are they numerically different. Or one is asking for a causal explanation: why did they end up different? None of this would work for Koch’s argument, as the issue concerns numerical identity and non-causal explanation. Second, isn’t the fact that a has a property that b lacks better seen as conclusive evidence that a and b are different, rather than something like a metaphysical ground or explanation of their difference? If so, then the support that the fact of their having different properties provides for them being different is epistemic, not metaphysical. Although I agree that these are legitimate and good concerns, I would like to leave them aside for now. Reasonable people can clearly disagree about these issues, and even for those who accept that grounds for identity and difference facts can be demanded, it is completely unclear how this could lead to an argument for idealism. We should thus for now go along with the basic setup of demanding grounds for identity and difference, of connecting these grounds to the having of properties, and then see how one might turn this starting point into an argument for idealism. If we accept this basic setup, we should next investigate what properties might be tied to the grounds for these identity and difference facts. A first attempt of an answer can be object-dependent properties. a is different from b because a has the property of being a, while b lacks it. But this quite clearly can’t be the answer. The question will just be pushed back to why the two properties being a and being b are different properties. The natural answer to that is because a and b are different objects. And thus the order of priority goes the other way round: the object-dependent properties are different because the objects are different, not the other way round. A second attempt is to hold that a and b are different simply because they differ in some purely qualitative property. A purely qualitative property here is simply an ordinary property like being round: a property that is not object-dependent and that can in principle be shared by other things. Thus for any two things a and b there will be some regular property that a has, but b lacks. And for different objects it can be different properties, although for any pair of different objects there will be some property or other in which they differ. And also the other way round: the identity of a and b is grounded in them having the same general, qualitative properties. In other words, numerical identity is determined by qualitative identity: it is having all the same qualitative properties that determines numerical identity. But this can’t be right, since it is possible for different objects to share all their purely qualitative properties. It is possible that the world is completely symmetrical and that there is a duplicate of every object on the other side of some center of symmetry. In one famous example, the world might contain only two iron spheres

forms of idealism we should reject 41 of the same size.2⁴ Qualitative properties by themselves thus can’t ground facts of identity and difference in general. There is a real issue here whether one could hold a kind of a disjunctive view: qualitative properties do ground identity and difference if there are no symmetries, and if there are symmetries, then something else does that job. But this has the ring of claiming that the proposal applies when it works, and when it doesn’t work some other proposal will apply. In general it would be preferable to have a uniform account of what grounds these facts, not one split into two cases. And this is just what Koch proposes, and I will thus leave this disjunctive approach aside and consider Koch’s positive uniform proposal instead. If neither purely qualitative properties nor object-dependent properties are the grounds for identities and differences, but properties nonetheless are what explains the sameness and difference of material objects in spacetime, then what properties could be the ones that do this work? And here one can see the proposal by making clear that we do not have think about this situation from nowhere, as if we described the world from the outside, while not being a part of it. From the outside, the symmetric world does indeed seem like as if nothing makes a difference: no properties can break the symmetry. But from the inside there is an obvious difference. If I describe the world from the inside, with a particular perspective onto the world, then there will be the following difference: that cup in front of me is right here, whereas its twin, the cup on the other side of the symmetry, is way over there. One can take such a perspective to be of only epistemic significance, or one can hold that it has true metaphysical significance. On the former, my perspective onto the world is merely my way of gaining information about the world and my personal way of accessing the world. But the perspective is not part of the world itself, only an aspect of my epistemic situation. On the more metaphysical understanding of perspectives, my perspective on the world has metaphysical significance, since the perspective is an aspect of the world itself. Thus when I see the cup in front of me, I do not only find out where the cup is located relative to me. I also find out that it has a special property of being here, which we can call hereness. Some other, far away object instead has the property of being way over there, or thereness for short. Thus from within, claiming metaphysical significance for my perspective, there is a difference even in a symmetric world: my cup has the property of being here, while any other, qualitatively identical cup has the property of being over there. The difference between the two cups can then be grounded in one having hereness, while the other has thereness. Koch calls properties like the property of being here and being there indexical properties. Even though the use of ‘indexical’ is often reserved only for features of representations, not the properties themselves,

2⁴ See (Black, 1952) and (Adams, 1979) for a discussion of the two spheres example, and (Hofweber, 2005) for an argument that the identity of objects is not determined by purely qualitative properties even if complete symmetry is ruled out.

42 idealism and the harmony of thought and reality I will nonetheless follow Koch’s use of the terminology here. Indexical properties like hereness are to be understood as not being identical to properties like being at location L. The property of being here is not the same, on this proposal, as the property of being in front of TH. Instead, it is a primitive further property. This is analogous to certain temporal properties on a tensed theory, or A-theory, of time.2⁵ The property of being present, on such views, is not identical to the property of being at 12:53 pm, June 29, 2020. Rather it is a further, primitive temporal property. And similarly, the property of being here is a further, primitive property, not identical to the property of being in Chapel Hill. It is these indexical properties that break the symmetry and are responsible for grounding identities and differences, or so Koch’s argument. It is because subjects have a perspective onto the world from within that objects have indexical properties. The subject is required for there to be a perspective onto the world, and that in turn is required for any object to have indexical properties. It is because of subjects being the sources of indexical properties like hereness that subjects are required in a spatiotemporal world of objects. The need to be there somewhere and at some time, no matter when, to guarantee a breaking of the symmetry. Without them the needed grounds for facts of identity and difference could not be guaranteed. And without such grounds there can be no material objects, or so Koch’s argument for the subjectivity thesis in a nutshell.2⁶ But this argument can’t be quite right, even if we grant the need for grounds for identity and difference and the coherence of indexical properties. The issue is not simply resolved by adding indexical properties to the mix, and by taking subjects to be the source of indexical properties. It is true, uncontroversially, that from my own perspective the symmetry is broken: one side of the symmetry is special and distinguished, the one I am at. But that is so far just a distinction for me, not by itself a distinction that gives us properties that make the difference. The real difference is to come from the indexical properties that objects have because of subjects whose interaction with the world somehow generates them. But in a truly symmetric situation, there will be another subject across the symmetry who will look at the cup in front of them, and who will also think that the cup is here. Thus both I and my duplicate will generate a hereness property for the cup in front of us. And with that our problem reappears: Are these two hereness properties the same or different? If they are the same, then the symmetry is not broken, since besides all the same purely qualitative properties, the cups would also have the same indexical properties, and thus indexical properties can’t ground that they are different cups. Furthermore, and maybe even worse, the cups would not just have hereness, they also both would have thereness. I generate hereness for the cup in front of me and

2⁵ See, for example, (Prior, 1967) or (Zimmerman, 2005). 2⁶ For a more detailed presentation of the argument and more discussion, see, in particular, the works by Koch cited above, as well as (Hofweber, 2015).

forms of idealism we should reject 43 thereness for the cup in front of my duplicate, whereas my duplicate does it the other way round. But hereness and thereness are naturally seen as incompatible properties: nothing can coherently have both of them at the same time. This is analogous to temporal properties of being past and being future according to an A-theory of time. Thus if the properties are the same, then it doesn’t seem to help and threatens to lead to incoherence. On the other hand, if the hereness properties of my cup and of my twin’s cup are different, then the question arises what grounds the difference of those properties? Are the properties relative to, or somehow to be indexed by, two subjects: TH and TH*? If so, then the hereness property I generate is either the property of hereness relative to TH, or it is the property herenessTH . Either way, the issue of the difference of the indexical properties will be derivative on the difference of the subjects that generate them. And if subjects are just further material objects, as Koch believes, then this can’t be the source of the difference of material objects in general. Thus by itself relying on perspectives and indexical properties won’t solve the problem. But that does not mean the argument can’t be saved. The criticism given above suggests that one can’t simply accept that each perspective on the world is equal. It would be sufficient to break the symmetry if one of the perspectives onto the world were distinguished and special. That, somehow, the world is always to be understood as the world from a particular perspective. We can call perspectivalism the view that any description of the world that isn’t given from a perspective within the world is bound to be incomplete.2⁷ Perspectivalism rules out a complete description from nowhere, but it faces the problem of dealing with the different perspectives that different subjects have and how they come together in one and the same world. Perspectivalism guarantees that no complete description of the world from nowhere is possible, but it does not guarantee that a complete and coherent description from within is possible. In particular, if there are more than one subject in the world, each of which has a perspective onto the world, can we then give a complete and coherent description of the world even from within? And can we employ perspectivalism to solve our problem about the grounds for identity and difference? Here there is a real tension: Either we downplay the perspectives and just see them merely as the source of more properties— the indexical properties—which arise equally for all perspectives. But then the symmetry is not broken. Or else we elevate the perspectives, and claim that the world comes with a distinguish perspective, which is the only one which gives rise to indexical properties. Then we break the symmetry, but we face the problem of justifying why this perspective, but not the perspective of my symmetric duplicate,

2⁷ Koch calls a somewhat different position the perspective thesis in (Koch, 2010). Koch addresses the (Hofweber, 2015) version of the argument, given here as well, in (Koch, 2016), where he also elaborates on his version of perspectivalism, which I understand to be similar to what I below call ‘solipsism light’. See also (Moore, 1997).

44 idealism and the harmony of thought and reality is the perspective of the world itself. Why think that my perspective, but not that of my twin, is the perspective of the world, in particular if we are both material beings in a material world? We seem to be pushed into a version of solipsism light, where I accept that there are others like me, but I can’t accept that their perspective on the world is just like mine. Their perspective is diminished, while mine is elevated. But why would I be so special that my perspective is the perspective of the world itself? In other ways I am not relevantly different from other subjects that have a perspective onto the world as well. It is thus hard to justify that although all subjects have a perspective on the world, mine somehow is the perspective of the world itself, while the other perspectives are not. Solipsism light is hard to hold onto. So far the view does not work, but there is one radical option left to consider: maybe all perspectives are elevated equally, and each perspective is the perspective of the world, equally. This would suggest that my cup has hereness and also thereness, even though these are incompatible properties. And that in turn suggests that there is no coherent and complete description of the world at all: if we mention all the indexical properties that are generated from all the perspectives, then that description ascribes incompatible properties to objects, and it is thus incoherent. Our final option then is this: we take all perspectives equally metaphysically serious, but we deny that reality is coherent. It might seem that this is going too far now. However, there is a way to make sense of it, one that is congenial to Koch’s own view. It deserves to be discussed separately, since it gives rise to special difficulties as well as several possibilities for formulating and defending idealism. The way to make sense of this radical option is fragmentalism, a position formulated and defended by Kit Fine in the philosophy of time, with applications to other areas, like the special status of a perspective onto the world. I will discuss fragmentalism in the next section, and evaluate there whether it can help with Koch’s argument for the subjectivity thesis and whether it is defensible on its own. But without saying more about how to understand the status of perspectives, I would like to conclude for now that either the argument for the subjectivity thesis requires solipsism light, and somehow one and only one perspective is metaphysically special, or else it leads to incoherence, since all perspectives are equally the source of indexical properties, objects have incompatible properties, and the world is incoherent. Neither of these options seems acceptable, but fragmentalism promises a way out of this difficulty.

2.4 Fragmentalism Fragmentalism is an intriguing and radical position about how reality as the totality of facts should be understood. It was proposed by Kit Fine in (Fine, 2005) with the primary purpose of defending a non-standard realist theory of tense. Although Fine’s primary focus in (Fine, 2005) is the philosophy of time, he also discusses several other applications of fragmentalism, including the metaphysical

forms of idealism we should reject 45 status of perspectives and selves. The main fragmentalist idea is that whereas standard views of reality hold that all the facts that obtain are coherent with each other, the fragmentalist denies this and holds that the totality of facts contains facts that are incoherent with each other. Instead of one coherent totality of all facts, we only have coherent fragments of reality: maximal sub-collections of all the facts such that their members are coherent with each other. There will be many different such fragments, each of which is internally coherent, but none of them are coherent with each other. To illustrate, let’s first consider a standard realist theory of tense. A realist theory of tense holds that there are irreducibly tensed facts, like the fact that a particular event, say the year 2020, is present. This fact is not identical to the fact that the year 2020 is the same year as the year when this token inscription was produced. The tensed language used to represent this fact is thus essential for its representation. Tense is therefore not merely a matter of a particular way of representing the world, but a feature of the world itself. A standard realist theory will hold that which tensed facts obtain changes over time. The fact that 2020 is present obtains this year, but next year it will stop obtaining, and a new fact will start obtaining: that 2021 is present. How such change among tensed facts themselves can be understood is notoriously problematic, and connected to McTaggert’s argument for the conclusion that such change is incoherent, and with it that time is unreal.2⁸ We won’t discuss any potential problems with a standard realist theory of tense here, but see (Fine, 2005) for more. Fragmentalism is a non-standard realist theory of tense. It is realist theory of tense, and thus holds that tensed facts irreducibly obtain. But it is nonstandard in that it holds that these tensed facts don’t change. Rather they all obtain: that 2020 is present, that 2021 is present, that 1781 is present, and so on. For each year, the fact that that year is the present year obtains. But any two of these facts are incoherent with each other: only one year can coherently be the present year. Fragmentalism accepts this consequence: facts obtain that are not coherent with each other. Thus the totality of all the facts that obtain is incoherent. But nonetheless, there is a sense in which reality is coherent. There are numerous coherent fragments of reality: one including the facts that 2020 is present and that 2021 is future, another one that includes the facts that 2021 is present and 2020 is past, and so on. Fragments are not merely coherent collections of facts, but maximally coherent ones: all the facts in a fragment must be coherent with each other, and no other fact could be added without losing coherence. If a fragment contains the fact that 2020 is present, then it is only coherent with certain other tensed facts: that 2019 is past, that 2021 is future, and so on. Reality can thus be seen in two ways: first as the collection of all the facts. Following Fine, we can call this über-reality. Or reality can be seen simply as a particular fragment: say the one that contains the fact that 2020 is present and all 2⁸ See (McTaggart, 1908) and (McTaggart, 1921) for the original argument, and (Markosian, 2016) for a short contemporary presentation. The argument is also presented in (Fine, 2005).

46 idealism and the harmony of thought and reality facts compatible with it. Über-reality is incoherent, since it contains facts that are incompatible with each other. Reality, understood as a fragment, is coherent, since it only contains facts that are coherent with each other. But in neither sense does reality itself change over time. There are simply different fragments which contain different facts about which year is present. McTaggart’s problem is avoided, but the coherence of reality overall, i.e. of über-reality, is given up. This is obviously only a rough outline of the main idea. But even so far, it is clear that there are some questions that should be answered. First, how should we understand the notion of two facts being coherent with each other? That 2020 is present is coherent with that 2021 is future, but not with that 2021 is past. One way to understand coherence of 2020 being present and 2021 being future would be in terms of that the two facts could both obtain together. But this won’t do, since both of those facts do obtain just as the fact that 2021 is past obtains: all of them are part of über-reality. It also won’t do to understand coherence in terms of both facts obtaining in the same fragments, since the notion of a fragment is understood in terms of coherence: fragments are maximal collections of facts that cohere with each other. Fine suggests in (Fine, 2005, 281) that the notion of coherence should likely to be taken as fundamental, primitive notion, not to be spelled out further. This, I will argue, is a source of a problem that makes fragmentalism vulnerable in a way that its alternatives are not. Second, why does fragmentalism not immediately imply contradictions? After all, the fact that 2020 is present obtains, and so does the fact that 2021 is present. But that 2021 is present seems to imply that 2020 is not present. So, both the facts that 2020 is present and that 2020 is not present obtain. Contradiction. A fragmentalist likely will on the one hand endorse this, since on their view reality is indeed contradictory and incompatible facts are part of über-reality. But on the other hand they will want to avoid such conclusions when understood as regular assertions. The natural move here, also adopted by Fine in (Fine, 2005), is that an assertion like “2020 is the present year” is to be evaluated not in über-reality, but in a particular fragment of reality. Thus an assertion is true when the expressed fact obtains in a particular, contextually determined fragment, not when it obtains in über-reality. And here it is natural to hold that an assertion made at a time t is to be evaluated at a fragment in which t is the present time. If this were so, then my assertion that 2020 is the present year would be evaluated in a fragment where the year in which the assertion is made is the present year, which would turn out to be correct, while my assertion that 2021 is the present year would turn out to be false. So understood, assertions are only evaluated in fragments, and since these are coherent there will be no correct contradictory assertions, even though reality contains incompatible facts.2⁹ Fine’s primary concern in his defense of fragmentalism was the philosophy of time and what the best form of realism about tense is. But Fine also discussed what he calls first-personalism: a form of realism about the first-personal aspect of our

2⁹ See also (Lipman, 2015) for more on fragmentalism.

forms of idealism we should reject 47 experience of reality, see (Fine, 2005, §12). This position is closely related to Koch’s perspectivalism discussed above, which gives not just epistemic, but metaphysical significance to the perspective we have onto the world. Perspectivalism by itself is merely a form of realism about our perspective onto the world. According to it, facts about what is here and there, what is close and far, etc., are analogous to facts about what is now or future on a realist theory of tense. And just as realism about tense can come in standard and non-standard forms, so realism about perspective can come in these forms. In particular, one form of non-standard realism about perspectival facts is a fragmentalist theory. And it is just the kind of approach that carries the hope of avoiding the problems raised for Koch above: how we can avoid solipsism, on the one hand, and accept that there are other subjects metaphysically just like me, but, on the other hand, have symmetry breaking, because my perspective is objectively special and distinguished. The fragmentalist solution to this problem is analogous to the fragmentalist solution to the problem of having a realist theory of tense that, on the one hand, avoids presentism and maintains that other times are just as real as the present, but, on the other hand, holds that the present is truly and objectively special. The solution was that, on the one hand, for each time there is a fragment where it is the present, and so in each fragment the moment that is present in that fragment is objectively special. But, on the other hand, every moment is the present one in some fragment, so all moments are equal after all: they are all present in some fragment or other. Similarly for subjects on a fragmentalist realism about perspectives: my perspective is the objectively special one in the fragment centered around me, but for each subject there is a fragment where their perspective is the special one. So, my perspective is indeed objectively special, but so is every other perspective. And since I am objectively special in the fragment centered around my perspective, the symmetry is broken in that fragment. Identity and difference can be grounded in indexical or perspectival properties, as Koch requires. If a fragmentalist realism about perspective would work, this seems to be just the position we need to support the subjectivity thesis and its associated form of idealism. Furthermore, perspectivalism all by itself would be a form of idealism, at least assuming that perspectives are tied to subjects. If such a perspectivalism were true, then reality would fragment around perspectives and thus around subjects. This would make subjects metaphysically special, since they, just like time, are a source of the fragmentation of reality. But I think fragmentalism can’t be right, and not just because it holds that reality as a whole, über-reality, is incoherent. A first problem with fragmentalist realism about perspective is that it is not easy to see how it can have it both ways: on the one hand that my own perspective is special and on the other hand that other subjects are nonetheless equal to me. The intended way is that my perspective is special in the fragment centered around me, but every other subject is equal in that they also are the center of some other fragment. But when we revisit the proposed way in which we evaluate utterances according to fragmentalism, as discussed above when we considered the problem whether fragmentalism leads to inconsistency, then such an utterance would have

48 idealism and the harmony of thought and reality to be false. An utterance is evaluated not in über-reality, but in a fragment. But in which fragment is an utterance evaluated? Fine’s proposed answer is the fragment where the time of the utterance is present, and analogously, the fragment where the person making the utterance is the center, i.e. has the objectively distinguished perspective in that fragment. So understood I would speak truly when I say that my perspective is special, since in the fragment in which this utterance would be evaluated—the fragment centered around me—it is true. But I would speak falsely when I say that other subjects are equal to me, since they, too, are centers of some fragments. That utterance would also have to be evaluated in my fragment, and in my fragment others are not equal to me: I am truly special and distinguished. This is a real tension. In a sense the problem arises in a familiar way that also affects many other philosophical theories: Fragmentalism can be accused of being a theory that includes a proposal about when assertions are true or false, but when that theory is applied to some of the claims made in the formulation of fragmentalism, then those claims turn out to be false. Maybe such worries can be overcome, after all, we have only seen the most elementary proposal about how to evaluate utterances in a fragment. But on the face of it this objection is legitimate and pressing. But the real problem for fragmentalism is the second one: how to explain the coordination across fragments. The issue is simply this: über-reality is an incoherent mess, that includes for every year the facts that this year is past, and that it is present, and that it is future. In the case of perspectives, it contains for many objects both the facts that the object has hereness, and also that it has thereness. But then fragments form in this lovely way: in each fragment, only one year is present, and all earlier years are past, and later years are future. And in each fragment, only things close to a particular person have hereness, and those far away have thereness. Everything falls into place as we would hope. But why is that so? Why do only these collections of facts form fragments, and not also other ones? Officially, the answer is because only those are collections of facts which are coherent with each other. But that notion of a coherence is like taken as a primitive notion, (Fine, 2005, 281), and so this is not much of an explanation. But something needs to be explained here. To illustrate, consider the table tennis example again we discussed in connection with phenomenalism above, on page 24. As the ball moves away from me towards you, it loses the property of hereness in my fragment, but gains it in your fragment. It might seem that this is easy to explain, since the ball gets closer to you and further away from me. But that doesn’t explain it, since the property of hereness is a primitive further property, not reducible to other, non-perspectival properties like being close to X. These properties are distributed over and above the other properties, but they do not appear randomly in fragments that actually form. Thus we need to explain why there is this order. Über-reality itself can’t help explain this, since it contains all perspectival facts equally. There is also no reductive explanation, since these perspectival properties

forms of idealism we should reject 49 do not reduce to others. And furthermore, these facts are easily explained on other view, like reductive ones, that take perspective less seriously, and either give them only epistemic significance, or understand perspectival properties like hereness as being reducible to something like being close to X. But without some other explanation, fragmentalism is on thin ice.3⁰ Obviously, much more can and should be said about all these issues. but those, in outline, are the reasons for why I do not accept fragmentalism. Without fragmentalism, I see little hope for perspectivalism, and consequently little hope for Koch’s defense of the subjectivity thesis. Giving metaphysical significance to perspectives is an intriguing option for a motivation of a version of idealism, but I don’t think what has been proposed so far along those lines succeeds. There seem to be fairly in-principle obstacles along that path, ones too big for me to see how they could be overcome. So far then it does not look too promising for idealism. But there are still further options. I would like to now move away from trying to tie idealism to the existence of objects or of matter, and instead concentrate on versions that focus on the having of properties and the obtaining of facts. These will get us closer to the view defended in the next chapter, and they are more promising in general.

2.5 Idealist theories of properties and truth So far our focus has been on ontological idealism, that is, on objects and matter, and what significance minds have for them. Are objects themselves mental, or at least derivative on the mental, as phenomenalism would have it? Or is merely which composite objects there are tied to minds, as conventionalism about composition holds? Or does the very existence of material objects require the existence of minds, as the subjectivity thesis maintains? I have argued that we should reject all of these positions. Of course, my discussion was brief and such views could be defended in other ways as well, but the problems we have seen don’t seem to be 3⁰ One especially good, but a little more complicated, example of the problem raised here arises with the fragmentalist interpretation of the special theory of relativity (STR) proposed by Fine in (Fine, 2005) and critically discussed in (Hofweber and Lange, 2017). Fine’s proposal is that instead of one 4-dimensional spacetime we have several fragments corresponding to a 3-dimensional space and time, each of which coordinated with each other via the Lorentz transformations. The problem for this fragmentalist interpretation of STR pushed in (Hofweber and Lange, 2017) is to explain why fragments are coordinated with each other in such a way that the Lorentz transformations hold between them. What explains this coordination across fragments? The non-fragmentalist interpretation of STR has a simple answer: one 4-dimensional reality presents itself differently from different perspectives, separating spacetime in related ways into space and time. No wonder there is coordination, since one reality appears in different ways from different perspectives. But the fragmentalist can’t give this explanation and has no resources for a different explanation of why the Lorentz transformations hold, or so (Hofweber and Lange, 2017). See also (Lipman, 2020) for a defense of the fragmentalist interpretation of STR against the argument in (Hofweber and Lange, 2017), and (Hofweber and Lange, 2020) for a reply.

50 idealism and the harmony of thought and reality tied too closely to the particulars of our discussion, and so it is natural to think that it is not just these particular formulations of those views which are mistaken. If so, then neither that there are objects at all, nor which objects there are, is in general connected to our human minds. Strong ontological idealism thus does not seem to be a promising route towards strong idealism. If human minds are central to reality, it looks unlikely that it is via their special relationship to objects. To reiterate: there are other options to defend a version of ontological idealism, and I certainly won’t be able to rule them all out. Still, it looks like the wind is blowing in a different direction, and that is away from ontological idealism and towards ontological realism. By ‘ontological realism’ I simply mean the opposite of ontological idealism, that is, the view that minds are not metaphysically central to reality understood as the totality of things. But even if so, that doesn’t answer the question about idealism, since it so far only concerns ontological idealism, not alethic idealism. Alethic idealism holds that minds are metaphysically central for reality understood as the totality of facts, but not necessarily also the totality of things. In this section we will consider some ways in which one might defend alethic idealism. I will argue that these forms of alethic idealism are also mistaken. In the rest of the book I will defend a different version of alethic idealism instead, in particular one that is compatible with ontological realism.

Mind-dependent properties A natural first step towards alethic idealism is to put all the idealism not into objects, but instead into the properties of objects. That is, to hold that even though ontological realism is true—again, in the innocent sense that just denies ontological idealism and thus denies the metaphysical centrality of minds to reality understood as the totality of things—nonetheless, our minds are central to how these things are. Our minds then are central for what properties things have, even if they aren’t central to which things exist. There are several ways this could go, and at first none of them look too promising. The most radical one is to hold that our minds are somehow responsible for the division between object and property itself: It is because of us that there is a division between objects and properties in the first place. On such a view our minds somehow impose the object-property structure onto the world, and that is why the world contains objects having properties. But this line is hard to combine with ontological realism: if we are not central for there being objects, and thus objects exist without our doing, then how can it be that we are responsible for the division between objects and properties? And if there are objects independently of us, what else would they be like other than having properties? Thus it is hard to see how we can be the source of the object-property division itself without also holding onto ontological idealism, something that has so far not looked too promising.

forms of idealism we should reject 51 The next step down would be to hold that although there are objects that have properties independently of us, nonetheless we are responsible for which properties these objects have: It is because of us, somehow, that objects have these rather than those properties. But this proposal faces similar problems as conventionalism, in particular the explicitness-compatibility tension between spelling out the relevant sense of dependence and reconciling the dependence of properties in that sense with what we know to be the case. One can’t easily hold that we are causally responsible for objects having the properties they have, for a series of straightforward reasons: objects had properties before we existed, we are objects as well, and our causal influence does not extend to all parts of the world where objects with properties are found. Whatever sense of dependence of properties on us there is supposed to be, it can’t be causal dependence. Similar remarks apply to counterfactual dependence. Clearly, some objects having some properties causally or counterfactually depends on us, like my banana having the property of being peeled. I will simply leave these obvious cases aside, and only focus on the more large-scale issue of a whole range of properties depending on us in some sense. The next best option seems to be to take inspiration from a primary-secondary quality distinction and hold that although there are certain properties that objects have independently of us, there are also other properties that depend on us, and do so non-causally. These dependent properties are “projected” onto the world, or “spread” from the mind outwards towards the world.31 And if there are enough of these properties and thus a good part of reality is tied to these projected properties, then this might well justify accepting that we are central to reality. The hard part in this proposal is to make sense of projection, which so far is merely a metaphor. In particular, projection can’t be understood as involving an error, as when we project something that isn’t there, and thus we falsely think that the world is a way that it in fact isn’t. Rather projection has to be understood as veridical for this proposal to make sense: the objects have these properties, but somehow it is because of us that they have them. There should be some non-causal sense of dependence in which these properties depend on our minds. There are two main ways one could try to spell out the metaphor of projection, but neither one of them is good enough to defend a version of idealism in our sense. The first takes its inspiration from Simon Blackburn’s quasi-realism,32 the second takes projected properties to be response-dependent properties. Both of these views naturally deserve a more substantial discussion, something we can’t engage in here, but I would nonetheless like to briefly present my reasons for why I think neither one is well-suited for a defense of a version of idealism.

31 Famously, David Hume asserts this metaphorically: “. . . the mind has a great propensity to spread itself on external objects . . . ” (Hume, 1739, Book I, §3.14). 32 See (Blackburn, 1993) and elsewhere.

52 idealism and the harmony of thought and reality The main idea of quasi-realism is to combine an expressionist theory of some parts of discourse with a minimalist theory of truth and fact. The former holds that some part of discourse, often evaluative discourse like an assertion that stealing is wrong, is not descriptive of the world, but expressive of an attitude the speaker has. The latter holds that to say that it is a fact or a truth that stealing is wrong essentially comes to the same thing as saying that stealing is wrong.33 This view is intended to be on the one hand anti-realist, after all, murder being wrong is tied to our attitudes, but also very much like realism, since it holds that there are facts about murder being wrong. And this combination of views might then be used to defend a form of idealism, not in the narrow sense of classical idealism, where matter itself is mental, but in our broader sense that minds are central to reality. If reality is understood as the totality of facts, and it being a fact that stealing is wrong comes down to stealing being wrong, and if stealing being wrong is an expression of our attitudes, then parts of reality are tied to our attitudes and thus our minds. If those parts are big enough, then maybe reality is in good parts tied to our minds, and thus idealism is true, or so the idea. Quasi-realism famously walks a thin line between on the one hand conceiving of itself as anti-realism, but on the other hand also as realism, adding up to quasirealism.3⁴ If realism is simply concerned with the obtaining of facts, then it is a form of realism, since it is a fact that stealing is wrong. If realism is concerned with what corresponds to descriptive discourse, then it is anti-realist about evaluations of stealing. It is consequently important to distinguish two widely used senses of ‘realism’. Both are relevant for our discussion here, and especially also for what is to come a bit later. One is simply the anti-idealist sense of ‘realism’. In this sense it is a denial of idealism, and thus realism in its anti-idealist sense is simply the position that minds are not metaphysically central to reality. Besides this anti-idealist sense of ‘realism’, there is also another one, which we can neutrally, but possibly unhelpfully, call the anti-antirealism sense. Here realism goes not primarily against idealism, but against antirealism. Antirealism in this sense is paradigmatically exemplified by expressivism, which holds that the function of a particular part of discourse is not descriptive, but expressive. The expressivist account of moral discourse as expressing attitudes is a good example of that. But these two sense of ‘realism’ need to be carefully distinguished. Someone who holds onto minimal theistic idealism, and thus that matter is an aspect of the mind of God, is not thereby an antirealist about matter discourse. They likely will take such discourse to be fully descriptive, but they make a claim about the nature of the things this discourse is about. And similarly, someone who holds onto an antirealist position about moral discourse, and takes it to be expressive of our attitudes, but not descriptive at all, is not thereby an idealist. They make no claim 33 For the details, see (Blackburn, 1984) and (Blackburn, 1993). 3⁴ See (Dreier, 2004) and (Sampson, 2018).

forms of idealism we should reject 53 about the metaphysical place of minds in the world, only a claim about what the function of a certain part of our speech is. Although in spirit antirealism and idealism are very different views, in letter we need to be more careful, in particular if we adopt a minimalist conception of facts. To do this, we need to articulate a distinction about how to understand ‘reality’ on the quasi-realist picture. This picture includes a minimalist theory of facts and fact-talk, and thus it endorses the minimal biconditional: (4) It is a fact that p iff p. Here ‘p’ can be any statement, be it expressive or descriptive. (4) is certainly plausible for descriptive ‘p’, but it can be affirmed for any ‘p’, even assuming that some sentences have an expressive, but not descriptive, function. We can then think of reality as the totality of facts in two ways: more narrowly or more broadly. To mark that difference, we can call a fact an expressive fact if it corresponds to an expressive sentence in the minimalist schema (4), and we can all a fact a descriptive fact if it corresponds to a descriptive sentence in that schema. Reality can consequently be understood as either containing only all the descriptive facts or else all the fact, descriptive and expressive.3⁵ On the former, reality does not settle whether stealing is wrong, favoring antirealism, while on the later it does settle it, favoring realism. The question remains which notion of reality is the relevant one for the question whether quasi-realism supports idealism. In our discussion so far, we took reality to be the totality of all facts, but that might have implicitly assumed that all facts correspond to descriptive discourse, and thus are descriptive facts in present terminology. If we widen that conception of a fact to include expressive facts via minimalism, then we might or might not want to widen our conception of reality with it. And with it we might or might not widen our conception of what counts as an idealist view. Part of this issue is clearly merely terminological, but part of it concerns the true spirit of the idealist position, and of our concept of reality. Here it seems clear to me that the intended notion of reality includes only facts in the sense of corresponding to descriptive discourse. If declarative sentences have not just a descriptive but also a further expressive function, then this is better understood as a view that some parts of our speech is not concerned with depicting reality. Such speech is 3⁵ Although this characterization might suffice for now, it is not quite as good as it can be. So far the issue is left open about what to do with facts that correspond to no instances of ‘p’ in our language, and which is thus an ineffable fact, a topic that will be important later on. If we can assume that at least the expressive facts are effable, then we can characterize the descriptive facts as all those facts which are not expressive, and the expressive facts as all those that correspond to an expressive sentence in the schema. Furthermore, the distinction is more properly made not for sentences, but for assertions of sentence in a context, but my main point remains the same either way. That expressive and descriptive facts are different, i.e. the same fact never corresponds to both an expressive sentence and also a descriptive sentence, is accepted on standard expressivist views and assumed here as well.

54 idealism and the harmony of thought and reality thus not about reality, but rather does something else: expressing attitudes. And that means that the proper conception of reality we should be working with is the narrower one, the one that corresponds to the descriptive facts and is the target of our descriptive speech. Idealism thus concerns the place of minds in reality as the totality facts, narrowly understood. And made precise this way, it is clear that expressivism is not a form of idealism, even if we accept minimalism.3⁶ It is consequently no surprise that Blackburn himself does not identify as an idealist. Quasi-realism isn’t idealism, but a form of antirealism. As such it is domain specific, applying only to domains of discourse that are expressive, not descriptive. What place minds have in reality is left open by this, although it does support that not all of our speech is aimed at reality. Trying to spell out the metaphor of projection in terms of response-dependent properties fares no better as a defense of idealism. The basic idea of this proposal is inspired by thinking about paradigmatic potential examples of secondary qualities, in particular colors. On such a view being red is a property that needs to be specified in part by talking about us human beings. The property of being red is not something like the property of reflecting light of a particular wavelength. Instead it is something like the property of appearing reddish to normal humans in normal circumstances. Here “reddish” is a phenomenal appearance, something mental, and since appearing reddish is a response by minds, the property overall is a response-dependent property. The proposal then is that a large range of properties of ordinary objects are such response dependent properties. At least the manifest image, how the world seems to be ordinarily, is full of such response dependent properties. These properties are real properties of things, the proposal continues, and so reality is in part tied to us and our minds. We are central to reality via the properties of things, and thus idealism is true, or so the idea in a nutshell. But this is not enough. Simply because a certain property can be expressed with reference to us does not make us central, not even for having that property. The property of being red, on the above proposal, might well also be the property appearing bluish to certain unusual dogs in certain unusual circumstances, and if so then this does not mean that dogs are central to color. Similarly, it is not enough to argue for the metaphysical centrality of measurement devices simply because certain properties like the property of being charged can be expressed by referring to measurement devices in some way or other, e.g. would result in such and such a reading on a measuring device in certain circumstances. Obviously, whether or not being red can be expressed as a response-dependent property is controversial, and so is whether in the end it can be expressed as a dog-response-dependent property or a measuring device-respondent-property. But the point remains the same: even if it could be expressed this way, no conclusion about the metaphysical centrality 3⁶ See also (Fine, 2001), where Kit Fine proposes that we must accept a distinctly metaphysical notion of reality, one that is likely a primitive metaphysical concept, to formulate the question of realism.

forms of idealism we should reject 55 of those mentioned in that expression follows, since it might well be expressed in many such ways, and it might well be an artifact of the expression of the property rather than the property itself. One might try to do better by claiming that it is a distinct part of the essence of certain properties that they are tied to minds in a certain way, thereby focusing on the property itself, and not merely its expression. But this is a long shot, since it will be tough to defend that properties have such complex essences, for example, that being red by its essence is tied not just to humans, but normal humans in normal circumstances. I therefore think that response-dependent properties are a dead-end for an attempt to defend a version of idealism. They might well have other good philosophical uses, but they won’t help us to motivate idealism.3⁷

Mind-dependent truths and facts Given our discussion so far, it is hard to see how we could be central for what objects there are or for what properties these objects have. But not all idealist hope is lost. A more promising route toward alethic idealism, if there is such a route at all, is to focus on truth and facts directly, and not on objects and their properties individually. Maybe we are central to reality understood as the totality of facts, since facts themselves are somehow tied to us and our minds. Facts are closely connected to truths, and thus essentially the same issue will arise for whether truths are somehow connected to us. To make this connection more explicit, we talk about truths and facts directly with that-clauses, like ‘that snow is white’. There is a plausible equivalence between facts and truths, which is an extension of the minimalist biconditional (4) above: (5) It is a fact that p iff it is a truth that p iff p. In particular, for any fact that p there is a corresponding truth that p, and the other way round. That leaves open whether facts are just identical to truths, whether facts are identical to true propositions, or whether they are all different, but nonetheless correspond to each other. We don’t have to take sides on this issue now, but we can note that among all propositions, the true ones correspond to facts, while the false ones do not. Truth is also attributed to sentences or assertions, as when I say that ‘Snow is white’ is true. The sentence or assertion does not directly correspond to a fact, it is a linguistic thing after all, but it is associated with a proposition: the proposition which is the content of an assertion of this sentence. And if the assertion is true, then the proposition which is the content of the assertion is true 3⁷ See (Rosen, 1994) for a more detailed discussion of response-dependent properties and their connection to idealism. Rosen reaches essentially the same conclusion.

56 idealism and the harmony of thought and reality as well, and thus corresponds to a fact. So far the basics that anyone can accept in outline. If our minds are supposed to be responsible somehow for truth and facts, the obvious first ways to think of this is that either we are responsible for there being facts at all, or else we are responsible for which facts obtain, and thus for what is true. The former is hard to make sense of: what would it mean for there to be no facts at all without our involvement? There being no facts at all seems to be incoherent: if there are no facts at all, then it is true that there are no facts at all, and thus it would be a fact that there are no facts at all. But maybe this incoherence only appears “from within” and does so only to us. After all, once we think about what is and isn’t coherent, then we are already at a stage after our contribution to there being facts at all has happened. But even if so, this does not help us make sense of that there being facts at all is due to us. It is no help, of course, to insist that there being facts is a transcendental condition on the possibility of thought or the like; this at best just puts a label on what needs to be made sense of. And it is us human beings who need to make sense of it. If there being no facts at all is simply incoherent by our own best standards of coherence, then it seems to be impossible that we can make sense of that there being facts at all is due to us. Our doing so would not rule out a coherent possibility. What is supposed to be due to us is something that we must already accept as being guaranteed to be the case, so we can’t coherently take a lot of credit for it. This way of defending alethic idealism would simply go too far.3⁸ But maybe we can make sense of that we are centrally involved in which facts obtain, even if we can’t make sense of there being facts at all is somehow due to us. One way to think of this possibility is as follows: there is some domain of propositions, some of which are true, some of which are false. The true propositions correspond to the facts: the proposition that p is true just in case the fact that p obtains. If we were somehow involved in which propositions are true, then we would be involved in which facts obtain. Clearly what is at issue here is again a sense of non-causal dependence of the truth of propositions on us and our minds.

3⁸ One very recent attempt to defend a view of the dependence of there being objects and facts at all on us and our language is in (Gaskin, 2020). There Richard Gaskin hopes to defend what he calls “linguistic idealism,” which is “the doctrine that the world is essentially a product or precipitate of language” (Gaskin, 2020, viii). But how can that be, given that the world was there before us? Gaskin maintains that “Human beings and their language depend empirically for their existence on the existence of the world. But the world depends transcendentally for its existence on the existence of language, a mere human invention” (Gaskin, 2020, 249). However, neither the notion of “depending transcendentally” nor that of being a “product or precipitate” of language are spelled out further in more explicit terms. They are intended non-causally, but that is not much of a positive account. Without a more explicit characterization of these notions such a view would not meet our explicitness constraint, and it is hard to see how whatever arguments are given for this view would support it if it isn’t characterized more explicitly. And similarly, without spelling out what transcendental dependence is supposed to be, it is hard to see how the world can transcendentally depend on us, even though it was there first. This won’t solve the dependence problem.

forms of idealism we should reject 57 Uncontroversially, the truth of some propositions causally depends on us. But can we make sense of the truth of propositions more generally to depend on us in a non-causal way? At first, it seems unclear how this could be, but there are some options worthy of consideration. The key to seeing this is to consider the distinction between construction and constraint. If which propositions are true is due to us, then we in effect construct the world: we make it the case, somehow, which among the eligible candidates for truth, the propositions, are a truth. And with it we make it the case which facts obtain and thus what reality is like. Or to put it differently: which propositions are true depends on us. Alethic idealism via construction seems like a dead-end, for much of the same reasons that undermine other forms of idealism already discussed: not only is it unclear in what sense it might depend on us which propositions are true, it also seems to be incompatible with what we know about our place in the world. Construction seems to be out, but this leaves open that we constrain the world. To constrain the world is to set some limits on what the world can be like. For us to constrain the world is for these limits being properly due to us: Either because of something we do, or some feature of our minds, or for some other reason tied to us. It is natural to think that constraint is a consequence of construction. If we are responsible for which propositions are true, then this might well lead to a constraint on the world. Certain propositions might be ruled out from being true, since they fall outside of what can be constructed by us. Constraint will likely be a result of construction, but we have reason to think that construction is mistaken. The question remains whether we might constrain the world without constructing it. Can we have constraint without construction? Although it is not clear how it could be that the world is constrained by us without also being constructed by us, this seems to be the most promising option left on the table. In fact, I will be arguing in favor of a version of this option in the rest of this book. But before we can get to the particular version I defend here, we should look at one alternative view, which I will reject, but which can also be seen as a version of constraint without construction: an epistemic theory of truth. One way the world might be constrained by us, without being constructed, is that truth itself is constrained by us: there are some conditions and limits on what can be true that are tied to us, somehow. The most popular way to defend this is via an epistemic theory of truth: truth itself is tied to something like verifiability or knowability. One popular version of such a view is to maintain that truth requires in-principle knowability, by us: What is true must be knowable by us in principle. On such a view it would be a mistake to think that knowledge can be analyzed in terms of truth, as something like true justified belief. Rather it goes the other way round: truth is analyzed in terms of knowledge. Any such view could be seen as putting a constraint on reality that comes from us. For it to be a fact that p it must be true that p, and thus, according to the epistemic theory of truth, it must

58 idealism and the harmony of thought and reality be knowable by us that p. Thus reality as the totality of facts is constrained by our minds and what we can know, or so the idea.3⁹ But this is a bit too quick. Simply because any truth must be knowable does not support that our minds are metaphysically central in the world in some intimate way. The question remains why it is that any truth must be knowable. And here this could be for reasons that have nothing to do with idealism. It could just be luck: the world is simple enough and we are good enough knowers that in principle we can know any truth. Such luck would be no reason to accept idealism. It would not support that we are metaphysically central to reality, only that we happen to be well-placed in it. On the other hand, any truth could be knowable not by luck, but for a reason. However, that reason still might not support idealism. For example, it could be that God finds it unbearable to have creatures like us in the world that are bound to ignorance, and thus God created the world and us in such a way that any truth is knowable by us. That is some coordination between the world and what we can know, one that supports that every truth is knowable, but it does not support idealism, in particular not strong idealism. Thus not just any reason for a correlation between what is true and what is knowable would support idealism, only the right kind of reason. Here the intended reason why every truth is knowable is supposed to come from the theory of truth itself: somehow truth itself is such that it must be knowable by us. It is part of the nature of truth, to put it that way, that it be knowable by us. Of course, this can’t just be proclaimed, it must be supported by an argument. In particular, it seems that we can give several examples of pairs of propositions where we have reason to think that one of the other of that pair must be true, but we will never know which one. For example, we know that dinosaurs once roamed the earth, and thus some number of dinosaurs lived on planet earth. But is that number odd or even? No matter how we make precise what a dinosaur is and what is required for having lived, we will never know the answer. A defender of such an epistemic theory of truth will likely accept the case, but argue that it doesn’t refute their proposal, since with an epistemic theory of truth comes a rejection of classical logic. So, they will accept that it is true that some number of dinosaurs roamed the earth, but for each number n, it is not true that n dinosaurs roamed the earth. And they will accept that although it is true that the number of dinosaurs is either odd or even, it is not true that the number is odd, and it is not true that the number is even.⁴⁰ To reject classical logic to deal with apparent problems is one thing, but what positive reason do we have to accept such an epistemic theory of truth, and with it a likely rejection of classical logic? One proposed reason comes from the 3⁹ For a detailed discussion of various versions of, arguments for, and problems for such an epistemic theory of truth, see chapter 7 of (Künne, 2003), which also contains main further references to the literature on this topic. ⁴⁰ Michael Dummett is the locus classicus for this proposal. See, for example, the essays in his (Dummett, 1978c).

forms of idealism we should reject 59 philosophy of language and the theory of meaning.⁴1 Maybe meaning has to be such that it can’t lead to verification-transcendent or knowledge-transcendent truth conditions. After all, how might one be able to learn words with such meanings, and how could these meanings be fixed by our language use? There is certainly an argument worth discussing here, but however that turns out, it won’t lead to idealism in our sense. At best such an argument will lead to a constraint on truth as it applies to our assertions and our sentences. But idealism is not the view that our minds are central to our assertions or even to all truths that we can assert. Instead idealism holds that our minds are central to reality, in particular reality as the totality of facts. No constraint about the totality of facts would follow from a constraint about what we can truly assert or what meanings sentences in human languages can have. The latter is about us and what we can represent in language, the former is about the world and what it is like. Idealism concerns reality overall, not just us and our languages. Uncontroversially, our minds constrain what we can truly assert, but this does not imply that they constrain the world. To put it differently, an argument from the learnability of meanings can at most constrain truth as it applies to assertions, not truth as it applies to propositions. Since facts correspond to true propositions, but not necessarily to true assertions, such an argument would not lead to a constraint on the totality of facts. At least not unless the bridge between what propositions are true and what we can truly assert is crossed someone, and it is somehow shown that there is a close connection between our own language and the totality of facts. But without such an argument, an epistemic theory of truth motivated via considerations about meaning won’t lead to a constraint on reality as the totality of facts. Not that those who hold onto such a theory of meaning, in particular Michael Dummett, make such idealist claims. Their target is generally to defend a form of antirealism about some domain or other: claims about the future, or claims in mathematics. ‘Antirealism’ here is supposed to be in the ballpark of something being merely projected, in the sense discussed above, or of something not being there independently of us, but only as some form of consequence of our activity. How to spell out these metaphors that characterize antirealism is notoriously hard, as is the question whether there even is a real issue here. For Dummett, the question of realism is closely tied to the obtaining of bivalence: either a statement or its negation is true. If the meaning of our assertions cannot have verification transcendent truth conditions, then it is arguable that bivalence fails, since there will be cases where we can neither verify or know a statement nor its negation. It makes sense to connect antirealism to our assertions; it concerns our activity after all. But this isn’t a route towards idealism.⁴2

⁴1 See (Dummett, 1978b). ⁴2 There is also a historically prominent connection between idealism and a different theory of truth: a coherence theory of truth. I do find a coherence theory harder to motivate or make sense of than an

60 idealism and the harmony of thought and reality Overall then, an epistemic theory of truth for assertions is not a promising route towards idealism. Even if it were correct, it would not support that minds constrain the totality of facts, but only the totality of assertable facts, something we should believe anyways. But if we could show that every proposition is assertable by us, then one could extend an epistemic theory of truth to the truth of propositions, and with that we might constrain the totality of facts. If every proposition is assertible, then any constraint on what can be truly asserted is a constraint on which propositions can be true, and thus on which facts can obtain. That would indeed motivate idealism. But on this route towards idealism, an epistemic theory of truth plays only a minor role, if any. The real idealist core of such a proposal would be that every proposition is assertable by us human beings. This certainly should not be expected to be the case. Why would it be that we human beings have all the representational resources to assert every proposition or to represent every fact? If such a connection between what propositions there are and what we can assert can be made, then this in itself might be closely tied to an idealist position. The real work will go into showing that these connections between what we can represent, on the one hand, and what is true and which facts obtain, on the other hand, can be established and what the reason is for why this connection obtains. It would lead to an approach that is motivated via considerations about what can be represented by us, not via considerations about what can be known by us. And this points to what seems to me to be the correct way to motivate idealism. And with that it is finally time for me to present my positive proposal.

epistemic theory, so I focused on the latter. For more on a coherence theory, see (Young, 2018), which illustrates just how hard it is to make sense of such a theory. Another association between a theory of truth and idealism that is sometimes made is that with an identity theory of truth. On such a theory the content of true thoughts is the same thing as a fact, which might be taken to suggest that the world as the totality of facts is somehow mental. But this would be a mistake, as it confuses the thoughts as mental events with the contents of the thoughts, which can simply be seen as propositions, and thus as worldly things. They relate to thoughts as being their contents, but are not thereby mental. Facts could be identical with true propositions, but that is also not a form of idealism, only a form of the view that the world is thinkable by some thinkers who can have thoughts with these propositions as contents. Such an identity theory of truth is endorsed by, for example, (Trueman, 2021) and (McDowell, 1994), with both correctly claiming that it is not a form of idealism.

3 Conceptual idealism: the basic idea Let us briefly take stock of where we are. Our main starting point was the Big Question about the place and significance of human beings in the world overall. Besides the standard naturalistic answer that we are not special and the theistic answer that we are special, there was also the possibility of a third, idealist answer. This answer holds that we are special, not because of a relationship to God, but more directly, since there is an intimate connection between our minds and reality itself. Idealism should be understood broadly as the claim that minds are metaphysically central to reality, not just narrowly as that the mental grounds the non-mental or that matter itself is mental. And it can be understood strongly, as that our human minds are metaphysically central to reality, not just weakly that some minds or other are central. A strong version of broad idealism would give us a third answer to the Big Question about our place in reality. If it were true, then it would support the claim that we indeed are special, not via a connection to God, but directly, via our connection to reality itself. But as we saw in the last chapter, it is not easy to defend such an idealist answer. Since reality can naturally be understood either as the totality of things or the totality of facts, one could aim to defend (strong) ontological idealism, the view that our minds are metaphysically central to reality as the totality of things, or (strong) alethic idealism, the view that our minds are metaphysically central to reality as the totality of facts. Neither one of these options looks promising at first. It is hard to see how we could be central for the totality of things and for what exists in general. That objects in general are somehow derivative on our minds, as a phenomenalist would have it, is hard to reconcile with what we know about the world. That which objects there are depends on our minds, as a conventionalist would have it, is difficult to defend and make sense of, in particular how the notion of ‘dependence’ is supposed to be understood. And that the existence of objects at all requires the existence of minds, as the subjectivity thesis holds, is equally problematic, since the argument given for it either leads to incoherence or doesn’t solve the problem that motivated the view in the first place, as argued in the last chapter. Ontological idealism might be defended in other ways as well, but on the face of it it does not look very promising. Alethic idealism does not seem to fare much better. It is hard to make sense of how we could be central to the totality of facts. To hold that we are somehow responsible for the properties that independently existing objects have was less than promising in our discussion in the last chapter, and so was making sense of Idealism and the Harmony of Thought and Reality. Thomas Hofweber, Oxford University Press. © Thomas Hofweber 2023. DOI: 10.1093/oso/9780198823636.003.0003

62 idealism and the harmony of thought and reality how we might be responsible for which facts obtain in general. Somewhat more promising was the idea that although we do not construct the world, we constrain it. This option of constraint without construction would seem to be supported by views like an epistemic theory of truth, where every truth is required to be knowable by us. But as discussed in the last chapter, the motivation for such an epistemic theory of truth can at best apply to our assertions and thoughts, but not to all propositions or to all facts. It is thus suitable to support antirealism in a widely used sense, but not to support idealism in our sense here. Still, the general option of constraint without construction is promising, and in this chapter I hope to argue that this is indeed the correct option and that a version of idealism is true that falls under this general heading. To defend such a view we need to make sure that it meets all the constraints that any reasonable form of idealism has to meet, as spelled out on page 8: it needs to be explicitly formulated without mysterious primitive notions of dependence, it needs to be supported with a coherent argument, and the resulting position needs to be compatible with what we know to be true. In this chapter I would like to put all my cards on the table. I will formulate a version of idealism, present the main argument for it, discuss some problems for this view, outline how they can be overcome, and point to what overall position this leads to. In the chapters to follow I will then develop various aspects of this position and argument in more detail: I will strengthen the version of idealism outlined in this chapter in a later chapter, I will spell out the positive argument in more detail, and I will develop the general approach to metaphysical questions that is congenial with the main argument for idealism. I prefer to proceed this way so that the overall position can come into view as soon as possible, rather than to develop each part of it separately and only put things together towards the end. This chapter is thus the most central one in this book.1

3.1 Idealism via harmony Although ontological idealism is unpromising, alethic idealism might do better. Alethic idealism concerns reality as the totality of facts, not of things. And although it is so far unclear why alethic idealism might do better than ontological idealism, one key difference might be the relationship between things and facts, on the one hand, and our representations of them, on the other hand. Such representations of facts can either be in thought or in language. In the case of linguistic representation, a thing is paradigmatically represented by a name, whereas a fact is represented by a sentence. Names are generally simple representations, 1 This chapter overlaps with (Hofweber, 2019b). However, the overlapping material has been revised and updated. (Hofweber, 2017b) contains an earlier presentation of the view.

conceptual idealism: the basic idea 63 they have no relevant internal complexity to them, whereas sentences exhibit such complexity, as do complete thoughts or judgments, in the case of mental representations of facts. And here there is a notable connection between facts and our representations of those facts. Reflecting on this connection is a key first step in the right direction. To start we can note that facts are often similar to each other in particular ways. For example, the fact that Sue is tall is similar in one way to the fact that Joe is hungry. Both facts are facts of an object having a property. We can say that these facts share a structure: they have an object-property structure. Talk of structure is supposed to be taken in an innocent sense here, if at all possible. That these facts have an object-property structure is, in the relevant sense of the phrase, a consequence of their being facts of an object having a property. To say that facts have a structure in our sense does not endorse a particular metaphysics of facts, nor does it use a substantial metaphysical notion of structure. It only puts a label on the obvious: some facts are facts of objects having properties, other facts are different kinds of facts.2 The structure of facts seems to have a connection to our thoughts and sentences. We represent facts in conceptual thought or in language, and these representations in turn have a certain complexity as well. To take the case of linguistic representation, the fact that Sue is tall is represented by the sentence ‘Sue is tall’. This sentence in turn has a particular form: it is a sentence with a subject term and a predicate. We can say that such sentences have a subject-predicate form. And although the complexity of mental representations is less apparent, it is reasonable to hold that however we represent the world in our minds, there will be an aspect of a representation of the fact that Sue is tall that corresponds to a subject term, standing for Sue, and a predicate, standing for being tall. Unsurprisingly, there is a connection between the thought and the fact it is about. The form of the thought seems to match up perfectly with the structure of the fact. A subject-predicate thought represents a fact with an object-property structure. In this simple case there seems to be a perfect match between the form of our thought and the structure of the fact that it represents. But why is there this match? Does this correspondence of form and structure need, and allow for, an explanation? There are two straightforward ways in which this correspondence could be explained, which are based on two different directions of what is explanatorily more basic: the form of our thoughts or the structure of the facts. The antiidealist realist will hold that our thoughts have their form because the facts have 2 Whether each fact has a unique structure is controversial, with Frege being a likely exception to the more standard view that they do have a unique structure. Frege famously held that contents can be carved up in different ways, and this naturally can be understood as being associated with the view that facts can have more than one structure. See (Frege, 1884). I hope to make clear below that this issue is largely irrelevant for us here.

64 idealism and the harmony of thought and reality the corresponding structure. And an idealist can see it the other way round: the facts have the structure because our thoughts have the form. At first it must seem that the realist got it right. The realist has a perfectly good explanation of why some of our thoughts have a subject-predicate form, which in outline goes as follows: Our minds developed in a world full of facts that have an object-property structure, i.e. a world of objects having properties. It would be quite inefficient for our minds to have a separate representation for each fact, especially since the same object usually has many properties, and the same property is often had by many objects. Thus our representations developed to exploit the structure of the facts and their components. We developed separate representations for the object and the property—a subject term and a predicate—which get combined somehow to represent the whole fact. And thus our minds have representations that have a subject-predicate form, which exactly corresponds to the objectproperty structure of the facts. The realist thus has a perfectly good explanation of why some of our forms of thought correspond to the structure of some of the facts. And the realist can employ the same strategy for any other form of our thoughts that we might find. In other words, the realist can explain why our forms are correct: the forms we have correctly correspond to the structure of the relevant facts. But the question remains whether our forms are complete: whether for every structure that occurs in some fact there is a form of some of our thoughts that corresponds to it. The realist might naturally be inclined to accept at least the possibility of structures among the facts that go beyond the forms of our thoughts. After all the facts with their structure are simply there, with the structure they have, ready to be represented. We should not expect, on the realist’s take, that all the structure to be found in the facts corresponds to some form in our thoughts. Our forms are correct, but maybe not complete, or so it is natural for the realist to hold. However, here the idealist will see things differently. If the facts have the structures they have because of the forms of our thoughts, then it is natural to hold that all the structure there is to be found in the facts corresponds to the forms of our thoughts. An idealist would thus naturally disagree with the realist here, and hold that our forms are, and have to be, complete when it comes to capturing the structures of the facts. This difference between realism and idealism leads to a possibility of formulating and defending a version of idealism. If we had reason to think that the structure of the facts does not, and cannot, go beyond the forms of our thoughts, then this might support idealism. There might be an explanation why our forms of thought are complete when it comes to capturing the structure of the facts, and this explanation might be an idealist one. This is the strategy for formulating and defending a version of idealism that I hope to pursue in this chapter. Let us call a fact that we human beings cannot represent in thought or language an ineffable fact. This notion is so far unclear, since it is unclear how ‘we’ and ‘cannot’ in its definition are to be understood. ‘We’ could be understood narrowly

conceptual idealism: the basic idea 65 or widely, possibly varying across time and across linguistic communities. What human beings can represent might differ over time and in different languages. Similarly, ‘cannot’ can be understood narrowly or widely: that which cannot be represented given how long we in fact live and how much we can say in a lifetime, or what we cannot represent in principle, even with more time. Neither way of making the notion of the ineffable more precise is better than the other by itself. But since our concern here is the place of humanity in reality in general, we should take the notion in its wide sense in each case. The ineffable should be seen as that which cannot be represented by any human being, no matter what language they speak or when or how long they live. The ineffable in this sense is a limitation of humanity. If there are ineffable facts in this sense, then this would point to a real mismatch between our minds and reality. We would be limited in what we can represent about the world, not just because of a limitation of the particular language we happen to speak, or because of our lives being just a little too short, but because our mind is just not suited to represent some parts of reality.3 There could in essence be two main reasons why a fact is ineffable for human beings in principle. They can be illustrated by two reasons why we might be unable to represent a particular fact with a subject-predicate representation. First, it could be that the fact is a fact of an object having a property, but we are somehow unable to either represent the object or the property. Second, it could be that in order to represent the fact we need a different kind of representation, one with a different form than a subject-predicate representation. The fact then is not one of an object having a property, but a different kind of fact. We can consequently distinguish two kinds of ineffable facts. First, a fact F is structurally ineffable if none of the forms of representations we have available are suitable to represent a fact with the structure of fact F. The structure of F would require a form of representation that goes beyond the forms we have access to.⁴ Second, a fact F is content ineffable if it has a structure that matches one of our forms, but somehow we are unable to fill in the relevant parts: we can’t represent an object, or a property, or some other aspect of it. Structurally ineffable facts are truly alien to us, while content ineffable facts are not all that alien, since they are at least facts of the same general kind as facts we can represent. We can now say that our minds and reality are in structural harmony just in case there are no structurally ineffable facts. Our minds and reality are in complete harmony just in case there are no ineffable facts at all: neither structurally ineffable

3 For a more detailed discussion of the notion of the ineffable and different ways to make it more precise, see (Hofweber, 2017a) and (Jonas, 2016). Whether different human languages differ in what they can represent is discussed, for example, in (von Fintel and Matthewson, 2008). We will discuss all this in more detail below in Chapter 5. ⁴ If facts can have more than one structure, then we take structural ineffability in the strongest sense: for none of its structures do we have a matching form. In light of this it should become clear later that it won’t really matter whether facts have a unique structure.

66 idealism and the harmony of thought and reality facts nor content ineffable facts. Harmony by itself, be it structural or complete, does not guarantee idealism. If harmony obtains, then this could be by accident or for a reason. It could be, for example, that all facts are just facts of objects having properties. Maybe the world is simple in this way, and then our minds would be good enough to represent all the facts at least in their structural aspects, and maybe even completely. There would then be no structurally ineffable facts, not because of an intimate connection between our minds and reality, but because we got lucky in that reality is simple and uniform enough so that the forms of our thoughts are good enough to match the structure of all the facts. But we would be lucky if that were the case, and we should thus not expect it. But not all the facts are this simple, since many facts we can represent don’t have simply the structure of an object having a property: there are conjunctive facts, universal facts, and so on. We have more forms of thought than simply subject-predicate representations, and since we have good reason to think that some of those representations represent accurately, we have reason to think that the facts that obtain don’t all have the structure of an object having a property. So why should we think that all the structure that might be realized in facts is structure corresponding to one of our forms of thought? The realist should expect that structurally ineffable facts are at least possible and not ruled out in principle. If there aren’t any, then we got lucky, but there is no guarantee that we should get lucky. The idealist, on the other hand, could turn this around and aim to support idealism via an argument that structural ineffability is ruled out in principle. The reason why there aren’t, and can’t be, any structurally ineffable facts might support idealism, since it might make clear that there is an intimate connection between the form of our thoughts and the structure of the facts. It might be that our minds limit the range of facts that could in principle obtain in that any fact that could obtain is required to have a structure corresponding to a form of our thought. And if so, then we might be central in reality after all, since we play a central role in reality understood as the totality of facts. The facts might have to conform to our form of thought, not by accident, but for a reason that makes clear that our minds are central in reality.⁵ To try to formulate and support idealism via considerations about the harmony of thought and reality is so far only a strategy for a defense. If it were successful it might well support a rather different version of idealism than the versions discussed above. Whether this strategy is at all fruitful will depend on two things. First, whether there is a good argument that harmony has to obtain in the first place. Second, whether this argument can be seen as providing the right kind of

⁵ Thomas Nagel is slightly unusual, but I believe correct, when he in (Nagel, 1986, 93ff.) takes the real issue about idealism and realism to be whether the world might outrun our representational capacities. Nagel, of course, rejects idealism so understood.

conceptual idealism: the basic idea 67 reason for why harmony obtains, namely the kind of reason that would support idealism. In the following I would like to argue that this strategy is indeed successful.

3.2 Harmony via internalism In this section I will present the argument that structural harmony must obtain. The argument will be slightly unusual for its desired conclusion in that it comes from considerations in the philosophy of language, in particular, what we do when we talk about facts or propositions. Obviously, facts about natural language are controversial and non-trivial, and I won’t be able to argue for these largely empirical claims in details now. I will instead present two sides of an ongoing debate, and argue that harmony follows if we take one of those sides. So, for the most part my argument in this chapter is of conditional form: if we take this side in the debate in the philosophy of language, then harmony must obtain and idealism follows. For almost anyone, both sides in the debate in the philosophy of language should seem reasonable. But the conclusion is not merely conditional in the end. I will argue in the next chapter, and have argued in detail elsewhere,⁶ that the side of the debate in the philosophy of language that leads to idealism is indeed the correct one. Thus for now, in this chapter, I would like to focus on the conditional claim. I will also have to simplify in two ways, mostly out of necessity to keep the discussion short enough to fit into one chapter. First, I will largely ignore context-sensitive expressions for now. I will explain below why this simplification is legitimate, and we will revisit this issue in detail in Chapter 5. Second, I will focus on the language in which I write: English. There is a more complex question whether the considerations given below carry over to other human languages as well. The argument to follow does not depend on all human languages being the same in this regard, but the situation gets more complicated if there is variation among human languages in certain ways. We will discuss this issue more thoroughly below, but leave it largely aside for now.⁷ I will thus use English as our language to investigate. Even with these simplifications, the argument to follow should seem significant enough, since it would seem that no conclusion like idealism could possibly follow from considerations about natural language alone, even if that language is English. After presenting the argument I will therefore look more carefully at how an argument like this could possibly establish a metaphysical conclusion like idealism.

⁶ See, in particular, chapters 3, 8, and 9 of (Hofweber, 2016b). ⁷ See also (Hofweber, 2006).

68 idealism and the harmony of thought and reality

Talk about facts and propositions When we talk about facts in English we generally do so most directly with a thatclause or a phrase like ‘the fact that p’. I will call instances of both fact-terms. They occur in examples like: (6) That p is surprising. (7) The fact that p is surprising. That-clauses do not always stand for facts. Sometimes they stand for propositions, as when someone believes that p, but it is not the case that p, and thus not a fact that p. But for any true that-clause there will be a corresponding fact that p. Whether facts just are true propositions, or whether they merely correspond to true propositions, won’t matter for us here. What matters instead is this question: when we use a that-clause or fact-term, are we thereby referring to some thing or entity? Are fact-terms like names for entities which are facts, or are they nonreferential expressions? When I utter (6), am I referring to some entity and say of it that it is surprising? Or am I doing something else with the that-clause, something other than referring? There are good reasons to support either answer, and which one is the right one is actively debated in the literature in the philosophy of language.⁸ One strong reason against that-clauses and fact-terms being referential is the substitution argument. It seems that these expressions do not have a feature expected from referential expressions: that they can be substituted for a co-referential expression without change of truth-conditions. This does not always seem to be the case. For example, there appears to be a difference in truth-conditions between: (8) John fears that his mother will find out. (9) John fears the proposition / the fact that his mother will find out. The former is fear concerning John’s mother, the latter is proposition phobia, fear of propositions themselves, which is different. One of these can be true while the other one is false, and thus substitution of one fact-term for another one seems to be illegitimate. Nonetheless there are also good reasons in favor of fact-terms being referential, and first among them are the quantifier inferences. From (6) as well as (7) it follows that:

⁸ See (Bach, 1997), (Moltmann, 2003b), (Schiffer, 1987), (Schiffer, 2003), (King, 2002), (Rosefeldt, 2008), (Hofweber, 2016b), and many more.

conceptual idealism: the basic idea 69 (10) Something is surprising. But for (10) to be true, it seems that there must be some thing or entity which is surprising. And that thing or entity seems to be just what the that-clause of factterm is referring to. So, substitution speaks prima facie against fact-terms being referential, quantifier inferences speak prima facie for them being referential. Reasonable people can and do disagree on what we should say about this. It is simply a question in the philosophy of language. For all we know, it might turn out one way or the other. If fact-terms are referential, then something needs to be said about the substitution arguments. If they are non-referential, then something needs to be said about quantification. Let’s think a bit more about the second one: what is going on in the quantifier inferences, in particular if that-clauses are not referential?

Quantification over facts and propositions If fact-terms are not referential, how should we understand the quantifier inferences? It won’t do to simply insist that they are not valid, not only since they quite clearly are valid, but also because quantification over facts and propositions plays an important role in communication, and shouldn’t just be dismissed. Instead, we should accept something like the following view of quantification in natural language. This view is congenial to a non-referential picture of that-clauses, but it can be motivated quite independently of it. Whether it is the best such view and whether it is the correct view of quantification in natural language is again something reasonable people can disagree about, but let us simply see where it would take us.⁹ The view is the following: Although quantifiers are often used in just the way indicated above, where they make a claim about a domain of entities, they are not always used in this way. Instead, quantifiers are used in two different ways, and they systematically have two different readings. One reading is the familiar one, which I will call the domain conditions reading, since when we employ it we impose a condition on the domain over which the quantifier ranges. When I say ‘Something fell on my head’ I make an assertion that is true just in case the domain of all objects contains at least one thing which has a certain feature: having fallen on my head. But quantifiers also have another reading. On this further reading they are used for their inferential role. In the case of ‘something’ the inferential role is simply to be able to infer from F(t) that something is F. ‘t’ can hereby be any expression of the appropriate syntactic type, with no regard to its semantic function. ‘Something’ on this reading is more like a placeholder for a ⁹ Alternative views compatible with non-referential that-clauses can be found, for example, in (Schiffer, 1987) and (Prior, 1971).

70 idealism and the harmony of thought and reality particular part of the sentence, in the sense that one can always validly replace a term ‘t’ with ‘something’ without going from truth to falsity. Let us call this reading the inferential reading. On the inferential reading, quantified sentences inferentially relate to other sentences within one’s own language, as opposed to drawing on a language external domain of entities. In this sense the inferential reading is internal to a language, relating sentences in it to each other, while the domain conditions reading is external to it, drawing on a language external domain of entities. In light of this, we can also call the inferential reading the internal reading, and the domain conditions reading the external reading. The terminology of internal vs. external in connection to what there is is generally associated with Carnap’s view in (Carnap, 1956), where Carnap holds that external questions about what there is are defective, while internal ones are trivial. In contrast, on the present proposal, internal and external readings of quantified statements are fully on a par: neither is defective and both have the same general status. They are simply two different readings of the same expression. Quantifiers are polysemous, with different, but equal, readings. That quantifiers have these two readings can be motivated quite independently of our issue of talk about facts. There are a number of quantifier inferences that seem to be valid, but that also seem to be hard to understand on the domain conditions reading of quantifiers. On the inferential reading, however, they are completely trivial, as they seem to be. Examples include: (11) I need an assistant. Thus I need something. (12) I want a unicorn. Thus I want something. To bring out the difference between the two readings with another example, consider: (13) Everything exists. On the one hand, (13) seems to be true. All the things we quantify over, all the things in the domain of quantification, exist. But on the other hand, (13) seems to be clearly false: we know many counterexamples to this universal claim. We know many examples of things that don’t exist: Santa, the Easter Bunny, etc. So, how can everything exist when we know of things that don’t exist? The tension arises, this view of quantification maintains, since two readings of ‘everything’ are at work here. On the domain conditions reading it is true that everything exists, while on the inferential reading it is false. The inferential reading does not allow for counterexamples of the form of ‘t does not exist’. But the domain conditions reading can allow for true instances of ‘t does not exist’, as long as ‘t’ does not refer to an entity in the domain.

conceptual idealism: the basic idea 71 Whether this view of quantification is correct is a topic that reasonable people can again disagree about, just as about whether that-clauses and fact-terms are referential. It is an issue about the function of quantifiers in natural language and ordinary communication. It is tied to what we do in communication, to which phrases in our language do something other than referring, and so on. None of those are obvious or trivial. I won’t aim to try to settle this issue about natural language here, of course, but merely investigate what connections it might have to our larger metaphysical questions. Let us thus take this view of quantification seriously for the moment. Quantifiers are polysemous; they can be used in two different ways: in their domain conditions reading and their inferential reading. How should we understand the inferential reading more precisely? What contribution to the truth-conditions does it make such that the quantified sentence has the inferential role for which we want it? Focusing just on a simple case again, the inferential role of ‘something’ is that any instance is supposed to imply it. That is to say, any instance ‘F(t)’ is supposed to imply ‘something is F’. An instance here is understood simply grammatically, where ‘t’ is an expression in our language of the proper syntactic type that can be combined with a predicate ‘F’ to form a sentence. Since the inferential reading inferentially relates sentences within our own language, the instances that we want to imply the quantified sentence are those in our own language. After all, we want an inferential reading that allows us to infer ‘something is F’ from all the instances ‘F(t)’, and the instances for which we want this for is first and foremost our instances, that is the instances in our own language. Now, what contribution to the truth-conditions would give ‘something’ this inferential role? Here there is a simplest and in a sense optimal solution. We can see what that solution is by first considering the even simpler case of wanting a sentence that has the inferential behavior of being implied by sentence A and also sentence B. Here, too, there is an optimal solution: the desired sentence has to be truth-conditionally equivalent to the disjunction of A and B. It could be the disjunction itself, A ∨ B, or some other sentence equivalent to it. Those are the strongest truth-conditions that have the desired inferential behavior. The same holds for our case with inferential readings of quantifiers. The strongest truthconditions that give ‘something is F’ the inferential role that any instance ‘F(t)’ implies it are being truth-conditionally equivalent to the disjunction of all the instances. Those instances are all the instances of grammatical expressions in our own language, English, that form a sentence ‘F(t)’. Thus the strongest truthconditions and the optimal solution to our problem of what truth-conditions give a quantified sentence its inferential role is this: being truth-conditionally equivalent to the disjunction of all instances F(t) in our language, which we can write as ‘⋁ F(t)’. And since the optimal solution to the problem what truth-conditions give the quantifier the inferential role for which we want it, it is not unreasonable to

72 idealism and the harmony of thought and reality think that those are indeed the truth-conditions of the inferential reading of the quantifier. The truth-conditions for ‘everything’ on its inferential reading are correspondingly equivalent to the conjunction of all the instances, which we can write as ‘⋀ F(t)’ and which gives ‘everything is F’ the inferential role of implying each instance ‘F(t)’. However, this can only be an outline of what the truth-conditions of the inferential reading of quantifiers are in full. We neglected contextual contributions to content, we looked only at the simplest cases of quantifiers, not generalized quantifiers, and so on. The treatment of inferential quantifiers outlined here is thus only an outline. We will discuss quantifiers in more detail below, in Chapter 4, but for now, let us work with the version outlined so far. On the inferential reading of the quantifier the inference from ‘that p is surprising’ to ‘something is surprising’ is valid. ‘that p’ is a grammatical instance of ‘t is surprising’, which implies the quantified sentence on the inferential reading. This inference is valid whether or not ‘that p’ is referential. If ‘that p’ is referential, then the inference is also valid on the domain conditions reading, but even if it is not referential, the inference is still valid on the inferential reading. Thus using quantifiers on their inferential reading in these cases goes together nicely with the non-referential picture of that-clauses and fact-terms. On the other hand, the domain conditions reading goes together nicely with the referential picture. If fact-terms aim to pick out entities in the domain, then quantified statements that quantify over facts should correspondingly make claims about that domain as well. These two combinations are two ways in which our talk about facts might be coherent. On the one hand, fact-terms might be referential and quantifiers are used in their domain conditions reading, on the other hand, they might be nonreferential and quantifiers are used in their inferential reading. Other options are in principle available as well, but these two are the two options that make the most sense of our talk about facts. To put a label on these options, let us call internalism the view that that-clauses and fact-terms are used non-referentially, and quantifiers over facts are used in their inferential reading. On the other hand, externalism is the view that thatclauses and fact-terms are used referentially, and quantifiers over facts are used in their domain conditions reading. This terminology, employing the internalexternal metaphor, seems appropriate, since on the referential picture talk about facts is about something external to the language—a domain of entities which is presumably simply there, waiting to be referred to—while on the non-referential picture talk about facts is not about some language external domain of entities, and quantification over facts inferentially relates to the instances internal to one’s own language. Neither internalism nor externalism should be understood as claiming thatclauses are absolutely always used one way or another. It is up to speakers to use expressions any way they want. The question cannot reasonably be whether

conceptual idealism: the basic idea 73 ‘the fact that p’ is always used referentially or non-referentially, only whether there is a standard use one way or the other, or whether they in general are used one way or another. I can name my cat ‘the fact that snow is white’, and thus use that phrase referentially when I call my cat this way. But this is not what matters for our issue here, and it does not refute internalism nor would a similar example of nonreferential use refute externalism, although such cases need to be acknowledged. What concerns us here is how these phrases are in general, normally, or standardly used. Anyone who uses them otherwise would use them contrary to how they are normally used. And anyone who uses them differently than they are standardly used would not speak of facts as we normally do. Both internalism and externalism are simply views about what we do when we talk about facts. To decide between them we need to look at issues about language, the role of quantifiers in communication, the substitution behavior of fact-terms, and so on and so forth. None of these issues seem to presuppose anything substantial about metaphysics. They are metaphysically unloaded questions about our actual use of that-clauses, fact-terms, and quantifiers. What the right thing to say is here should again be an issue where reasonable people can disagree. Maybe the evidence will point one way or another. And which way it will go will to a large extent be an empirical issue about what we in fact do. But here is the rub: the question whether idealism is true is closely tied to the question how this issue in the philosophy of language turns out. In the next sections I hope to make clear why and how that is so. After that we will discuss how there could possibly be such a connection, one between broadly empirical issues about our own language and a metaphysical issue like idealism.

Internalism and structural ineffability Suppose, at least for the rest of this chapter and until we can revisit this issue in more detail in Chapter 4, that the empirical evidence points one way and internalism turns out to be correct. Suppose that our talk about facts and propositions is as the internalist picture has it. What then becomes of our question about the harmony of thought and reality and its connection to idealism? This question was a question about whether there is a guaranteed harmony between the form of our thoughts and the structure of the facts. And this question in turn is closely tied to the question whether there are structurally ineffable facts, and if not, whether such facts are ruled out for a reason or whether they merely happen not to obtain. These issues can now be resolved. There is a straightforward argument that shows that if internalism is true, then such harmony is guaranteed. Internalism, simply a view about our talk about facts, guarantees that ineffable facts are ruled out and our minds and reality are in harmony. The argument is simply this: If internalism is true then our talk about facts is in accordance with the internalist

74 idealism and the harmony of thought and reality picture, which is to say: fact-terms are non-referential and quantifiers are used in their inferential reading. This internalist picture applies to our present discussion of facts, and it therefore applies to our question whether or not our minds and reality are in structural harmony, i.e. the question whether or not there are any structurally ineffable facts. The thesis that there are such facts we can call the structural ineffability thesis, either for facts or for propositions: (14) There are structurally ineffable facts. (15) There are structurally ineffable propositions. This contrasts with the structural effability thesis, which in turn says that: (16) Every fact is structurally effable. (17) Every proposition is structurally effable. The structural effability thesis claims that every fact or proposition is such that it can be represented in thought or language with a representation that has one of the forms of our representations. The effability thesis contains quantification over facts or propositions, and according to internalism, such quantified sentences involve the inferential reading of the quantifier. This inferential reading, in turn, is truth-conditionally equivalent to the conjunction of all the instances in our own language. Thus the structural effability thesis is truth-conditionally equivalent to one big conjunction, in the case of propositions: (18) ⋀ that p is structurally effable. While in the case of facts it is the following slightly more complex conjunction: (19) ⋀ if that p is a fact, then that p is structurally effable. No matter which case we consider, the result is the same. These conjunctions are true just in case each conjunct is true. But every conjunct is just an instance, in our own language: that snow is white is structurally effable, that grass is green is structurally effable, and so on. Each one of these instances is true. Some instances might be very long and complex, involving billions of words. Such instances might not in fact be representable by any actual human being. The representations involved are just too long; our brains would run out of space and our lives would be over before we would be done representing them. But even in these cases the facts are structurally effable. The form our representations is enough to represent them, even if the size of our brains or the length of our lives is not. The forms of our representations are thus good enough to match the structure of any fact.1⁰ 1⁰ A different argument that structural ineffability is impossible is given by Krasimira Filcheva in (Filcheva, 2020). Filcheva argues that structural ineffability is conceptually ruled out, and that no metaphysical conclusions like idealism follow from this.

conceptual idealism: the basic idea 75 And this is no accident, it has to be so. On the relevant reading, it can’t be that there are some facts that are structurally ineffable. This last sentence can be understood in two ways, corresponding to either there being further facts, or else to there being less representational power of our minds. Of course, we could be worse off and not be able to represent some straightforward facts like the fact that snow is white. If our brains all get damaged, then we might not be able to represent that fact any longer. But the issue is not how we might be worse off while the facts remain otherwise the same. The issue instead is whether the facts could be different while we remain the same so that some facts are now structurally ineffable. Could it be that there are facts with a structure that does not match any of the forms of our thoughts as we now have them? The answer is again: no, assuming internalism. In the very question I just asked I used a quantifier over facts. Such quantification, according to internalism, is equivalent to the disjunction over the instances, our instances. That question is thus equivalent to the question whether it could be that (⋁ that p is structurally ineffable). The instances for ‘that p’ here are just the same as before: all the instances in our present language. And keeping fixed what we in fact can represent, it is false that this could be. All of the ones that are, and all of the ones that could be, are structurally effable.11

Internalism and idealism Whether internalism is correct is a substantial and largely empirical question that goes beyond the scope of this chapter, but which will be the focus of the next chapter. However, let us continue to assume for now that internalism indeed turns out to be correct. We can then conclude that structurally ineffable facts are ruled out, not by accident, but for a reason. No fact or proposition is or can be structurally ineffable. Thus the facts and our thoughts are in harmony: the structure of the facts exactly corresponds to the form of our thoughts. Reality, on one way of understanding it, is just the totality of facts, and thus reality so understood is in structural harmony with our human minds. Reality as the totality of facts is tied to us in just this way. There is a guarantee that our minds are structurally good enough to present every fact that does or could obtain. Or to put it differently, the totality of facts is constrained by our form of thought: it doesn’t and can’t go beyond it. Thus we are central to reality understood as the totality of facts: the limits of our human thoughts are the limits of reality. The overall story of reality will have to mention this connection between our minds and reality. And thus strong idealism is true. We might not be central to the totality of things,

11 How internal quantification interacts with modality is discussed in more detail in (Hofweber, 2006), as well as below in section 4.4. The issue is a little more complicated once inferential quantifiers are formulated to allow for context-sensitive instances, but the conclusion remains the same even then. We will see more on all this in Chapter 4.

76 idealism and the harmony of thought and reality and thus ontological idealism might well be false. But we are central to the totality of facts: our forms of thought are limiting what facts do and can obtain. Idealism is thus true; not ontological idealism, but alethic idealism. In general, we might have nothing to do with what there is, but we are central for what is the case. Not because the obtaining of facts is, in general, tied to us, but because the range of what facts can in principle obtain is tied to us. These two ways in which facts might depend on us should be clearly distinguished. Since facts correspond to true propositions, let us say that a fact truth-depends on us just in case its obtaining counterfactually depends on us. In this sense facts in general do not depend on us. Which propositions are true is in general, of course, not counterfactually dependent on anything having to do with us. But we now also have another notion of dependence to consider. Let us say that the totality of facts range-depends on us just in case the range of all the facts, that is, which facts can in principle obtain, is tied to us. And in this sense facts depend on us, assuming internalism: that every fact is effable is not an accident, but obtains for a reason tied to us. The range-dependence of the totality of facts on us, but not their truth-dependence, is how we constrain the world without constructing it. We limit what reality as the totality of facts can be like, since any fact in that totality must be representable by us. But we are not responsible for which of these facts obtain, i.e. for which of these true propositions are true. Constraint without construction is therefore possible after all, and this is how can be. Overall then, we can call conceptual idealism the version of alethic idealism which holds that reality as the totality of facts is range-dependent on us.12 Conceptual idealism is the form of idealism defended in this book. It supports that our human minds constrain the world without constructing it, and it makes clear why and how our minds are metaphysically central to reality as the totality of facts. The range dependence of reality as the totality of facts on our minds is closely connected to harmony obtaining for a reason. The argument given above hoped to show that internalism guarantees that harmony has to obtain, and that thus reality as the totality of facts range-depends on us and our minds, as conceptual idealism would have it. The structures of the facts are tied to the forms of our thoughts so that there has to be a harmony between them. And that is idealism. This argument must seem highly suspicious. How could considerations merely about our own language show that there is a harmony between our language and reality, or correspondingly between our minds and reality? And how could such an argument possibly support a metaphysical thesis like idealism? 12 The name ‘conceptual idealism’ is also used by Nicholas Rescher for a completely different view in (Rescher, 1973). This reuse of names for versions of idealism is unfortunately hard to avoid given how many forms of idealism have been defended and named in the past. ‘Conceptual idealism’ is a fitting name for the view defended here, since it concerns the aspect of reality that are represented conceptually in a thought or judgment. That aspect of reality is not independent of our conceptual representations, according to the view, while other ones might well be.

conceptual idealism: the basic idea 77 There is a more specific and a more general objection in this neighborhood. The more specific one concerns whether I made confusion of levels when I argued that harmony obtains for a reason tied to us. Harmony by itself could obtain by accident, and thus by itself it is not enough to support idealism. But if harmony obtains for a reason, and the reason is somehow tied to us, then this can support idealism. Similarly, the range-dependence of the facts on us requires more than simply the obtaining of harmony, it requires that harmony obtains for a reason tied to us. Now, I argued above that there is such a reason: that every fact is effable is true because, in a nutshell, ‘every fact’ is a quantifier that generalizes over our own instances. Thus it is no accident that every fact is effable, but it holds for a reason connected to us. But one could object that this argument confuses a reason for the truth of ‘every fact is effable’ with a reason for every fact being effable. The former is metalinguistic and concerns reasons for the truth of a sentence, while the latter is metaphysical and concerns reasons for facts being effable. Maybe we have seen that there is a reason tied to us for the truth of the sentence, but that is not enough for idealism, since for that we wanted a reason tied to us for every fact being effable. It is true in general that a reason for the truth of a sentence does not have to be a reason for the obtaining of the fact expressed by that sentence. That the sentence ‘every fact is effable’ is uttered in English, say, might be a reason for its truth, but that this sentence is in English is not a reason for the obtaining of the expressed fact. But sometimes there is a connection between these levels that carries over. In our case it works like this: every fact is effable, since ‘every fact’, as it was just used, generalizes over our instances, and so each of these facts is effable. Here there is a connection between the obtaining of a particular fact, namely that every fact is effable, and the language we used to express that fact, in particular the quantifier ‘every fact’. We want to find a reason for the obtaining of a particular fact, a fact we represent in our language. What that fact is like, and under what conditions it obtains, is thus tied to the language that was used to express it. We can see that this fact obtains just in case each and every instance obtains, since those are the truth conditions of the quantifier used to express this fact. And these instances are the instances in our language. Thus part of the reason why the fact that every fact is effable obtains is tied to us: it is because the instances in our language figure in what needs to be the case for every fact to be effable. And thus there is a reason tied to us for why it is no accident that every fact is effable. It is not a lucky accident that every fact is effable, but it has to be so for a reason, a reason tied to us. But besides this particular worry about the argument, there is also a general worry about the whole approach. All we did is look at our minds and language; reality didn’t seem to play any role at all. But to show that our minds and reality are in harmony it would seem that one has to look at two things—our minds on the one hand, and reality one the other—and show that these two go together in

78 idealism and the harmony of thought and reality the right way. But it can’t be, or so is natural to think, that one can just look at one of the two sides and from that alone conclude that the two go together. Furthermore, it is natural to think that it can’t be that one can draw metaphysical conclusions simply from considerations about language. The above argument must be mistaken, since it would need to overcome the insurmountable languagemetaphysics gap: the impossibility of drawing metaphysical conclusions from premises about language alone.13 From considerations about language one can’t draw conclusions about what reality is like, only about how we represent reality to be. And thus the above argument must be mistaken. To be sure, one could argue for metaphysical conclusions from considerations about language together with the further assumption that certain representations in that language represent reality correctly. But I didn’t make any assumptions about which sentences are true, only what the semantic function and general use of that-clauses and quantifiers is. Such considerations are merely about our representational goals and what we hope to do, and those alone aren’t a guide to larger features of reality. And from our hopes we can’t conclude much about what reality is like, since our hopes might remain unfulfilled, and our attempts at representation might fail and be completely inadequate to capture reality. The above argument must therefore be mistaken, and there are a couple of places where it might be tempting to argue that it goes wrong there. Or the argument must not be based on as innocent assumptions as I made it out to be. Maybe I smuggled in some idealism at the beginning in what I claimed were only largely empirical considerations about our own language. However, I will not draw these conclusions here. Instead I hope to show that these reservations are mistaken. The main argument given here does indeed assume only something about our own language, something largely empirical, and it does not smuggle in any metaphysical assumptions. Nonetheless, idealism follows. Idealism follows simply from considerations about our own language. Idealism can be established on largely empirical grounds, by thinking about our own language, and the language-metaphysics gap can be bridged, or so I hope to make clear now.

3.3 How the argument is possible Let’s first look at an example of how the language-metaphysics gap can be bridged. This example should motivate that it is indeed sometimes possible to draw metaphysical conclusions simply from considerations about language alone, making no assumptions about which sentences are true or which phrases carry out their semantic function successfully. As my example I would like to consider the

13 Heather Dyke calls drawing such conclusions ‘the representational fallacy’ in (Dyke, 2008).

conceptual idealism: the basic idea 79 question whether or not facts exist. That is to say: are facts, like the fact that 2 + 2 = 4, entities which exist? Does the totality of all that exists include facts besides other things? This is a metaphysical questions about what the world is like, not a question about our language or our minds. But still, we can answer it from consideration about language alone, and here is how. Consider again the question whether that-clauses are non-referential expressions. Being non-referential can be understood in two different ways: either about language alone or else about the relationship between language and reality. The second way to be non-referential is paradigmatically exhibited by an empty name: a proper name that fails to refer to anything. Such an empty name has a semantic function—the function to refer to an object—but it fails to carry out that function. It is non-referential even though it aims to refer, since what it aims to refer to is not there. Whether a phrase is non-referential in this sense is thus not about language alone but also about reality. But other expressions are non-referential in a different sense. Such expressions have a completely different semantic function than referring. They include expressions of a grammatical category quite different from that of a name or singular term, but they also include phrases that can appear in subject position. Take ‘nothing’ as an example. It is a quantifier, and quantifiers are not referring expressions. Their semantic function is to make claims about a domain of entities, at least in its most familiar use, which I will rely upon here to illustrate my main point about reference. ‘Nothing’ is non-referential, not in the sense that it aims to refer and fails to achieve its aim, but in the sense that it does something completely different semantically than referring. Quantifiers are not alone in this regard. On most views about natural language, most expressions are non-referential in just this sense. They include expressions like ‘very’, ‘many’, ‘if ’, ‘tall’, and so on and so forth. Although these expressions are of various different syntactic categories, and have various different precise semantic functions, they all have in common that they are non-referential in the sense that they semantically do something other than referring. If internalism is true, then that-clauses are non-referential in the same sense in which ‘nothing’ is non-referential. It is not that they aim to refer and fail to achieve their aim, but that what they do is something altogether different. And this has consequences for what reality is like. Let me illustrate this first with a different example: ‘nothing’. Let us call, by stipulation, ‘The Nothing’ whatever the word ‘nothing’ refers to, if anything. Is there such a thing as The Nothing? We can conclude, from considerations about language alone, that there is no such thing. ‘Nothing’ is non-referential, since it does something else semantically than referring, and so none of the things that there are is The Nothing. The Nothing just was, by stipulation, whatever ‘nothing’ refers to. But that phrase does not refer to anything. So, whatever there might be, whatever reality might contain, none of it is The Nothing. Reality might contain all kinds of things, but we know at least this: none of them is The Nothing. And that none of them is The Nothing was

80 idealism and the harmony of thought and reality determined by considerations about language alone, together with a stipulation of what The Nothing is. Similarly with that-clauses and talk about facts or propositions. Facts, if there are any at all, are just the kinds of things that fact-terms stand for: the fact that snow is white, the fact that 2 + 2 = 4, and so on. But if internalism is true, then thatclauses and fact-terms like ‘the fact that snow is white’ and ‘the fact that 2 + 2 = 4’ are non-referential. They are non-referential not because they aim to refer and fail, but because they do something else semantically than referring. So, whatever things there might be, whatever entities might be part of reality, none of them is the fact that snow is white. To make this clear, just consider the last part of the last sentence just written: ‘none of them is the fact that snow is white’. The quantifier ‘none of them’ ranges over the domain of things on its intended use, the things that are part of reality. But ‘the fact that snow is white’ as it was just used, according to internalism, is not a referring expression. It does something different than picking out some entity in the domain. Thus none of these things in the domain is picked out by this phrase, and so none of them is the fact that snow is white. And similarly for all the other examples of facts. Whatever things there might be, none of them is the fact that snow is white, just as none of them is The Nothing. Even though ‘the fact that p’ and ‘The Nothing’ are non-referential, the sentences ‘no entity is the fact that p / The Nothing’ are perfectly grammatical, according to internalism. They are grammatical and true, and we can see that they are true simply by reflecting on what it means to be non-referential. Thus if internalism about talk about facts is indeed correct, then none of the things contained in reality are facts. No facts exist, there is no ontology of facts, and facts are not entities or things. There could be all kinds of things, but whatever they might be, none of them is a fact, again assuming internalism. Facts still obtain, but they do not exist. It is no help to try to avoid this conclusion by insisting that one takes a fact just to be, say, an ordered pair of an object and a set of objects, or some other specified entity. There might well be such ordered pairs, but if internalism is true, then those are not facts. If internalism is true, then the fact that John is tall is not an ordered pair. The fact-term is not referential, and thus does not pick out any object in the domain, including any ordered pair that might be a member of the domain. And the same holds for any other fact-term. No entity in the domain can be identical to the fact that p, no matter which instance of ‘p’ we consider.1⁴ And this is a metaphysical conclusion drawn solely from considerations about our own language. Internalism guarantees that there are no facts, understood as involving the domain conditions reading of the quantifier, and it guarantees that facts do not exist. Simply from considerations about our own language alone we are thus able to draw a conclusion about reality.

1⁴ The significance of internalism for ontology is developed in detail in (Hofweber, 2016b).

conceptual idealism: the basic idea 81 This bridges the language-metaphysics gap, at least for this particular case. But it does so in a different way than the way that is objectionable. The objectionable connection was to conclude from our attempts to represent the world that the world is more or less as we represent it. That poses an insurmountable gap, since no guarantee can be given, without further argument, that we succeed in what we aim to do when we represent the world. No such assumption is being made in the above argument, but it still must seem puzzling how we can conclude from merely facts about our language alone, with no regard to which representations are true or accurate, that reality beyond us and our language is a certain way. In particular, how can we do this unless the conclusion we hope to establish is an analytic or conceptual truth? To get clear on this it is useful to focus not on the answer we hope to establish, but on the question we originally asked. The question we hope to answer is stated using our representations: we ask the question in language or formulate it in thought. As such there is a connection between the question and our representations. In particular, it might well be that some of the representations employed in the question are such that the answer must go a certain way. And just this is the case in our example. When we ask whether facts like the fact that 2 + 2 = 4 exist, then we are using fact-terms in the question itself. And if it turns out, by thinking about language alone, that these are non-referential expressions, then we can answer this question by thinking about language alone, even though the answer is not an analytic truth. All this can only be an outline so far, but the shift towards focusing on the questions we ask, and not just on the answers we hope to give, will be an important theme in what is to come. I will elaborate in more detail below how this method for answering metaphysical questions can work in general as well as in this particular case. We will return to this in detail in Chapter 6 below. I hope that for now at least the worry that no argument of this kind could possibly establish a metaphysical conclusion can be put aside until we revisit this issue more thoroughly below. That leaves a second worry, namely that even if this argument is correct, it couldn’t possibly establish what I claim it does: idealism. This is what we need to pay attention to next.

3.4 But is it really idealism? It must seem that the whole argument given so far is a cheat or a trick. It doesn’t seem like the right kind of argument that could support idealism. And this is the natural way to feel about it, I know the feeling myself. After all, the question is about the harmony between two different things: the facts and our conceptual representations of them in thought or language. No argument looking at just one of the two should be able to conclude that both of them are intimately related. But I want to maintain nonetheless that the above argument does indeed

82 idealism and the harmony of thought and reality establish idealism, assuming internalism. The reason ultimately is that internalism shows that the totality of facts and our representations of the facts are not two independent things. And since they are connected, looking at one can inform us about the other. Internalism is not only a view about our language, it in addition leads to a larger metaphysical picture of the fact-like or propositionlike aspect of reality. And it is now in order to not just argue that the languagemetaphysics gap can be bridged, but to make clear how these considerations about our language indeed support idealism. In this section I would like to work this out by considering a series of objections, why they are misguided, and why idealism indeed follows from all this. The first objection to the main argument is that if internalism is indeed correct, then idealism was badly stated above. Internalism guarantees that there are no facts, that is to say, that facts do not exist. And if facts don’t exist, then the totality of all facts doesn’t exist either. But then, how can it be legitimate to think of reality as the totality of facts? And how can it be legitimate to think of idealism as alethic idealism, i.e. as a position concerning our place in reality understood as the totality of facts? If internalism is indeed true, then it was a mistake to think of reality in this way. And consequently it was a mistake to think of idealism as being either ontological idealism or alethic idealism. Idealism must be understood differently, the objection continues: either only as ontological idealism or in some completely different way. Thus the above argument misses its target, especially if its starting point is correct: If internalism is true then this undermines the given formulation of idealism rather than establish idealism. Although this objection might at first seem compelling, it is nonetheless mistaken. What is uncontroversial here is that facts obtain, what is controversial is whether in addition facts exist. That 2 + 2 = 4 is a fact can be agreed upon by all in this debate, but whether there is an entity which is the fact that 2 + 2 = 4 is what is at issue. As long as one does not deny that facts obtain one can understand reality as the totality of all those facts that obtain, and for this it does not matter whether or not the totality of facts itself is an entity. All the facts that obtain together give us one way of thinking of what reality is like. Thinking of reality as either all there is or all that is the case does not depend on facts themselves existing. If anything, it is the other way round. Those who think that facts exist must also hold that the totality of things, which is the totality of what exists, includes the totality of facts. After all, the facts are just some of the things or entities which exist. The facts are thus subsumed under the things. But if a substantial distinction is to be drawn in this regard—what exists is one thing, but what is true or what is the case is a further issue—then facts must themselves not be part of what exists. Those who proclaim that the world is the totality of facts, not of things, usually have such a substantial distinction in mind. And that would be just so if facts obtain, but do not exist. Whether they exist is, of course, a substantial question, but their existence should not be seen as required for the distinction between two senses of reality.

conceptual idealism: the basic idea 83 To the contrary, the distinction is more significant if facts obtain, but do not exist. Thinking of reality as all that is the case is legitimate as long as facts obtain, and thus our conclusion about the place of our minds in reality stands, since the obtaining of facts is not denied, only their existing. Still, the worry remains that the idealist only wins in letter, but not in spirit. Maybe we are central to the facts, given internalism, but that just shows that the real metaphysical question is not about the facts, but about something else. Maybe we need to hone in on a different, more metaphysically appropriate conception of reality, or at least focus on some other, more suitable question. When we ask whether or not all the facts are tied to us, then we are in a sense too involved in the question itself, assuming internalism. We are then simply generalizing over the instances in our own language and it is thus no wonder that we are central. But maybe there is a better question to ask, one that isn’t so closely tied to us and that more properly captures the real issue tied to idealism. Let us call the proper question the question we should be asking when we ask about idealism. Internalism shows, or so the objection, that the proper question is not to be asked in terms of facts. Internalism does not answer the proper question, but rather points to an error in the original articulation of what we took to be the proper question. All internalism shows is that we need to formulate the question differently, not what the answer to the question is. But how should we state the question instead, and why should we think that this is indeed a better way to state it? There are two main ways to try this, one moderate and the other more radical. The moderate way attempts to state the proper question in other familiar terms. Maybe it is the question whether we are central for all truths, or all true propositions, or all contents, or something like it? But it is not hard to see that this approach will go nowhere, for at least two reasons. First, whatever reasons we will have for thinking that internalism is true for talk about facts will carry over to a reason for internalism about talk about truths, true propositions, contents, and so on. All of those are talked about with that-clauses, and the question of internalism vs. externalism is essentially the same for all of these cases. That is why internalism or externalism will apply uniformly to the propositional more broadly, all the things we talk about with that-clauses: facts, truths, propositions, reasons, contents, and so on. Second, there are close connections between the concepts of a fact and a truth and a true proposition, and others that belong to the propositional. For every fact there is a truth, and for every truth a fact. And for every true proposition, there is a truth, and so on. Whether facts are identical to truths, or just correspond to them, is thereby left open, but the correspondence has a status close to a conceptual truth. Thus the real issue is about the propositional more generally, and it is about a whole web of notions that are closely tied together. A moderate change from one part of this web to another won’t lead to a completely new question. If internalism is true, then we are not just central to the facts, but to the propositional generally. If we didn’t ask

84 idealism and the harmony of thought and reality the proper question already, then moving to a nearby alternative is not the way to get to it. But maybe something more radical will do better. Talk about the facts is too closely tied to us to be suitable in stating the proper question, one might object, in particular, since quantification over facts on the internal reading merely generalizes over our instances. We can also quantify externally over facts, after all, speakers can use quantifiers on any reading they want, and both readings are always in principle available. But when we do so such quantification is always vacuous, since, as we saw above, the domain of entities does not contain any facts. Nonetheless, we might hope to find some other entities in the domain, entities that are not tied to us and our representations, and then externally quantify over them to state the proper question. Facts won’t work here, but some replacement of facts, call them facts*, might. Facts* could be some entities or other, maybe sets of worlds, or ordered pairs of things and sets of things, or maybe some things that have features fitting a pre-theoretic conception of what facts are like, or what have you. The proposal then is to state the proper question in terms of facts*, maybe as the question whether we are central to the totality of facts*. This radical proposal strikes me as better than the moderate proposal above, but it, too, should be rejected. It is not a problem to introduce facts*, this can be done coherently in many different ways. And we would indeed ask a quite different question when we ask about our place among the facts*. The problem with this proposal is why we should think that we are now asking a better question. And here it is hard to see what reason we might have for thinking so. Facts* are not facts, despite having a similar name. We have good reason to find out about the facts, this is just what we do when we try to find out what is true. Maybe some facts* correspond to facts in a way that can be made clear, and then we would have a reason to find out about those facts*, via our good reasons to find out about the facts. But it is hard to see what reason we would have to find out about the facts* and our place among them that goes beyond finding out about the facts. Facts have a special status when it comes to carrying out inquiry. Inquiry aims at the truth, and the truth corresponds to the facts. Thus it is at first natural to care about the facts, and thus to see reality as the totality of facts. It might also be nice to find out about facts*, whatever they might be more precisely, but should we replace finding out about the facts with finding out about the facts*? Should we give the facts* a more distinguished place than the facts? Is asking about the facts* an improvement over asking about the facts? The answer has to be ‘no’. The activity of trying to find out itself has a constitutive aim: the truth. Finding out aims at what is true and thus at the facts. Finding out is inquiry, which aims at the truth and the facts, and thus at reality understood as the totality of facts. To properly pull off the radical strategy, one has to be even more radical: one has to give priority not just to the facts* over the facts but also to the other connected starred notions over their non-starred equivalents. Thus we

conceptual idealism: the basic idea 85 should better carry out inquiry* which aims at the truth* and what facts* obtain or exist, rather than plain old inquiry, truth and facts. But what reason could we have to abandon inquiry in favor of inquiry*, to find out what is true* rather than what is true? From our present standpoint, while thinking about what we should and should not do, we are assessing what reasons we have for or against a change in our projects, including the project of inquiry itself. But there seems to be little wrong with inquiry as such. It might be suspicious that inquiry concerns the facts and if internalism is true, then we are central to the domain of inquiry itself. That might seem too idealistic and a disappointing outcome for some, but it is hardly a good reason to overthrow the whole web of notions connected to the propositional and replace them with starred substitutes, be it for inquiry generally or in the part of it that is metaphysics. Nonetheless, a feeling of a limitation given our situation must remain. To bring it out, we can imagine simpler creatures than us who speak a limited and primitive version of English, without some of the forms of representations that we have. They can go through the same argument given here for their language and conclude, as they would put it: ‘There are no facts which are structurally ineffable for us!’ However, we know that the structure of the facts is not constrained by their thoughts. Some facts obtain which are structurally ineffable to them and we could give an example relying on our extra forms. But then, why shouldn’t we think that other creatures, powerful aliens, say, could look down at us like we look down at them? Such creatures might conclude that we human beings can’t capture all the structure among the facts and they might even give examples of facts structurally ineffable for humans. Maybe then we can’t quite articulate how we are limited, but we should conclude nonetheless that we are limited, and not the constrainers of reality.1⁵ But while this feeling of a limitation is undeniable, we can reason quite conclusively that it is misguided, assuming internalism is true. First, we should note that those speaking the impoverished version of English speak truly when they utter ‘There are no facts which are structurally ineffable for us!’ But this sentence does not mean what it means in English. In particular, it does not mean that there are no facts which are structurally ineffable for them. These sentences have different truth-conditions and mean different things in full English and impoverished English. Both are true, but only one means that there are no structurally ineffable facts, namely the one in full English. Furthermore, we can conclude, given internalism, that no other creature can truly say that there are facts structurally ineffable for humans. They might utter sentences that sound like this, but these sentences would either not mean that or else they would not be true. We know, assuming internalism, that all facts are structurally effable by

1⁵ See (Nagel, 1986, 95f.) for an argument of this kind.

86 idealism and the harmony of thought and reality us, so no one can truly say the opposite. And so, even though it seems at first plausible that other creatures can say something similar about us than what we can say about simpler creatures, we can reason conclusively, assuming internalism, that this feeling is misguided. We can truly say about the simple creatures that there are facts structurally ineffable for them, but no one can truly say that about us. None of this should be taken to imply a form of relativism about facts. It is not to say that which facts there are is relative to a language, and that there might be fewer or more facts relative to poorer or richer languages. Some languages can be poorer and represent fewer structures of facts, but no language can be richer, since we already can represent the structures of absolutely all facts. Instead there is only one totality of the facts, reality as all that is the case, but it is tied to our human representational capacities. That is not relativism, but idealism. But maybe the advanced aliens can say something analogous about us to what we say about the simpler creatures, maybe not about facts, but about something else? Maybe they can state in some other way that we are limited? And, of course, often they would be correct. We are limited in many ways: we might not have the spaceships or the laser guns the aliens have, but such limitations are very different than the one that is at stake. We are limited in many ways, but are we limited when it comes to inquiry and representing reality and the facts? The answer, at least concerning structure, is ‘no’, and no other limitation supersedes this. We cannot rationally accept that asking about something else, facts*, and our place among them is a better question to ask than asking about facts. So, even though we are clearly limited in many ways, we are central where it counts. Internalism makes clear that the totality of facts is constrained by our human minds and thus we are unlimited even though other creatures are not. A feeling of a relevant limitation surely remains, but we have good reasons to conclude that this feeling is misguided.1⁶ Instead of seeing internalism as a limitation that should be overcome, I would like to suggest instead that it gives us an insight into what reality is like. We did state the proper question all along when we asked whether we are central to reality and whether there can be structurally ineffable facts. Reality can be understood as the totality of facts, even if and especially when facts are not entities. So understood reality is tied to us, since there are no structurally ineffable facts, in principle and for a reason, assuming internalism. That we asked this question concerning idealism by talking about facts is not undermined by internalism about talk about facts. The question remains intact as the proper question. Instead of undermining 1⁶ The problems and outlined solutions from this section will be developed in much more detail in Chapters 5 and 6 below.

conceptual idealism: the basic idea 87 the question, internalism points to its answer, and that answer is idealism. Instead of thinking of internalism as a limitation for what we can say, we should think of it as uncovering a metaphysical picture of the propositional or fact-like aspect of reality. And this picture holds that what exists might well simply be there, with no involvement on our part, but what is the case is different. Facts obtain, but the totality of facts is not simply there, waiting for us to represent them. Which facts obtain is not due to us, but the range of the facts which can in principle obtain is tied to us: they must be representable by a representation of the kind we have available. This points towards how ontological realism might be compatible with alethic idealism. What there is is independent of us, but no matter what there is, it can only figure in facts with a structure matching our forms. Since we have forms like subject-predicate representations, there is no tension between this kind of realism and that kind of idealism. There can be all kinds of things, and we might have nothing to do with which ones those are, but whatever they are, they can only figure in facts that have a structure that matches one of our forms. Since we have subject-predicate forms, there is no restriction on which facts with object-property structure these things figure in. Thinking about harmony makes clear how all this can coherently be so. The totality of facts range-depends on us, but facts do not truth-depend on us. This range-dependence holds, since any fact that can obtain has to be representable with a thought of the form that our thoughts have, and this limits what reality can be like. Range-dependence is, in essence, the dependence requirement that arises from harmony being required. And thus thinking about the required harmony of thought and reality leads to a sense of dependence of reality on us and our minds, and with it to idealism. This defense of idealism meets the constraints listed above on page 8: It explicitly formulates idealism and whatever notion of dependence it might rely on, since range-dependences is explicitly stated. It is compatible with what we generally know to be true, since conceptual idealism does not hold that the obtaining of the facts depends on us, only the range of the facts does. And it can be supported with an argument: the argument outlined in this chapter together with a more detailed defense of internalism. All this meets the goal we set for a defense of a strong form of idealism and a third answer to the Big Question about our place in the world. The overall metaphysics of all of reality will have to bring out this connection between us and reality. The metaphysical story of reality will have to mention us human beings in the main text, not just in a footnote. And so we are metaphysically central to reality. Our minds are central, not because reality is mental or constructed by our minds, but because there is a guaranteed structural harmony between which facts

88 idealism and the harmony of thought and reality could in principle obtain and which facts can be represented by us. Our minds limit what can be the case. We constrain reality without constructing it.

3.5 Filling in the picture How such a version of idealism works more precisely as a metaphysics is largely left open by what I have said so far. This chapter doesn’t aim to work out the idealism that would follow, nor did it aim to defend internalism, which would imply it. My goal in this chapter was simply to argue that a strong anthropocentric form of idealism can be defended from an unexpected angle: via a defense of internalism, a position merely about what we do when we talk about facts. Even though the assumption in this argument is only about our own language, nonetheless substantial metaphysical consequences follow. The language-metaphysics gap can be overcome in this case, and metaphysical conclusions can be drawn from considerations about our own natural language alone. To argue for this at least in outline was the goal of this chapter. Hopefully it is clear that this version of idealism is distinctly different from the ones discussed in Chapter 2. But before we can truly contrast conceptual idealism with other versions of idealism, and argue that conceptual idealism does not suffer from the problems that the other versions of idealism faced, we will need to see some more of the details and spell out the picture better. In particular, there are several issues that arose in this chapter that need to be filled in for the picture to come together properly. The goal of this chapter was merely to give an overview of the idealist position I hope to defend, to outline the argument that is supposed to defend it, and to raise some issues and concerns that need to be addressed in this defense of idealism. In the next couple of chapters I hope to develop these issues more properly, which will not only fill in some gaps, but in particular will also allow us to see that conceptual idealism, once properly formulated, is a notably stronger position than spelled out in this chapter, and has more significant consequences than what could become clear so far. To get there we need to address several issues in turn. First among them is the formulation and defense of the internalist picture of talk about facts. We only saw an outline of how this position should be formulated, and what reasons we might have for its truth. In particular, above we only discussed the context-insensitive version of internal quantification over propositions and facts, and we know that this can’t be quite correct. In the next chapter I will present my reasons for why we should think that internalism is indeed correct, and I will formulate a context-sensitive version of it, which will become important later on, in particular when we need to discuss the relationship between ontological and alethic idealism. After internalism has been clarified we will be in a position to more properly discuss how the harmony of thought and reality should be understood and how

conceptual idealism: the basic idea 89 the impossibility of ineffable facts is compatible with various contingencies and changes about what we can express in our own languages. After that we will need to revisit how the argument for conceptual idealism was possible, and whether it merely highlights that we should change some of our key concepts, like that of a fact. The proper resolution of this issue will give rise to an approach to metaphysical question which is in certain ways neo-Kantian, and which leads to a larger project about how one can defend philosophical theses, a general project of which the present defense of idealism is just one special case. Finally, we will be in a position to discuss the overall metaphysical picture connected to conceptual idealism.

4 The internalist conception of facts In the last chapter I outlined an argument for a conditional claim: if a certain view about our talk about facts—internalism—is correct, then a particular form of idealism—conceptual idealism—follows. In the next couple of chapters this argument and the resulting idealist position will be clarified, developed, and strengthened in various ways. I will start in this chapter by defending internalism. Internalism about talk about facts is simply a view about what we do when we talk about facts. It is thus a broadly empirical position about our own language and our own speech. As such it needs to be defended by looking at what we in fact do when we talk a certain way. The question of the truth of internalism thus belongs to the philosophy of language and linguistics, not to metaphysics directly. This chapter will therefore be just about language, not any metaphysical consequences that might follow. And the reasons for and against internalism should similarly come from the kinds of considerations that are appropriate for an investigation into our own language. In particular, it is not a good strategy to argue from metaphysical considerations towards the truth or falsity of internalism. One can’t hope to establish internalism on the basis that one hopes idealism is true, together with a belief that idealism follows from internalism. And similarly, one can’t hope to argue against internalism on the basis that one finds idealism unacceptable. We will need to evaluate internalism and externalism on their own merits as position about natural language. After having done so we will look more closely at any alleged metaphysical consequences in the chapters to come. To settle any question related to natural language is a substantial task. I can’t hope to address all the issues that need to be addressed in this single chapter, but I can and will present my main reasons for why I believe internalism is true. I have developed internalism and the arguments in its favor in more detail in my last book, (Hofweber, 2016b), and I will refer back to the discussion there when we won’t be able to dig further in this chapter. For those who find the defense of internalism too brief, I would encourage them to look at the details in (Hofweber, 2016b) or to continue to treat the defense of idealism conditionally: if internalism is indeed correct then idealism follows. Even the conditional claim should be rather non-trivial, and many of the philosophical issues that arise in the following

Idealism and the Harmony of Thought and Reality. Thomas Hofweber, Oxford University Press. © Thomas Hofweber 2023. DOI: 10.1093/oso/9780198823636.003.0004

92 idealism and the harmony of thought and reality chapters apply to the conditional claim just as well as the unconditional one. I, however, hold the unconditional claim, and will present my main reasons for why in this chapter. But besides the issue of the truth of internalism, there is also an issue about its proper formulation. The version outlined in the last chapter was simplified in a way that matters for what is to come, and it is part of the task of this chapter to better formulate internalism and to get rid off the simplifications that will make a difference below. Internalism about (talk about) facts has two connected parts: the semantics of fact-terms and the semantics of quantifiers over facts. Fact-terms can either be simply that-clauses, as in ‘that snow is white’, or more complex, as in ‘the fact that snow is white’. Similarly, I will use ‘proposition-term’ to include that-clauses and more complex terms like ‘the proposition that p’. That-clauses can stand for facts or for propositions, and facts might just be true propositions, or facts and propositions might be different. The question about internalism is very similar for both talk about facts and talk about propositions. I will use talk about facts as our main example in the following, but much of what we will see carries over to talk about propositions as well. In fact, much will carry over to any talk that is tied to that-clauses: meanings, reasons, contents, and so on. For now, let’s start with talk about facts. The first part of internalism about talk about facts concerns whether fact-terms aim to refer to any entities. Do phrases like ‘that snow is white’ or ‘the fact that snow is white’ have the function to refer to an object or entity which is a fact? The internalist holds that such terms are non-referential expressions, in the sense that speakers normally do not use these terms with the aim to refer, and that their semantic function is something other than referring. The second part of internalism concerns quantification. Quite clearly, the following quantifier inference is valid: (20) That Sue ate the apple is surprising. (21) Thus: Something is surprising. Internalism does not take such quantifiers to range over a domain of entities, one of which is referred to with the term ‘that Sue ate the apple’. Although often quantifiers are used in just such a way as to range over a domain of entities, the internalist holds that not all uses of quantifiers are the same in this regard. Instead, sometimes quantifiers interact with non-referential complements for a certain inferential role. Quantifiers thus have two readings, the domain conditions reading and the inferential reading. And when they interact with fact-terms, they are used in their inferential reading, or so the internalist position in outline. We should now look at both parts more carefully: the internalist view of fact-terms and the internalist view of quantification over facts.

the internalist conception of facts 93

4.1 Fact-terms A term can be non-referential in two senses, and this difference was important for the main argument for conceptual idealism. First, there are those terms that aim to refer, but fail to succeed in their aim, and thus they do not refer. A paradigmatic case of this is an empty name, which semantically is just like any other name, but which does not succeed in referring, since the object it aims to refer to does not exist. We can call such terms de facto non-referential, since they aim to refer, but as a matter of fact fail to succeed in their aim. ‘Betty Crocker’ is likely an example of such a name. This name was used in advertisement for a presumed recipe author and brand figurehead, but doesn’t refer to any real person.1 Second, there are those expressions that do not aim to refer at all, but have a different semantic function altogether. They are non-referential not because they aim to refer, and fail, but because they do something other than referring altogether. We can call such expressions constitutively non-referential, since their very semantic function is something other than reference. Most expressions in natural language are likely constitutively non-referential, including words like ‘very’, ‘however’, and so on. In fact, many would accept that most expressions in natural language are constitutively non-referential, with the only real issue being about terms one could broadly classify as ‘singular terms’, or at least ‘terms’. Such terms here can be arguments of predicates, and in particular appear in subject-position in a sentence. Both proper names as well as fact-terms are terms, as indicated by examples like that: (22) a. Fred is revolting. b. That Fred kicked the cat is revolting. Since quantifiers can be terms in this broad sense, in particular definite descriptions, which on the standard Russellian line are quantifiers, not all terms are referential, since quantifiers clearly are not.2 But the question remains whether all terms are either quantifiers or referential. Those who accept that all nonquantificational terms are referential generally are driven by a picture of language which we could call the referential picture of language. On this picture, the semantic function of simple subject-predicate sentences is to either refer to an object and attribute a property to it, or to impose a condition on the domain of objects, in

1 Whether or not names like ‘Betty Crocker’ are truly non-referential is controversial, since some philosophers hold that they refer to abstract objects instead of material objects, and thus are referential. See (van Inwagen, 1977), (Thomasson, 1999), and (Zalta, 1999) for instances of such views. I think this is mistaken, but I won’t discuss the issue here. In the end it will not matter, since it will be hard to deny that some names in our language are non-referring and empty names in just this sense. Some names will be defective enough that they won’t manage to pick out anything at all, be it abstract or concrete. 2 See (Neale, 1990).

94 idealism and the harmony of thought and reality case the term is a quantifier.3 This contrasts with a picture of language that takes the semantic function of subject-predicate sentences to be more diverse, with only certain examples referring to objects and attributing properties to them, while other examples do very different things. In particular, if we reject the referential picture of language, then we will take the semantic function of terms to be more diverse than reference. Internalism is one view that rejects this referential picture of language. And there are good reasons to do so, which include the following. First, the referential picture of language naturally suggests itself if we take as our starting point first-order logic and aim to expand it to a semantics of natural language. This direction of analyzing natural language has some historical validity to it, but it is unjustified. There is no good reason to think that first-order logic is a good basis for the semantics of natural language. It was developed primarily to deal with certain simple valid inferences and the language of mathematics, not natural language in general, and has proven to be completely inadequate as a formal basis for modeling and understanding natural language. In standard first-order logic all terms are referential, and thus the referential picture of language applies to it. But whether it applies to natural language in general is a rather different issue altogether. Second, without looking at this issue through the lens of first-order logic, the referential picture must seem prima facie implausible. There are many examples that on the face of it do not look anything like names or referring expressions nor like quantifiers. That-clauses and other clauses are good examples of this. When I say that I wonder why she did it, then I do not seem to refer to something with ‘why she did it’, nor does it seem like a quantifier. Similarly, when I say that it is surprising that Sue ate the apple, then, prima facie, I am not referring to some entity and say of it that it is surprising, but rather I specify what it is that is surprising: I state or specify what is surprising, but don’t refer to it. Obviously, this prima facie implausibility of the referential picture can only go so far, but it should nonetheless not be ignored or put aside. Third, the referential view of that-clauses and fact-terms is in tension with the substitution argument briefly spelled out in section 3.2. To repeat, if that-clauses and fact-terms generally were referential, then co-referential ones should be substitutable for each other in ordinary cases without change of truth conditions. But just that does not seem to be the case, as a pair like: (23) a. John fears that his mother has read his diary. b. John fears the fact that his mother has read his diary. The former is a regular case of fear, while the latter is fact-phobia, fear of a fact itself. And those two statements thus seem to be different in truth conditions. 3 See (Neale, 1993) for an endorsement and defense, and (Hofweber and Pelletier, 2005) as well as chapter 8 of (Hofweber, 2016b) for criticism.

the internalist conception of facts 95 But if the that-clause in (23a) refers to a fact then it should refer to the same fact as the fact-term in (23b) and so they should be substitutable for each other.⁴ I take these reasons together to favor the non-referential picture so far. But the non-referential picture faces two obstacles: one serious and the other one less so. The serious one is tied to quantification, which is the topic of the next section, and which I will postpone until then. The less serious one is tied to the treatment of that-clauses in a compositional semantic theory. In a nutshell, the issue here is this: in a compositional semantic theory every phrase is assigned some semantic value such that the semantic values of the phrases in a sentence together with how they are combined into a sentence generate the truth-conditions of that sentence correctly. But that means that any successful semantic theory will have to assign some semantic values to that-clauses and fact-terms. And such semantic values are just the right candidates for what that-clauses refer to, and so they should be seen as referring to them, and thus referential expressions, or so the argument.⁵ But the close association of semantic values with reference made in the above argument is a mistake. For one, all phrases get semantic values in compositional semantic theories, even those that are quite clearly not referential, like ‘very’. Having a semantic value by itself is no reason to think that the relevant phrase is referential. Furthermore, even for referential phrases one should not identify the semantic value with the referent. This is made vivid by Richard Montague’s well-know treatment of proper names in (Montague, 1974). Montague assigned each proper name the set of all properties of the referent as its semantic value. Thus ‘Obama’ gets assigned the set of all properties of Obama. But ‘Obama’ does not refer to this set of properties, it refers to Obama, the person. Assigning a set of properties as the semantic value needs to be distinguished from referring to that semantic value, even if we assume that the phrase refers. Thus the treatment of that-clauses in a compositional semantics is largely neutral with regards to both whether such that-clauses are referential, as well as, assuming they are, to what kind of thing they refer to. Whatever semantic value they might get assigned in some compositional semantic theory by itself it does not speak for or against them being referential.⁶ The real issue for the non-referential picture of fact-terms is thus how to understand quantification over facts. Prima facie it is plausible that fact-terms are non-referential, but their interaction with quantifiers does give us good reason to think that this prima facie picture must be mistaken. To see if this is indeed so, we need to look more closely at quantification, which is the focus of this chapter.

⁴ The strength of these substitution arguments is obviously debatable, with, for example, (King, 2002) and (Schiffer, 2003) trying to defend the referential theory against them. Several philosophers have argued that these defenses fall short, see, for example, (Pryor, 2007), (Rosefeldt, 2008), (Hofweber, 2016b, §8.4), amongst others. For a mixed view, see (Moltmann, 2013). ⁵ For a version of this argument, together with an argument that that-clauses cannot be accommodated in compositional semantic theories, see (Schiffer, 1987). ⁶ This issue is discussed in much more detail in chapter 8 of (Hofweber, 2016b).

96 idealism and the harmony of thought and reality

4.2 Quantification Uncontroversially, quantifiers are sometimes, even often, used to make a claim about a domain of entities. When I say that: (24) Something fell on my head. I am making a claim about the domain of things, whatever there is or whatever exists, and claim that it contains at least one thing that fell on my head. Often we do not make a claim about the whole domain, but rather we restrict the domain contextually and we claim that something in that restriction has a certain feature. The question remains whether all uses of quantifiers are just like that. And here there is good reason to think that this is not so. One of the main reasons comes from the interaction of quantifiers with non-referential complements, be they singular terms or phrases of a completely different kind. Such non-referential complements are simply phrases that do not refer to anything, in either of our two senses of non-reference: de facto non-referential terms like the empty names ‘Betty Crocker’ and ‘Sherlock Holmes’, or constitutively non-referential ones like ‘very’. One example of a non-referential complement and how it interacts with quantifiers is this: (25) I need an assistant. This sentence implies that: (26) I need something. Thus ‘something’ interacts with ‘an assistant’ in such a way that this inference goes through. But ‘an assistant’ is not a referring phrase. In fact, the inference goes through even if there are no assistants at all, and thus if the domain is empty of assistants. So, it is hard to see how such a quantifier inference can be valid if it is understood along the lines of the common, domain focused, use of quantifiers.⁷ It is natural to hold that all this is exclusively due to the intensional transitive verb ‘need’ figuring in (25): a proper semantics of intensional transitive verbs will make clear why this quantifier inference is valid. This is a perfectly good and tried strategy for hoping to deal with such examples. But it isn’t clear how this would even address the real issue, since it doesn’t get around the fact that what I need is an assistant, there might not be an assistant, and nonetheless, there is something

⁷ See (Moltmann, 2003a). Moltmann’s own view is that ‘something’ belongs to a class of quantifiers she calls ‘special quantifiers’ that behave differently than other quantifiers. I will instead defend the view below that uniformly all quantifiers have two readings, with none of them more special than other ones.

the internalist conception of facts 97 I need, even if there aren’t in fact any assistants. I would like to suggest that there is a better strategy, one that does not depend on any special semantics for intensional transitive verbs, and is applicable generally to any examples of this kind. This better strategy is to hold that quantifiers have an inferential reading. And we can support this strategy simply by thinking about what functions quantifiers have for us in communication. Quantifiers have at least two separate functions in ordinary communication. Both are important, and neither one is derivative on the other. The first is perfectly exhibited by the domain conditions use of quantifiers. We represent the world as containing objects, and sometimes we need to make claims about the domain of objects: that some, or all, or most, etc., are a certain way. This use of quantifiers I have called the domain conditions reading or external reading above, in Chapter 3, and it is the one that is quite uncontroversial. But besides that, quantifiers are also used for the inferential relationships that they have with their instances. They also have an inferential reading or internal reading. This is just the case with (25) and (26). Here the quantifier is supposed to have the role of a placeholder, to occupy the position that can also be occupied with ‘an assistant’, but to contribute as little content or information as possible. And there is good reason why quantifiers have this reading, reason that comes from a general need in communication. The need for the inferential use of quantifiers can be nicely illustrated by the case of trying to communicate while being forgetful. I might not remember what it is that I need, but still remember enough that I need something or other. In that case I need to be able to communicate what I still remember, and I can do so by uttering (26). Here it does not matter if what I need is my coffee mug, or something completely different, like peace and quiet, or an assistant. Since ‘need’ is a transitive verb, I need to use some complement or other in the object position; I can’t just say ‘I need’. But that complement should be minimal in the sense that it contributes only what is necessary to make this claim at all, and thus in the sense that in contributes ‘minimal content’ or ‘minimal information’. In other words, no matter what it was originally that I needed, the function of the quantifier in this use should guarantee that (26) is true. And that is to say, the quantifier in (26) should be such that the inference from ‘I need X’ to (26) is always valid, no matter what ‘X’ might be. As long as ‘X’ is of the right syntactic kind to form a sentence in ‘I need X’, then the inference to (26) should be valid. Thus on its inferential reading, ‘something’ has the inferential role to go from ‘F(X)’ to ‘Something is F’, where ‘X’ is simply of the proper syntactic category. And mutatis mutandis for the universal quantifiers ‘everything’ and other quantifiers. Thus besides the familiar domain conditions reading, quantifiers also have an inferential reading. It is the same quantifier that has these two readings, and quantifiers are thus semantically underspecified: they can be specified or filled in to have either one of these two readings. Sometimes we use them one way, and sometimes the other, both of which are perfectly fine and meaningful.

98 idealism and the harmony of thought and reality This in outline is the idea behind why there is an inferential reading of quantifiers. I would like to now clarify several aspects of it: how to understand semantic underspecification, what motivates the underspecification view, how to understand the truth conditions of inferential quantifiers, and, maybe most importantly, how to formulate a context-sensitive version of inferential quantification. First, the claim is that quantifiers are semantically underspecified and have an inferential and a domain conditions reading. This is to be contrasted with a view that holds that quantifiers are ambiguous. Ambiguity occurs when two different words are spelled the same way, as in the case of ‘bank’. Semantic underspecification occurs when the semantic content of a word does not fully determine the content that it contributes to an utterance. In that case, there are different contributions that a phrase can make to the truth conditions of an utterance, but they are related in some way or other. Such underspecified expressions are polysemous, not ambiguous. In particular, quantifiers are claimed to be a case of semantic underspecification: there are at least two ways to fill in that content. The semantics of quantifiers, i.e. what contribution to content is made at the level of language, does not fully determine what the contribution to content and the truth conditions are of an utterance of quantifiers. This phenomenon of semantic underspecification is extremely widespread in natural language. Most words have a number of different readings and we easily, although somewhat mysteriously, can both fill in what is missing in an utterance and understand how it was filled in when hearing such an utterance. To consider just one example, take the verb ‘get’. It is used in several different readings in a sentence like (27) Before I get home I need to get some beer to get drunk. It is not that that verb is multiply ambiguous, having several unrelated meanings. Rather it is semantically underspecified, having several related readings, each of which makes a slightly different contribution to the truth conditions. Many verbs in English are underspecified in just such a way. It might be tempting to think that such underspecification belongs to the softer parts of language, verbs and adjectives, say, but not things that are more logical like quantifiers. But as is well known, paradigmatic logical expressions like ‘and’ are underspecified as well. Not only can ‘and’ combine different kinds of arguments (sentences, noun phrases, verb phrases, etc.), such combinations can be read at least collectively or distributively. The sentence (28) John and Mary danced. is semantically underspecified in that it can be read either collectively, that they danced together, or distributively, that they each danced. Those two readings have different truth conditions, and which one is the correct one is up to the

the internalist conception of facts 99 speaker of that sentence, but not the sentence itself. And similarly, I claim, are things with quantifiers: they can be read either inferentially or in their domain conditions reading. Quantified sentences are semantically underspecified, and what the proper truth conditions of an utterance of a quantified sentence is is not settled by the quantified sentence itself, but by the speaker of that sentence. But just as with an utterance of (27), speakers do not need to be explicitly aware that the relevant phrases are underspecified, nor how to specify them in an utterance. None of these aspects of language use are explicitly accessible to speakers by default, and it takes some training to notice them, even in simple cases like (27). Given how widespread semantic underspecification is in natural language, we should not at all be surprised to find it with quantifiers. On a similar note, the claim is not that there are two different quantifiers: say one for objects and one for facts. Instead the quantifier is the same in both cases. It is one single underspecified and polysemous quantifier that has different readings. All quantifiers have a domain conditions and an inferential reading, be they about objects or about facts. The question remains which of their readings they are commonly used in. And here there can be a systematic difference. It can be that when we quantify over facts, we commonly use quantifiers in one reading and when we quantify over objects, we commonly use them in a different reading. But even if so, both readings are always available no matter what we quantify over. Second, I would like to highlight how this view about quantification was motivated. At no point did I argue that this is the position to accept because it has certain metaphysical advantages, say because it is more nominalistic, or anything like it. The motivation simply came from a need we have in ordinary communication, not from something tied to philosophy or metaphysics.⁸ In fact, the underspecification view was not motivated with the use of examples like ‘something doesn’t exist’, which naturally could be affirmed with an inferential quantifier, but can’t coherently be affirmed with a domain conditions quantifier. Any such examples are always controversial and are too close to philosophical debates. Instead the motivation was simply from the need to have expressive devices available to communicate what we need to communicate in a language that has the complexity like ours. We can see that there are two distinct roles that quantifiers play: making a claim about the domain of things, and having inferential relationships to other sentences. We have a need for something to play each of those roles in ordinary communication. And although these two roles are different, it should be no surprise that one and the same underspecified expression can be specified in different ways to play each of these roles. The domain conditions

⁸ This contrasts with Einheuser’s claim that there are counter-conventional readings of subjunctive conditionals discussed in Chapter 2. Einheuser’s new readings of subjunctive conditionals are not tied to a need in ordinary communication, besides possibly to discuss conventionalism. The inferential reading of quantifiers instead has a role in ordinary communication, outside of philosophy.

100 idealism and the harmony of thought and reality reading and the inferential reading can be played by exactly the same fully specified expression in simplified circumstances: If every term in our language refers, every object is referred to with a term, and quantifiers only interact with terms, then inferential role and domain conditions exactly coincide. This is the case in standard first-order logic interpreted in term models, for example, but this isn’t the case with natural language interpreted in the world, where we have both non-referring terms, undenoted objects, and the interaction of quantifiers with other categories than terms, like ‘an assistant’. Third, it is important to clarify the truth-conditions of the inferential reading of quantifiers, as spelled out above in section 3.2. There I mentioned that several possible contributions to the truth-conditions could give the quantifier its inferential role for which we want it, but there is one simplest and optimal way: the strongest truth-conditions that give ‘something’ the inferential role F(t) implies F(something) is simply the disjunction over all the instances F(t), i.e. ⋁ F(t). Those are the strongest truth conditions, since they imply all other ones, for example all those that add more disjuncts. Similarly, the weakest truth-conditions that give the universal quantifiers the inferential role F(everything) implies F(t) is simply the conjunction over all the instances F(t), i.e. ⋀ F(t). It is reasonable to take those to be the truth-conditions of these quantifiers in their inferential reading, since they are optimal in just this sense. But this is not to say that sentences with inferential quantifiers in them are somehow infinitary or infinitely long. It merely means that the truth-conditions of inferential quantifiers are the same as those of some sentences that are in an infinitary language. In other words, we can specify those truth-conditions with infinitely long sentences but also in our ordinary natural language with utterances of simple finite sentences, namely ones that use the inferential quantifier. The inferential readings of quantifiers are not introduced into natural language with some infinitary machinery. Rather they are already there, and we can make certain aspects of them more explicit by using some infinitary machinery. But the proposal given so far has been too simple, not just because it was only an outline but also because it neglected one complexity that does really matter in the end: how to formulate inferential quantifiers that interact with context-sensitive expressions. This final point is important enough in general, and in particular also for our discussion later on, that it deserves its own section.

4.3 Context-sensitive quantifier inferences The inferential reading of ‘something’ inferentially relates quantified sentences to their instances. For example, ‘I need an assistant’ implies ‘I need something’, and it does so independently of what the semantic function of ‘an assistant’ is, in particular, whether or not it is referential. The simplest truth-conditions that give

the internalist conception of facts 101 the quantifier this inferential role is the disjunction of all the instances that are supposed to imply it: I need water or I need an assistant or I need a new car or I need peace of mind. . . . But this can’t be quite the full story, since it ignores contextsensitive expressions which can also interact with quantifiers. And that omission is significant, since such context-sensitive instances interact with quantifiers just as well as context-insensitive ones, but it is not clear how such context sensitivity can be accommodated in the proposed simplest account of the truth conditions of inferential quantifiers. To illustrate the problem, consider: (29) I need this. There are two context-sensitive expressions in (29): ‘I’ and ‘this’. ‘I’ is unproblematic for our purposes, as I will make clear below, but ‘this’ gives rise to a problem that seems to be serious, and it will be our focus now. The problem with (29) is that it, too, implies the quantified sentence (26). But it is hard to see how (29) can simply be an instance in the disjunction which specifies the simplest truth conditions of the inferential quantifiers. Those truth conditions are that I need money or I need an assistant or I need this or. . . . The problem is highlighted by the difference between ‘I’ and ‘this’ in how those truth conditions are spelled out above. When a speaker utters the quantified sentence (26), then they will have everything that is required to pick the referent of ‘I’, if any such thing is required: they intend to talk about themselves, about their own needs. But when they utter the quantified sentence, then they will have no referential intentions that could fix the referent of ‘this’. They might not be aware that the truth conditions of the quantified sentence somehow involve a demonstrative, after all. But without a demonstrative being backed up by such referential intentions it is hard to see how it could have a referent, and thus how the whole disjunction with the disjunct ‘I need this’ has determinate truth conditions. (29) can be used in essentially two ways: the first is where ‘this’ is a discourse anaphora, referring back to something that came up earlier in the conversation. Often one might prefer to use ‘that’ rather than ‘this’ in such a situation, but in any case, this case is not really a problem for the proposal. There would already be the relevant disjunct in the disjunction. If someone said earlier that they have a new lawn mower, and one replies with (29), then there is no problem with the truthconditions of the inferential quantifier as spelled out so far with the simplest truth conditions. After all, the disjunct ‘I need a new lawnmower’ is one of the instances that was already covered, and thus the inference from this utterance of (29) to (26) goes through on the original formulation without a problem and without any modifications being required. But the problem is the second way of using ‘this’: what if ‘this’ in (29) is used as a demonstrative, whereby the speaker intends to refer to a particular object? Then it seems that the inference might not always be valid. If the speaker refers

102 idealism and the harmony of thought and reality to something with ‘this’ that we also have a context-insensitive term for, then it might be that the inference still goes through. If this just refers to the Mona Lisa, then ‘I need the Mona Lisa’ will already be one of the disjuncts, and the inference is fine. And since terms are not just names, but include also definite descriptions, then there will be a large number of cases where the context-insensitive terms will have us covered.⁹ But we can’t assume that we always have such a term available. There is no guarantee that anything that we can refer to with a demonstrative is also something for which we have a context-insensitive term. After all, there seem to be more things than we have context-insensitive terms for. And if those come apart, then the semantics for the inferential quantifier seems to be mistaken. When I need this, then I need something, no matter what this is and whether we have a context-insensitive term for it. The inferential reading of the quantifier is supposed to make the inference from ‘I need X’ to (26) valid, no matter what ‘X’ is. But if ‘X’ can be context sensitive, then this thus seems to be false. To properly formulate the truth conditions of the inferential quantifier we thus need to accommodate context-sensitive instances with which the quantifiers can interact. And that is problematic for the view, since the speaker of the quantified sentence can’t be expected to recreate the referential intentions of the contextsensitive term. Thus having a context-sensitive term or a demonstrative in one of the disjuncts won’t do. The motivation for the inferential reading of the quantifier comes in part from communicating in ignorance, and thus we can’t assume knowledge about what it is that is needed. How then can the inferential reading be saved in light of context-sensitivity? If we at first focus just on contextual reference to objects, as in the above example, then there is a fairly straightforward solution to this problem. The key is to think about which object might in principle be available to be contributed to content in a context. For the truth conditions of the inferential reading of ‘something’ to be good enough we need to make sure that no matter which object might be contributed to the content by an utterance of a context-sensitive sentence, there is a corresponding instance in our disjunctive truth conditions of the quantified statement. And the best way to guarantee this is for any object whatsoever being accommodated. After all, the semantics of quantifiers should not pass judgment on which objects are available for demonstrative reference, and so from the point of view of what truth conditions the inferential reading of the quantifier should have, we should accommodate all objects. But how can we make sure that for any object there is a disjunct that involves that object in just the right way?

⁹ There is an important issue about whether a description denoting an object will lead to the same truth conditions as a name or demonstrative referring to that object. If not, then the range of cases already covered is small, if so, then it is larger. Either way, our main problem remains, so I will largely sideline this issue for now.

the internalist conception of facts 103 Let us consider this for our example (29): ‘I need this’, which implies (26): ‘I need something’. We can’t impose any restrictions on which object might have been referred to with ‘this’ in the original utterance. Thus we want the specification of the truth conditions to include besides the usual instances like ‘I need an assistant’ also for any object o a disjunct ‘I need o’. It might be tempting here to simply claim that the truth conditions for the inferential quantifiers should be spelled out in an artificial language that contains a name for each object, but we can do better. We can simply spell them out as follows: add a disjunct with a new free variable in it, like ‘I need x’, and bind that variable from the outside with a particular quantifier in its domain conditions reading. In other words, for our simple case, the truth conditions of (26) are ∃x ⋁ I need t[x]. Here ‘t[x]’ is any grammatically suitable instance in English with possibly our new variable ‘x’ in it. And this is just what we need. Now it doesn’t matter which object was referred to with ‘this’, the inference from (29) to (26) will be valid. The disjunction of the instances will now include ‘I need x’ and that ‘x’ is bound by a particular quantifier in its domain conditions reading. In particular, whatever ‘this’ might have referred to, it will be one of the values of ‘x’. This takes care of the case when there is only a single context-sensitive term in the original sentence. But things can, of course, also get more complicated. We might have started with ‘I need this more than that’, ‘I need this or that more than that other thing’, and so on. Thus to solve this problem generally, we should not just bind one single variable with a domain conditions reading of a particular quantifier, we need to bind as many variables as could possibly occur in a regular sentence. And since there is no upper finite bound on that, we need to bind countably infinitely many such variables, all at once. Thus for the general case, the truth conditions of ‘something is F’ in the inferential reading are equivalent to those of ‘∃x ̄ ⋁ F(t[x])’ ̄ , whereby ‘x’̄ is short for a countable infinite sequence of variables x1 , x2 , x3 , . . . , ∃ binds all of those variables, and t[x]̄ is a term that might contain some of the variables in the sequence x.̄ The quantifiers ∃ is used in its domain conditions reading here, ranging over whatever objects there might be. And it is an ‘infinitary quantifier’ that binds infinitely many variables at once, just as ⋁ is an infinitary disjunction symbol, disjoining infinitely many sentences at once. In both cases these infinitary logical expressions only affect countably many variables or sentences. These infinitary logical tools go beyond more familiar logical resources, but they are technically unproblematic, just less familiar.1⁰ This is how I think we should understand the truth conditions of inferential readings of quantifiers. So understood, they can accommodate contextual contributions to content via contextual reference, while still preserving the core

1⁰ In more technical terms, the truth conditions of the inferential reading of quantifiers are spelled out in a fragment of ℒ𝜔1 ,𝜔1 . See chapter 9 of (Hofweber, 2016b) for more on these truth conditions of natural language quantifiers, and (Keisler, 1971) for more on infinitary logic.

104 idealism and the harmony of thought and reality inferential role for which we want them. A universally quantified sentence ‘everything is F’ correspondingly has the truth conditions ‘∀x̄ ⋀ F(t[x])’ ̄ . We will only need to consider these two examples of quantifiers in this book and so I will only discuss them. I have discussed elsewhere how this account can be extended to a larger class of quantifiers, so-called generalized quantifiers, which include ‘most’, ‘many’, ‘four’, and so on.11 Of course, contextual reference is not the only contribution that context can make to the content of an utterance, but it is the most tricky one for the inferential reading. Other examples, like specifying the content of a semantically underspecified expression, are easier, since here we can simply form disjunctions using each of the fully specified contents. Since underspecification usually only involves a limited range of options, this can be done quite straightforwardly. But contextual reference is open-ended, there is no limit at first concerning what objects could be referred to in some context or other. This gives rise to a real problem for the inferential truth conditions. Those were at first simply language-internal: they related one sentence to others within a given language. But contextual reference threatens this picture: it threatens the picture of inferential readings as simply being language-internal. With the above account, that picture is restored while contextual reference is accommodated in the relevant inferences. All this naturally deserves more detailed discussion. I have defended this view of quantification in my last book, (Hofweber, 2016b), where I work out the argument in its favor in much more detail, contrast it with various alternative proposals, and develop the proposal to cover generalized quantifiers and context sensitivity. I won’t repeat all of that here, but I would like to invite those who would like to see more to consult that book. Unfortunately, a certain amount of overlap between this chapter and the material from the earlier book can’t be avoided to make this book self-contained. My apologies, and thanks, to all those who had already read the defense of the underspecification view of quantification in (Hofweber, 2016b) before reading this chapter. We do not need all the details of the underspecification view of quantification for our purposes here, since we will mostly focus on the internal reading of quantification over facts and propositions. But I would like to contrast the underspecification view defended here with some alternatives before moving on, including one alternative that I neglected in (Hofweber, 2016b). First, it is important to note that on their inferential reading quantifiers do not range over a different domain. To illustrate with an example, consider (30) Some things don’t exist, like Santa Claus.

11 See section 3.7 of (Hofweber, 2016b) for development of context-insensitive inferential generalized quantifiers, and section 9.5 for extending it to context-sensitive ones.

the internalist conception of facts 105 This sentence is true on its inferential reading, but the reason is not because the internal quantifier ranges over domain that contains non-existent objects. The inferential quantifier does not range over a domain at all. It merely inferentially relates sentences within one’s own language. This contrasts with a Meinongean understanding of the quantifier in (30). The Meinongean will hold that (30) is true, since the domain of quantification contains non-existent objects. Other uses of quantifiers are then taken to be restricted, in the sense of quantifier domain restriction, to just the existing things. The Meinongean thus takes the difference between the use of ‘some things’ in (30) and in ‘Some things fell out of my pocket’ to be one of being unrestricted or restricted to existing things. The internalist, on the other hand, takes the reading employed in normal utterances of (30) to not involve a domain at all, but rather to be the inferential one, which relates sentences inferentially to each other internal to the language. Thus internal quantifiers do not have a special domain associated with them. Second, on the internalist view, quantifiers over facts are not of a different order or of a different kind than quantifiers over regular objects. Similarly, the internalist view of quantification does not take quantification over facts or propositions to be over a different domain than quantification over regular objects. No matter what terms quantifiers interact with, fact-terms or singular terms for objects, there will uniformly be two readings of such quantifiers, as opposed to different kinds of quantifiers: one for objects, and one for facts or propositions. Such ‘higherorder’ views of quantification are technically perfectly coherent, but we have good reason to think that quantification in natural language is not like this. One of the best reasons is that we can easily quantify over both facts and things with the same quantifier, something that should not only be impossible, but even be ungrammatical, on the higher-order view. Consider: (31) There are two things that Sue finds disgusting: pizza with pineapple and that John got the promotion. On the internalist view, there is just one kind of quantifier, with no different kinds for different domains or orders, interacting with different kinds of terms. Instead the internalist holds that the one kind of quantifiers uniformly have two different readings. In each case there are always two ways a semantically underspecified sentence can be more fully specified, leading either to the inferential or the domain conditions reading.12

12 For a defense of the higher-order view, see, for example, (Rosefeldt, 2008) or (Williamson, 2013). See (Uzquiano, 2018) and (Trueman, 2021) for criticism of my own view from the angle of a higherorder quantification approach. I have responded to Uzquiano in particular in (Hofweber, 2018), where I spell out in more detail what reasons we have for thinking that the higher-order quantification approach is false for quantification in English, even though it is coherent in formal languages.

106 idealism and the harmony of thought and reality

4.4 Quantification and modality Quantifiers in their domain conditions reading range over a domain of objects, whereas quantifiers on their inferential reading don’t range over a domain, but inferentially relate to their instances within one’s own language. This difference is crucial in several ways, and one of them is how quantifiers interact with modality, a topic we will briefly consider in this section and which will be relevant in the next chapter. The interaction of quantifiers in their domain conditions reading with modality is reasonably well understood. On one way of looking at it, modality, which is connected to phrases like ‘it is possible that’ or ‘it has to be the case that’, is semantically modeled with the use of possible worlds. Such worlds have a domain of objects associated with them, which are simply the objects that exist at that world. When we quantify within the scope of a modal operator, as in (32) It is possible that something falls on my car. then we can evaluate this utterance by considering whether there is a possible world such that the domain of objects associated with that world has something in it that falls on my car. The quantifier in its domain conditions reading can thus be evaluated as ranging over different domains in different worlds, assuming different domains are associate with different worlds. Each world might have its own, possibly quite different, domain, and the truth of quantified statements in their domain conditions reading at such worlds depends on what this domain contains in these worlds. Subtleties aside, this is all pretty standard, and even though much more should be said about it, it is not the topic we need to make progress on now. What concerns us instead is how to understand the interaction of inferential quantifiers with modality. Here there is an important difference, since inferential quantifiers do not range over a domain of objects, and so the standard view about how domain conditions quantifiers interact with modality does not apply or carry over in an obvious way. Let’s consider our example of me needing something again, but add modality: (33) It is possible that I need something. Here ‘possible’ is supposed to be metaphysical modality, not epistemic modality: the issue is that it could be the case that I need something, not that for all I know I need something. How should we understand this interaction between modality and inferential quantifiers? There are essentially two main ways to approach it, which we can bring out by considering what the truth-conditions of (33) are on their proper contextsensitive formulation, namely:

the internalist conception of facts 107 (34) It is possible that ∃x̄ ⋁ I need t[x]. ̄ This gives rise to two questions: first, what does the domain conditions quantifier ‘∃x’̄ range over, and, second, what are the instances of ‘t[x]’ ̄ that figure in the big disjunction? As for the former, the answer should be clear: the domain of objects in the relevant world at which we are evaluating, just as we would normally evaluate domain conditions quantifiers that interact with modality, as outlined above. As for the latter, there are essentially two options: the straightforward option, which holds that the instances of ‘t[x]’ ̄ are simply the ones in our own language, and the world-relative one, where the instances are those in the language that we speak in that world in which we evaluate the inferentially quantified sentence. It is not hard to see that the straightforward solution is the correct one. First, when we evaluate the inferential quantifier at a different world, then we might not exist in that world at all. Some relevant possibilities are possibilities without us, and so we can’t rely on the instances that our language would have in that world. Second, even if we are in that world, and even if we speak a different language there, the issue is not about inferential relationships to sentences in that language. The issue still is the inferential relationship of the sentence we actually uttered to sentences in that very language in which we uttered it. We don’t need our sentence to inferentially relate to sentences in some other, possible language which we might possibly speak. We need our inferential relationships to be within our own language, even when we are taking about what could be the case. When we make an assertion of a sentence containing modality and an inferential reading quantifier, then it is us in our present situation that determines what was asserted and what the inferential quantifier inferentially relates to. It is not the case that we, so to speak, move to a different world and then evaluate the assertion in that world, inferentially relating the quantifier to sentences we would then use if we spoke a different language. The content of what was asserted is long settled before we go to other worlds and evaluate that content there. The straightforward way to understand the interaction of inferential quantifiers and modality thus takes the domain of the domain conditions quantifier to vary from world to world, but the instances to be the instances in our own, actual language. Thus other possibilities do and don’t interact with inferential quantifiers. They do interact with them, since the connected domain conditions quantifiers range over the domains of the worlds at which we evaluate a sentence. And they don’t interact with them, since the instances with which the inferential quantifiers relate to are simply the ones in our own language, not the ones in a different language what we would speak in a different world. This, at least in outline, seems to me to be the right thing to say about the interaction of inferential quantifiers with modality. Both the context sensitivity of inferential quantifiers and their interaction with modality will matter for us later on. But what we have seen so

108 idealism and the harmony of thought and reality far should be enough for our discussions there, and so it is time to wrap up and move on.

4.5 The internalist position Internalism about talk about facts is merely a position in the philosophy of language, concerning what we do when we talk about facts and how our own language works. It involves two connected claims: fact-terms are constitutively non-referential, and quantifiers over facts are used in their inferential reading. Fact-terms do not aim to refer to things which are facts, but rather they state or specify the facts in question. Stating or specifying facts does not have to involve reference to things which are facts. Quantifiers, on the internalist picture, are generally semantically underspecified. They always have at least two readings: a domain conditions reading and an inferential reading. The internalist position about fact talk maintains that in ordinary assertions of sentences that quantify over facts we use such quantifiers in their inferential reading. On this reading the quantified sentence inferentially relates to instances in one’s own language. Furthermore, the inferential reading of quantifiers over facts has the simplest and optimal truth conditions: it is truth conditionally equivalent, in the simplest cases, to a conjunction or disjunction of the instances to which it inferentially relates. The situation gets slightly more complicated when we try to accommodate contextsensitive instances, but we have seen how we can do so in the way outlined above: add new variables to the instances and bind all those from the outside with either a particular or a universal quantifier in its domain conditions reading. That in a nutshell is the internalist position about fact-talk. In this chapter I outlined my reasons for accepting the internalist picture. These reasons came simply from reflecting on the need we have for various aspects of fact-talk in ordinary discourse. In particular, the need to have an inferential reading of the quantifiers that can interact with non-referential complements came not from metaphysics, but from considerations about what we hope to do in certain ordinary situations of communication. This order of priority in the argument, a priority of considerations about language over metaphysical ones, is important for the overall argument for idealism. It isn’t the case that we start with some metaphysical picture, maybe a form of nominalism about facts as entities, and then argue for a special interpretation of language in light of one’s metaphysical persuasions. Instead I argued for the internalist picture from considerations about language alone. Metaphysical considerations did not motivate this picture, only considerations about language. In Chapter 3 I claimed that we can establish idealism from considerations about language alone. For this to come to pass it is important that both the argument for internalism as well as the argument that internalism implies idealism do not depend on some metaphysical assumptions

the internalist conception of facts 109 friendly to idealism. The argument for internalism given in this chapter satisfies this requirement. No metaphysical assumptions were made at all, and especially none related to idealism. The question whether the argument that idealism follows from internalism also satisfies this requirement will be taken up below. Internalism surely deserves a more detailed discussion, and as with any view about natural language, the issue is complex and involved. But nonetheless, I hope to have at least made the case above that internalism is a real candidate for being correct about what we do when we talk about facts. I believe that internalism is indeed correct, and I have argued for it in much more detail in (Hofweber, 2016b), where I develop many of the issues that we could only touch on here, and discuss several other related topics that I left aside for now. But I hope that even with only spending one chapter on internalism, we can see what the position comes down to and why it might well be true. It is fair enough to insist that more needs to be said about various arguments in this chapter, but my hope is that those who insist on this will consult (Hofweber, 2016b), and those who do not will have gotten enough of a sense about internalism from this chapter to be comfortable moving on to focusing more on the main topic of this book—idealism—and what significance internalism might have for it. Even though reasonable people can and do disagree about the strength of the considerations brought up in favor of internalism here, I predict that internalism is not the most controversial part of the position defended in this book. That honor presumably belongs to the argument that internalism implies idealism, and thus that a view merely about our own language implies a view about the metaphysical centrality of minds in reality. It is this position and that argument that we should now turn to in more detail.

5 Harmony and ineffability 5.1 From structural harmony to complete harmony? Any defense of idealism faces two immediate problems: first, how to even coherently formulate the idealist position, and second, to say what reason we could possibly have for idealism as formulated being correct. In Chapter 3 I outlined my proposed answer to both problems: first, we can formulate idealism using the notion of harmony, as the view that harmony obtains for the right reason, and second, we can conclude from internalism that harmony indeed obtains for the right reason. In the last chapter I have argued that internalism is indeed true. In this chapter we will more closely investigate the connection between harmony and idealism. Let me very briefly review the basic idea of the argument for idealism given in Chapter 3: I proposed that we should focus on a strong form of alethic idealism and defend that our human minds are metaphysically central to reality as the totality of facts. In particular, we should focus on the relationship between our conceptual representations of facts, on the one hand, and reality as the totality of facts, on the other hand. In Chapter 3 the emphasis was on the relationship between the structure of facts and the forms of our conceptual representations of those facts. Paradigmatically, subject-predicate representations represent facts with an object-property structure, and other forms correspond to other structures. The question remains whether this correspondence can be explained. Do our thoughts have these forms because the facts have those structures, which would be congenial to realism, or do the facts have the structure because our thoughts have the form, which would be congenial to idealism? The anti-idealist realist and the idealist should consequently expect a difference when it comes to whether all the structure that can be found among facts corresponds to some form or other of our representations. A realist should see no reason to expect this to be the case, since form follows structure, while the idealist should expect all the structure to match our forms, since structure follows form. The key difference thus can be stated as one about structural ineffability: are there any structurally ineffable facts? Here structurally ineffable facts are those that are ineffable for us, since representing them requires representations of a form that we do not have access to. This notion contrasts with content ineffable facts, which are facts that have a structure we can

Idealism and the Harmony of Thought and Reality. Thomas Hofweber, Oxford University Press. © Thomas Hofweber 2023. DOI: 10.1093/oso/9780198823636.003.0005

112 idealism and the harmony of thought and reality represent, but for some reason or other, we can’t fill in that structure with the appropriate representations of the components.1 If there are no structurally ineffable facts, then our minds and reality as the totality of facts are in structural harmony. If there are no ineffable facts at all, then they are in complete harmony. But harmony by itself is not enough to establish idealism, since harmony could obtain by accident: our minds might just happen to be lucky enough to have all the forms required to represent all the facts, or even lucky enough to represent all the facts completely. But if harmony obtains for a reason, then this might be good enough to formulate and defend idealism. In Chapter 3 I outlined the argument that structural harmony has to obtain, not by accident, but for a reason that supports idealism: harmony is guaranteed to obtain because internalism is true, and thus it is no lucky accident. So far the review of the basic strategy for the formulation and defense of idealism. But there are some real worries about this approach that need to be addressed. One pressing worry is that the argument for conceptual idealism from internalism seems to support not just structural harmony, but the much more radical complete harmony. After all, if quantifiers over facts are inferential quantifiers, and they thus generalize over the instances in our own language, then why doesn’t it follow that all facts are straightforwardly representable, and not just structurally representable. The outlined argument against structurally ineffable facts seems to generalize to an argument against ineffable facts full stop. But that seems to be going too far. For one, ineffable facts should be expected as long as which objects there are is independent of us. We can’t expect to be able to represent all these objects, and with it all the facts in which these objects figure. Furthermore, it is one thing to match all structures of the facts with our forms, but quite another to be able to fill in the content in each case. Shouldn’t we expect that some components of facts are simply beyond what we can represent? Complete harmony is much stronger and seems to be quite clearly false. But a case can be made that internalism implies it as well, putting pressure on internalism itself. To make progress on these and a few other issues is the goal of this chapter. And the best way to start is to think a little bit more about the relationship between ontological and alethic idealism. Conceptual idealism is a form of alethic idealism, not ontological idealism. It concerns our place in reality as the totality of facts, not in realty as the totality of things. We have reason to think that ontological idealism is false, and we saw some of these reasons in Chapter 2. But prima facie it seems problematic how alethic idealism could be true, while ontological idealism is false. After all, there is a close connection between the totality of things and the totality of facts. This connection includes that for every thing that exists, there is a fact that obtains, namely the fact that this thing exists, and the other way round. This seems to suggest that one needs

1 See page 65, where this distinction was introduced.

harmony and ineffability 113 to accept both ontological and alethic idealism, if one accepts any one of them. But if ontological idealism is false, then such a connection would support that alethic idealism is false with it. How then can we overcome the tension inherent in accepting alethic idealism while rejecting ontological idealism? At first this is fairly straightforward. Conceptual idealism so far was understood as a position about structural harmony: the structures of the facts must conform to the forms of our thoughts. In particular, structurally ineffable facts are ruled out: there aren’t, and can’t be, facts that have a structure that does not correspond to any of the forms of our thoughts. Thus our minds constrain reality as the totality of facts, since only facts with a structure that corresponds to a form of our thoughts can be part of reality. But all this is perfectly compatible with our minds not having any influence on reality as the totality of things. Any things whatsoever can exist and can figure in facts. All that is required is that these facts don’t have a structure that goes beyond the forms of our thoughts. As long as the facts have, say, an object-property structure, then there is no further requirement on which objects can figure in such facts. Our minds only constrain the structure of the facts, not the totality of things. Thus if structural harmony is all that is required for alethic idealism, then there is no tension between ontological realism and alethic idealism. So far the tension can be avoided. But a good case can be made that more is required, and that simply leaving things at structural harmony is not a stable place to be. There are several forces that push one all the way to complete harmony once one accepts structural harmony. These forces can either be seen as supporting a reductio of structural harmony, if one holds that complete harmony is clearly mistaken, or they can be seen as pushing conceptual idealism towards its natural and more insightful conclusion. The two key considerations here are the following: First, how can we make sense of structural harmony obtaining without complete harmony also obtaining? How could we constrain the structure of the facts without also constraining the facts more generally? If there are content ineffable facts which are structurally effable, as structural harmony permits, then facts in general are not completely tied to our language and our representations of them. But then, why do we constrain their structure at all, if not also their content? Second, it seems that the argument given for structural harmony should carry over straightforwardly to an argument for complete harmony. That argument was simply that assuming internalism is true, then quantification over facts is used in its inferential reading, which in turn, on its simplest truth conditions, generalizes over the instances. But that seems to suggest that facts are not merely structurally effable, but simply effable. We will look at these arguments more carefully shortly. But we can already note that if it were indeed so that structural harmony without complete harmony is unstable, then this would lead to a much more radical idealist position, maybe too radical to be true. Complete harmony is a much stronger claim than structural harmony. If structural harmony obtains, then there are no

114 idealism and the harmony of thought and reality facts that are structurally ineffable by us in our language. That would mean that the basic kinds of thoughts and the basic kinds of sentences we have available are good enough to represent all the facts, but it leaves open that many facts are not yet representable by us in our present language. There might be ways to fill in these structures with content that we cannot yet represent, and thus it allows for our expressive abilities to change over time. But if complete harmony is true, then there are no ineffable facts at all, and thus everything that can be truly said in principle can be said by us, using our present language. This goes noticeably further than there being no structurally ineffable facts, and it seems to go too far. There are several arguments against there being no ineffable facts at all, but maybe most pressing among them are the following two. First, it can’t be that every fact is expressible by us in our present language for simple technical reasons: our language has a finite vocabulary and finitely many ways to put it together. Thus overall there are only countably many sentences and thus it seems only countably many facts can be represented in our language. But there are uncountably many facts overall. To give just one example, there are uncountably many real numbers, and for each of them it is a fact that this number is a real number. Thus there are uncountable many facts, and so most of them must be ineffable in our language, contrary to complete harmony. In the same spirit, alethic idealism understood via complete harmony is not compatible with ontological realism. If ontological realism is true, then there can be who knows how many objects. But complete harmony only seems to allow for countably many facts, and thus only for countably many objects figuring in those facts. It thus seems that we can’t constrain the facts in the way suggested by complete harmony without also constraining the things. And we have good reason to think that we don’t constrain the things. Second, if there are no facts that are ineffable in present day English, then it means either that we got lucky to have just the right language to represent all the facts, or that facts themselves are somehow tied to English. But both of these should be expected to be false. English is not special among human languages, and contemporary English is not special in the history of the development of human languages. Instead we should expect that what can be represented in natural languages changes over time, hopefully for the better, with our languages developing and allowing us to represent more and more facts as time goes on. We have no reason to think that this growth of expressive power has reached its end, and it would be too miraculous to be believable that we just happen to be at the end of the expressive development of natural language. Neither of these two arguments against complete harmony have any bite against structural harmony. Structural harmony is perfectly compatible with there being more facts than representable facts, as long as all those facts have one of the allowed structures. And structural harmony is perfectly compatible with expressive change over time, as long as it concerns the content of what is expressed, but not the

harmony and ineffability 115 structure. The question thus is whether the argument given for structural harmony extends to complete harmony, and if so, whether that shows that the argument is mistaken somewhere, or that complete harmony is defensible after all. The key to making progress on this question is to look at the concept of the ineffable more closely.

5.2 The concept of the ineffable Although the concept of the ineffable is used in different ways by different people, we will use it in one particular way: A fact is ineffable just in case we cannot represent it conceptually. But even with this understanding of it, more needs to be said to make it clear. In fact, all of the parts of this specification of ‘ineffable’ are in need of further clarification. First, our issue here is the ineffability of facts, not the ineffability of objects. Sometimes the notion of the ineffable is used to apply to objects. Here, on one strict way of understanding it, an object is ineffable if nothing can be truly said about it. On a looser way of understanding it, an object is ineffable if we cannot comprehend its nature, or if we cannot say anything positive about it, or some similar restriction. God is sometimes claimed to be ineffable in one or another of these senses. But being able to say that there is an ineffable object in the strict sense arguably leads to a contradiction unless specified more carefully: if nothing can be truly said about an object, and thus by definition that object is ineffable, then something can be truly said about it after all, namely that it is ineffable. This association of the ineffable with paradoxes or contradictions has sometimes shed a bad light on this concept, or drawn the attention of those fond of contradictions.2 But in the way I use it here, no such paradoxes or contradictions arise. For our discussion, a fact is ineffable just in case our conceptual representations do not allow us to represent that fact. Here we need to distinguish two senses of representing a fact. The first, which we can call the aboutness sense, is simply representing something about a fact, as in ‘the fact that p is surprising’ or more indirectly ‘The most remarkable fact will never be known’. Here some fact is represented in the subject position of a sentence, which overall makes a claim about this fact. What represents the fact here is a fact-term. The second way to represent a fact, which we can call the mirroring sense, is with a declarative sentence or thought. In this second sense the fact that snow is white is represented with the sentence ‘snow is white’. This mirroring sense of representing a fact is the one relevant for our notion of the ineffable. What matters for us is representing a fact, not just representing something about a fact. An ineffable fact is thus not to be

2 See (Priest, 2002) for the latter.

116 idealism and the harmony of thought and reality understood in analogy to an ineffable object, as a fact about which nothing can be truly said. It is to be understood as a fact such that we have no sentence or thought available that represents this fact, in the mirroring sense of representing a fact. There is nothing contradictory about there being ineffable facts in this sense. It simply means that our representational resources are not enough to capture that fact. We can truly say about such a fact that it is ineffable, and thus truly represent something about this fact without being able to represent the fact. We can even have a term for such a fact, and still, no contradiction arises, as when one says that the basic laws of physics are ineffable. We can say something about these basic laws: that they govern the universe, that they are ineffable, etc. But we might not be able to represent these laws, in the mirroring sense of representing a fact. Second, our issue here concerns conceptual representation, not representation in general. Conceptual representation is a special kind of representation. It is representation in thought or language. Conceptual representations are distinct from other kinds of representations in that they can be true or false, whereas other kinds of representations are accurate or inaccurate, but not true or false. Not all conceptual representations are true or false, of course. Predicates or names by themselves are neither true nor false, only those conceptual representations that are put together in the right way to form a judgment or a declarative sentence. But only conceptual representations are true or false. Other non-conceptual representations like a photo or a diagram can be better or worse in various ways. They can be more or less accurate, more or less helpful, but not true or false. When it comes to conceptual representation of a given fact, a representation either gets it perfect right, and thus is true, or it misses completely and thus is false. Accuracy comes in degrees, conceptual representation of a fact is all or nothing. Representations in thought as well as in language are conceptual representations, but that does not mean that they are the same, nor that they can necessarily represent the same. When we wonder whether a fact can be represented conceptually, then this should be taken to ask whether it can be represented in either thought or language. As long as the fact can be represented in either one of these two ways of conceptually representing a fact it is not ineffable in our sense. Thus an ineffable fact is one that cannot be conceptually represented by us in either thought or language, even if we can represent some aspect of this fact to some degree of accuracy in non-conceptual ways, as we would with a diagram or picture.3 Third, it is important to note that the notion of the ineffable is not an epistemic notion. The issue at hand is not directly about what we cannot know, but instead about what we cannot represent. What we can’t represent we also can’t know, on

3 Not everyone uses the notion of the ineffable as concerning primarily ineffable facts. One alternative use of the notion of the ineffable focuses on ineffable knowledge, roughly knowledge which is not propositional and which cannot be explicitly articulated. See (Moore, 1997), (Moore, 2003), and (Jonas, 2016) for more on this sense.

harmony and ineffability 117 standard views, since knowing something requires at least a mental representation of what one knows. An ineffable fact is unknowable, but not the other way round. We will never know if the number of dinosaurs that ever lived was odd or even, but we can easily represent both options. Some unknowable facts are not ineffable, but all ineffable facts are unknowable, since we can’t even represent them. Our concern here is what we cannot represent, not what we cannot know. Fourth, who is the ‘we’ that is supposed to be unable to represent a fact if it is ineffable? Here the notion of the ineffable can be made precise in several ways, each of them leading to an interesting and legitimate notion of the ineffable so made precise. The question is which notion should we use for our purpose? Which one is the most interesting for us, given our main concern? The goal of this book is to make progress on the question about the place of humanity in reality, that is, what place human beings have in the world. Thus our main goal is to investigate humanity and the representational capacities of human beings. What matters for us here is therefore not primarily what is representable in this or that particular human language, but by human beings in general. This does not make the issue fully precise, of course, since it is unclear what human beings are and what they can be. How much of a change in our minds should we allow for someone still to be a human being? How much technical enhancement is allowed before we leave humanity behind and become post-human? This is both a hard and also not fully precise question, but we do not need to answer it now. We can simply take human beings to be the creatures that are pretty much like we are now, leaving open whether future descendants of us, possibly much enhanced with technology or genetic engineering, are still human beings or some post-human creatures instead. This will be good enough for what is to come. Even with this rough understanding of humanity, the contrast can be made between focusing on humanity vs. focusing on a particular human being, like oneself, or a particular group of human beings, like speakers of contemporary English. Fifth, how is ‘cannot’ to be understood in the claim that we cannot represent ineffable facts? When we ask what we can and cannot do, we usually keep certain things about ourselves the same, and allow others to vary. I cannot fly, keeping my physique the same and disallowing technological augmentations, but I can fly with extra wings or a plane. Similarly, when we ask what we can represent conceptually, what should we keep the same and what should we allow to vary? Here there is again no clear answer in itself. Instead, there are once more various ways to make the notion of the ineffable more precise, with each one at first being perfectly legitimate. The question again is, which notion of the ineffable do we want to employ in our discussion here? Our main goal is to determine the place of humanity in reality and how the human mind matches up with the world. Thus we should focus on a notion of the ineffable that fixes how the human mind is in general, and allows for variations in how human beings employ this mind. For example, human beings have a rather limited lifespan, and so any facts that require

118 idealism and the harmony of thought and reality representations of such complexity that more than one human life is needed are in a strict sense not representable by any human being. But that is not a limitation in a match between our minds and reality in the intended sense. It is simply a limitation in how we can use our minds given that we are embodied in these particular bodies. This limitation is thus too restrictive for our purposes. On the other hand, we might be able to represent anything at all if we had competently different minds, maybe a mind like God. This is too permissive for our purposes. Thus we should keep the basic setup of our minds the same, but be generous when it comes to time, memory, length of sentences, and the like. This way we are homing in on the right notion of the ineffable for our purposes. Other notions of the ineffable are equally legitimate, but not the right ones for us given our goal. Sixth, the issue should not be trivialized by making facts too fine-grained. On a fine-grained conception of facts, the fact that I am hungry might be different than the fact that TH is hungry, and the fact that tomorrow I’ll go shopping might be different than the fact that on Friday I go shopping, even though tomorrow is Friday. If we understand facts as fine-grained, then only I can represent the fact that I am hungry, and so it will be ineffable for everyone else. Overall then, a fact is ineffable if we human beings cannot represent it conceptually in either thought or language, even allowing for extra time and memory, while leaving our basic cognitive setup the same. If there are no ineffable facts so understood, then our human minds are a perfect match for reality as the totality of facts. If there are ineffable facts, then there is a mismatch between our minds and the world. Such an ineffable fact can be either structurally ineffable or content ineffable. The former obtains if representing a fact with that structure requires a form of a representation that is beyond those our minds have available. The latter obtains if we can represent facts with that form in general, but for some reason we can’t fill in the form in this case: either because we can’t represent a certain object, or we can’t represent a certain property, or we can’t represent some other component of the fact. If there are no structurally ineffable facts, and thus structural harmony obtains, then our minds are a pretty good match to reality: they have the right kinds of representations, and the right kind of representational setup, to represent reality, but they might fall short when it comes to filling in some of the contents. Structural harmony is good, but not quite as good as complete harmony, which is a perfect match, given our notion of the ineffable. If complete harmony obtains then any fact is representable by us, given the above idealizations. Harmony by itself is metaphysically neutral, since it can simply obtain by accident. But if there is the right kind of an explanation of why either structural or complete harmony obtains, then idealism follows. Structural harmony is more palatable, but it leads to a weaker form of idealism. Complete harmony is more radical, more interesting, and leads to a stronger form of idealism. The question remains which kind of harmony is supported by internalism, and this is the question we need to attend to next.

harmony and ineffability 119

5.3 Objects as a source of ineffability In light of this discussion of the concept of the ineffable, the gap between structural and complete harmony starts to disappear a bit. What is at issue is the relationship between our minds and reality as the totality of facts. In light of our discussion just above of what notion of an ineffable fact is the most relevant for us, we can see that certain de facto limitations we face are not relevant limitations for our main issue. Complicated facts that require a representation of great length, but are otherwise perfectly representable, should not be seen as ineffable in the sense of the term relevant for our main issue. Consider one example that brings home this point: the sand metric of planet earth. It is the fact of the precise distance of every grain of sand on earth to every other one. This fact is simply a complex conjunctive fact: grain 1 is x meters from grain 2 and grain 1 is y meters from grain 3, etc. No human being will ever be able to represent this fact in thought or language, and thus in a strict sense it is ineffable. But this strict sense of the ineffable is not the relevant one for our issue of the match between our minds and reality. Each of the conjuncts is easily representable by us, and the rest is just putting them all together. Our minds and the sand metric fact are not in a relevant kind of mismatch. Thus we need to home in on a different notion of the ineffable, one that does track the larger metaphysical issue, and thus one that idealizes to an extent. Simple complexity is not a proper source of ineffability in the sense relevant for us. But complexity might not be the only source of a limitation that we can represent. Another clear example of facts that no human being can de facto represent, but which are not ineffable in the sense relevant for us would be mortal facts: facts that are so depressing that any attempt to represent them conceptually would lead to the death of the person attempting to do so. I spare you the details of what such facts might be; after all, if there were any, I couldn’t represent them: they would be de facto unrepresentable by us as we are, even though they do not point at all to a mismatch between our minds and reality. This, too, is not a limitation of what we can in fact represent that is relevant for our main question. Similar considerations also apply to other representational limitations that we encounter, but that do not point to a proper mismatch between our minds and the world. In particular, they also apply to some limitations that we face when trying to fill in the content of structurally effable facts. For example, take an object that is simply too far away from us to be in any causal contact with human beings. Can we represent any facts that involve this object? It might well be that given how isolated this object is from us that we will never be able to represent such facts. One way this could be is if a causal theory of reference or singular thought were correct, which would require causal contact to refer to, or think about, an object. But even if so, this does not point to a mismatch between our minds and the world, only to an unfortunate placement of our minds within the world when it comes

120 idealism and the harmony of thought and reality to representing facts involving this faraway object. If we were closer to the object we could represent these facts without a problem. The issue is thus not a mismatch between our minds and the world in general, but a mismatch between where we are and where the object is. To get a clearer picture of what is a legitimate mismatch between our minds and the world, rather than merely an unfortunate placement of our minds in the world, it is helpful to turn things around. We should not focus so much on what we can de facto represent when faced with the world, but rather on what resistance our minds might put up when they are faced with various representations of facts. A good thought experiment to bring out the difference is to think about what would happen if some person tried to communicate a certain fact to us, and why they might be unable to do this. To illustrate, suppose we encounter a benevolent creature who speaks our language besides their own and who wants to communicate a certain fact to us. Suppose further that this creature, maybe an advanced alien life-form, knows this fact and can represent it in their own language and mind. What obstacles might such a creature face in trying to communicate this fact to us? Are there any facts where an attempt to communicate them is bound to fail, since our minds simply can’t think those kinds of thoughts and thus won’t be able to receive and absorb a representation of these facts? If so, then such a fact would be incommunicable to us. If a fact is incommunicable by a benevolent communicator, then it would be ineffable for us, and this is a case of ineffability that is relevant for us. The reason the attempt at communication would fail is due to our minds not being up to the challenge, not our unfortunate placement in the world or the like. The reason why this attempt at communication must fail is due to our minds and their limitations. Any fact that is incommunicable to us by a benevolent creature that can represent this fact would be ineffable for us in the proper sense, since the fact just doesn’t fit our minds, not because of an unfortunate relationship that our minds have to parts of the world. And this exactly tracks the sense of being ineffable for us that matters for our main discussion. I will call the incommunicability test for ineffability the test whether a fact could be communicated to us by a benevolent creature that can represent it. A fact should only seen as ineffable in the relevant sense if it doesn’t pass this test. If it passes, then the fact is either straightforwardly effable or it being de facto ineffable, in a strict sense of the ineffable, is due to our placement in the world or a similar reason, not due to a mismatch between our minds and this fact. Now, using the incommunicability test we can see that a far away object does not make a fact involving this object ineffable. A benevolent communicator who is able to represent that fact, and thus has a name for that object available, would be able to communicate it to us. They could simply allow us to piggyback on their use of the name for this object, and thus to allow us to refer to that object after all. This might require some setup on their part. They might first have to introduce the name to us, maybe say a bit about it, which would in turn allow us to piggyback on

harmony and ineffability 121 their use of the name to use it ourselves. The chain of reference would then go from us via them to the object. This process of relying on the use of names by others to be able to refer to an object with that name is very familiar. This is how we manage to refer to Socrates after all: with the help of others who used his name before us. And using the same mechanism of relying on others for reference, we will be able to refer to the object represented by the benevolent communicator. If we grant that any object that can be represented by a benevolent communicator could also be represented by us, in principle, and thus that such objects are not a proper source of ineffability, then this leaves open the question what to do with any objects that can’t be represented by such a benevolent communicator either. Maybe some objects can’t be represented in singular thought at all. Maybe there is something about them that makes it impossible for them to be the objects of reference, not just for us, but for any creature, actual or possible. Timothy Williamson calls such objects elusive objects in (Williamson, 2008, 16). Williamson imagines that maybe ordinary objects are constituted by such elusive objects: “Can we be sure that ordinary material objects do not consist of clouds of elusive subsub-atomic particles?” (Williamson, 2008, 17) I think we can be quite sure that this is not so. It is in general quite easy to have singular thoughts about objects. I can, for example, introduce a name for an object via a description and then employ that name in singular reference. So, if I name the largest volcano ‘Vulcan’, then I can think that Vulcan is a volcano, and thus have singular thoughts about Vulcan.⁴ This is also good enough for singular thought about really small things, like sub-sub atomic particles and material objects in general. If elusive objects are such material objects, and thus have spatial locations, then I can pick one of them out as, say, the one that is closest to a particular point 45 degrees to my left. Introducing a name for it via that description will allow me to have a singular thought about that allegedly elusive object. It thus seems that no material object can be elusive, since there can be a smallest, or closest, or hottest, etc., one, and such description of it is incompatible with it being elusive. If material objects can’t be elusive, maybe some abstract objects could be? But this is also rather problematic. First, it seems to be in conflict with the axiom of choice for (impure) set theory. The axiom of choice implies that every set can be well-ordered, including the set of elusive objects. But then, call the least element in that ordering Eli, thereby enabling singular thought ⁴ This, of course, is only singular thought in a minimalist sense: it is a representation of the fact that Vulcan is a volcano, but with the thinnest of concepts of Vulcan. I might have no other idea what Vulcan is besides how I introduced the term. Some philosophers hold that such minimalist singular thought is not good enough to be proper singular thought, and that some acquaintance condition must be met as well. But this is really an issue about what should count as ‘singular thought’, not a debate about what is required to represent the fact that Vulcan is a volcano. Our issue here concerns only the representation of facts coarsely understood. Names introduced via a description are certainly good enough for that. No richer integration into our mental life is required for our purposes. See (Jeshion, 2010) for more on the debate about the requirements for singular thought.

122 idealism and the harmony of thought and reality about Eli. Now, it is not clear if this is good enough, since that well-ordering might not be definable by me, even though it is there. Still, this is a common way to introduce names for objects in mathematics, and such names do seem to be in the business of reference, and in this case would refer to some alleged elusive object. Maybe there is a lack of determinacy here, but still, such names are nonetheless referential and sentences involving them can be true. Second, the issue is not what normal humans can refer to in ordinary circumstances, it is what any creature whatsoever, actual or possible, can refer to. Elusive objects must be impossible to refer to, no matter how things are. No creature whatsoever can be able to pick even one such object out, not even with a description. That means no creature can have the relevant concepts that would allow them to formulate such a description. It might seem reasonable at first to not rule out elusive objects, since why restrict the domain of possibility this way? But on the other hand, not ruling them out also restricts the domain of possibility: it rules out creatures with certain concepts as impossible. It rules out an all-knowing God. It rules out that it is possible that there is only one such object. And so on. Overall then, the case against elusive objects is strong, and I will therefore put them aside. The incommunicability test makes clear that objects are never a relevant source of ineffability. Any objects that we cannot de facto represent lead to a limitation that could easily be overcome if we were in a better position, say situated next to a benevolent communicator. Thus our minds are perfectly capable of representing facts involving any objects. Limitations to which objects we can de facto represent are thus similar to limitations in the complexity of the facts we can represent. They do not point to a mismatch between our minds and the world, but only to a mismatch in how our minds can execute their abilities in light of how they are embodied and how they are situated within the world. Our goal is to home in on the right notion of the ineffable such that the effability thesis stated with that notion—the claim that all facts are effable in the relevant sense—closely tracks our main issue of the question about the proper matchup of our minds to reality. Uncontroversially, on a strict notion of the ineffable, where we need to be de facto able to represent a fact, there are ineffable facts in that sense, since no human being will ever be able to represent the sand metric fact. But this notion is too strict for our purposes. I argued above that on a loose notion of the ineffable, that of being structural ineffable, internalism guarantees the corresponding effability thesis, namely the structural effability thesis. Although establishing the structural effability thesis is a substantial result, it is not the strongest notion of the ineffable, and with it it is not the strongest effability thesis, that we should consider. In light of the present discussion just above, we can see that the relevant notion of the ineffable is an object-permitting one: objects are never a proper source of ineffability. This adds to the fact that complexity is not a proper source of ineffability for our main issue. So, on the relevant notion of the ineffable, neither complexity nor objects are a

harmony and ineffability 123 proper source of ineffability. The question remains if there are further sources of ineffability, and thus whether there are ineffable facts even using the proper notion of the ineffable, which is object-permitting and tolerant of complexity. We can then formulate our target thesis as the complete effability thesis: all facts are effable, in the object-permitting and complexity-permitting sense of being effable. This is the thesis that we should aim to establish or refute. If objects are not a source of ineffability, then the next natural candidate to consider are properties. But in a sense the issue with properties is the same as the issue with propositions. A property might be ineffable, since it involves an ineffable object, as the property of being taller than x, whereby x is an ineffable object. This source of ineffability is illusionary on an object-permitting notion of the ineffable, just as it is for the case of propositions. Or properties could be ineffable for other reasons than involving ineffable objects. And here the issue will be quite parallel to the issue concerning ineffable propositions. The question will be, how we should understand quantification over properties. Is it internal, inferential quantification, or is it external, domain conditions quantification? It is not hard to motivate that whatever reasons one has for accepting internalism for talk about propositions will carry over to internalism for talk about properties, and similarly, that whatever reasons support externalism for one will support it for the other. Talk of properties and talk of facts and propositions will have to be understood along the same lines. I have argued for this parallel in more detail in chapter 8 of (Hofweber, 2016b), but we can leave it aside for now. We will return to the issue of properties in due course, but for now I would like to focus directly on how we can argue for the complete effability thesis as clarified. The object-permitting notion of the ineffable is the proper notion of the ineffable for our purposes, and with it, the complete effability thesis is the right formulation of the effability thesis for our purposes. I will thus from now on understand complete harmony such that it obtains just in case the complete effability thesis holds. All this is intended as a sharpening of our earlier discussion of this issue. And so understood, the thesis of complete harmony is a substantial further thesis, one that goes beyond structural harmony. Even if objects are never a proper source of ineffability, there are many other sources of ineffability that remain. One, of course, is structure and with it the possibility of structurally ineffable facts. But even if structural harmony obtains, there are many other ways in which complete harmony could fail: for each way to fill in the structure with content other than objects, the question remains whether we can indeed do this for every fact. Here the incommunicability test can be our guide towards which facts might be ineffable in the relevant sense. This test is closely tied to the issue whether our minds are in principle suitable to represent certain facts, and thus whether the true reason for ineffability is a limitation of our minds, or merely a misplacement of our bodies within the world. To make progress on this question we thus need to investigate whether complete harmony obtains.

124 idealism and the harmony of thought and reality

5.4 The argument for complete harmony In Chapter 3 I argued that structural harmony follows from internalism, and in Chapter 4 I argued in favor of internalism. The argument for harmony from internalism was simply that internalism applies to the very statement of structural harmony, since it involves quantification over facts. In particular, internalism is directly relevant for the truth conditions of: (35) Every fact is structurally effable. Assuming internalism, this sentence is true just in case the conjunction of all the instances is true. That conjunction, in turn, is true just in case each conjunct is true. All the conjuncts are of the form: (36) If it is a fact that p, then it is structurally effable that p. Each of the instances of this form simply results from replacing ‘p’ in (36) with a sentence of English. But then each of these instances is true, since no matter which sentences S one replaces ‘p’ with, the result is structurally effable. The structure of the fact that S does correspond to a form of our thoughts or sentences, since the very instance ‘S’ exhibits just the form that matches up with the structure of the fact that S. Thus (35) is true and structural harmony holds. So far the argument from Chapter 3. But with our discussion of the ineffable from this chapter and our more careful formulation of internalism from Chapter 4 we can now see that we can go further. We can not only argue that structural harmony holds, we can extend the argument towards complete harmony, and I will do so momentarily. In Chapter 3 we were not yet in a position to conclude that complete harmony holds for at least two reasons: first, even though any instance of ‘that p’ is structurally effable, it is not clear that it is also effable without some further clarification of the notion of the ineffable. An instance can be too long and complex for any human to actually be able to represent it in thought or language. But even for long and complex instances, we still have the right form of a representation available to represent a fact with this structure, and so long instances are not a problem for structural harmony, but they are a problem for complete harmony. However, in this chapter we saw that the proper notion of the ineffable, the one most apt for our discussion, is one that is permissive when it comes to complexity. Mere complexity is never a proper source of ineffability, in the sense of the notion relevant for our main discussion. Thus the complexity of instances should not be seen as an obstacle to moving from structural harmony to complete harmony. The second reason why we were unable to argue for complete harmony above is connected to the simple semantics for inferential quantifiers that was given in

harmony and ineffability 125 Chapter 3 and its lack of being able to accommodate context-sensitive instances. When we ask what can be represented by us, we need to distinguish what can be represented by us in the contexts that we are in fact in, and what can be represented by us in principle, using our present language, but allowing us access to contexts that we are not in fact in. The difference was illustrated above by considering a far away object, one we could easily refer to if we were closer, but possibly can’t refer to without help given how far away it is. This should not be seen as a relevant limitation for our main issue about the harmony of our minds with the world. Our unfortunate placement in the world is not a relevant limitation, although it might be a de facto a limitation. When it comes to structural harmony, context is irrelevant, since context does not affect the forms of our representations, only how contents are filled into these forms. But when it comes to the question of complete harmony, context becomes important, even crucial. When we wonder whether any fact whatsoever can be represented by us, the question is not whether any fact can be represented by us in a context-insensitive way, without demonstratives. Nor is the question whether we can represent every fact while being in the context we are in fact in. The question instead is whether any fact can be represented by us in principle, i.e. whether there is some context or other such that we could represent this fact in that context. Thus to consider complete harmony we need to consider a context-sensitive version of internalism. We have seen this version of internalism above, in Chapter 4, and thus it is time to see whether complete harmony does indeed follow from internalism as well. Suppose then, as I have argued in Chapter 4, that internalism is indeed correct. Complete harmony obtains just in case there are no ineffable facts, on the proper sense of ineffable worked out just above. That is to say, the following sentence is true: (37) Every fact is effable. Using the proper, context-sensitive version of the inferential reading of the quantifiers, (37) is truth conditionally equivalent to: (38) ∀x̄ ⋀ (if it is a fact that p[x], ̄ then it is effable that p[x]). ̄ This is true just in case for whatever objects ō might be the values of the variables x,̄ and whatever instance we might pick for ‘p[x]’ ̄ , then if it is a fact that p[o], ̄ then it is effable that p[o]. ̄ Suppose that we fix some objects ō and instance p[o], ̄ which is indeed a fact. Is it effable that p[o]? ̄ Since we allow for arbitrary complexity in the instances as effable, the complexity of ‘p’ is not a source of ineffability. And since ‘p’ is an instance of our own language, it is effable. The question remains whether we are able to also pick out all the relevant objects ō that figure in ‘p[o]’ ̄ . But as we saw above, the notion of ineffability relevant for our topic is object-permitting.

126 idealism and the harmony of thought and reality Objects are never a proper source of ineffability. And so every instance is effable, no matter what objects are the values of the relevant variables. And thus (37) is indeed true. Internalism does not only guarantee structural harmony, it guarantees complete harmony when properly formulated. The context-sensitive semantics of the inferential reading of the quantifiers is perfectly matched with the objectpermitting notion of effability to guarantee that every fact is effable. And this is just as well. It was puzzling how it could be that we constrain the structure of the facts, while not constraining the range of all the facts with it more directly. Now we can see that this worry is unfounded. We do constrain the totality of facts directly, in that every fact that can obtain must be effable by us in the object-permitting sense of effability. And via this constraining of the facts in general do we also constrain the structure of the facts. In a similar spirit, we can now see why alethic idealism is compatible with ontological realism not just in the case of tying alethic idealism to structural harmony but also in the case of tying it to complete harmony. Complete harmony is to be understood as involving an object-permitting notion of the ineffable, and so understood, there is no tension between us having nothing to do with which objects there are, and us being able to represent all facts in the object-permitting sense. Complete harmony is a more stable position than mere structural harmony without complete harmony. But complete harmony is also a much more radical position. It is now time to look at what is implied by complete harmony, both problematic and uplifting.

5.5 The consequences of complete harmony On page 114 we saw two arguments against complete harmony. One was that it faces a cardinality objection: there are more facts than there are representable facts. The argument was that since our language has a finite vocabulary and only finitely ways to combine it we can represent at most countably many facts. But there are uncountably many facts that obtain. For example, for each real number r it is a different fact that r is a real number. Thus there are uncountably many facts, since there are uncountably many real numbers. But we can now see that this argument is mistaken. It is true that only countably many facts can be represented in a context-insensitive way, since there are only countably many sentences that can be formed in our language. But this is not what matters for the question which facts are ineffable, since the same sentence can represent different facts in different contexts. When we wonder which facts are effable, what matters is which are representable in a context-sensitive way. In particular, what matters is which facts are ineffable in the object-permitting sense of the ineffable. And here we can see that the real number example is not a problem, since real numbers are objects and objects are not a source of ineffability in the relevant sense. In fact, the statement motivating the objection

harmony and ineffability 127 (39) For every real number r there is a fact that r is a real number. is true on the properly formulated internalist view. To write it semi-formally in our usual notion, it comes to: (40) ∀r (if r is a real number, then ∃x̄ ⋁ (that p[x]̄ is the fact that r is a real number)). And this is true, since one of the instances of ‘p[x]’ ̄ is ‘x is a real number’, and one of the values of ‘x’ is r. This kind of objection to complete harmony relies on there being lots of objects, real numbers in this case, which are taken to outnumber the representable facts. But with the proper notion of the ineffable as worked out above, objects by themselves can’t outnumber the effable facts, since effable facts are object-permitting.⁵ But not all objections to complete harmony can so easily be put aside. The deeper worry about complete harmony is tied to expressive completeness and the impossibility of cognitive leaps for humanity. To see this, consider that if complete harmony obtains, then every fact is expressible in principle by us, in our present language. That means that we can’t improve in what we can express or represent in principle: we can already represent every fact, and no one can do better than that, including our future selves. But this is rather counterintuitive. We know that other creatures can represent less than us, since their minds are simpler. A dog, for example, might well be able to represent various things about its environment, but it won’t be able to represent that there is no complete axiomatization of arithmetic. Its mind is too limited to represent this fact, even if we allow for extra time and memory. Still, a dog is doing better than even simpler creatures like ants. This suggest that there is a hierarchy of what can be represented by different creatures, starting with simple creatures like ants or even lower, moving up to dogs, and finally reaching humans. But why should we think that the hierarchy ends with us? True enough, among the creatures that presently roam on earth, we might well be the best when it comes to representing the world. But we might merely be a temporary high point. We can imagine that in the future we ourselves or other creatures will go much beyond that. And we can also imagine being visited by much more advanced extraterrestrial creatures who developed for millions of years longer than us, and who have vastly superior minds and technology. Don’t we have to accept that such aliens could be representationally better than us, and thus that they can say more things and represent more facts than we can? But if complete harmony obtains, then we can already represent all the facts, and thus no creature can represent more, no matter how many millions of years they had to practice. And similarly, if we can ⁵ This and similar arguments are discussed in more detail in (Hofweber, 2006).

128 idealism and the harmony of thought and reality represent all the facts, then we can represent all propositions and thus all contents. Thus nothing can be said by anyone that can’t be said by us already. This goes contrary to what the thought experiments with more advanced creatures would suggest, and thus goes against internalism and the complete effability thesis. So, either internalism is false after all, or it doesn’t imply complete harmony, or representationally more powerful creature are not possible after all, despite their initial plausibility. That there are no ineffable facts surely must seem completely implausible, but the issue is not that easily resolved. It is plausible that expressively vastly superior future generations of humans or aliens with much greater representational abilities are possible. And it is plausible that what can be expressed by various creatures comes in a scale, with us being reasonably far up on that scale, but not as far as one could be. This conception of an expressive scale allows for us to grow over time and to achieve truly novel expressive powers, possibly via some cognitive leap, some radical break from what we could do before, which takes us to new levels of what we can say or think. And possibly after several such leaps we might reach the level of our aliens, who can think about the world in radically novel and better ways than we can so far. All this is very plausible, but it being plausible is, of course, no guarantee that it is true. Whether there can be cognitive leaps of this kind, whether we are somewhere on a scale when it comes to what can be expressed, and whether we can already represent every fact there is in principle to represent are all real questions that are not to be settled via plausibility, but by more substantial considerations. Which considerations can resolve these questions is itself a tricky question, one also not answered simply by plausibility considerations. All this is closely tied to the question whether there is a completely independent totality of facts that is waiting to be represented. Here, too, it is plausible that there is, but in the end this is also a topic that needs to be settled not just by plausibility. A key dividing line on this issue is connected to whether internalism or externalism about talk about propositions and facts is true. Internalism takes quantification over facts to generalize over the language-internal instances, externalism takes it to range over a domain of entities. If facts form a domain of entities, in particular independent entities, then the scale conception of what can be represented, with us somewhere on the scale, seems to be the better candidate. But if internalism is true, then things look rather different. Which of those two, if any, is correct is not to be settled on prima facie plausibility, but via considerations about our own language. Internalism rules out aliens that can represent more facts than we can. Should we take the thought experiment of expressively vastly superior aliens itself as evidence that internalism is false? They certainly make internalism prima facie implausible. But what prima facie implausible consequences a view has that are quite remote from what the view is directly about is not a strong reason to reject the view. Implausibility alone is not enough for rejection, in particular in light of

harmony and ineffability 129 the fact that we have seen an argument above for internalism. What is implausible can nonetheless be correct, and what matters is the precise reasons for or against a view, not how well it chimes with what seems right to us beforehand. Obviously, there is a real issue about whether internalism is correct after all. I have argued, in Chapter 4 above and in (Hofweber, 2016b), that it is, and, of course, that could be mistaken. But the reasons why it is mistaken must in the end be tied to why it is mistaken to think that that-clauses and fact terms are non-referential, or why it is mistaken that quantifiers over facts are used in their inferential reading with the simplest truth conditions. It is one thing to uncover the mistake in the above argument for internalism, and quite another to merely point to the implausibility of its consequences. The reasons for internalism being mistaken can’t simply be that it would lead to unwanted metaphysical consequences or that we would be wrong about what seems initially plausible in certain thought experiments. There certainly is something correct in these thought experiments, even for an internalist. It is perfectly coherent that much more advanced aliens visit us. They can be more advanced than us along several dimension: better technology, a better, more just society, more elegant and efficient ways of representing the facts, and so on. But what is not coherent, assuming internalism, is that the aliens can represent facts that we can’t represent, not just de facto but also not in our looser sense. But assuming internalism, we can represent all the facts and the totality of facts is not completely independent of our minds. Still, we should have a closer look at how to think about the possibility of advanced representers and of the possibility of advances in what human beings can represent. Here there are several aspects of this issue to consider that promise to illuminate the positive proposal besides clarify the thought experiments. It is first important to note that if internalism is true for talk about facts and talk about propositions, then it will also be true for other talk connected to thatclauses: talk about contents, or meanings, or reasons, etc. There are two main reasons for this: First, there is a close, likely even conceptual, connection between propositions and meanings, or reasons and facts, or truths and facts, and so on. This connection essentially guarantees that larger philosophical views about one also carry over to the other. Examples of such connections are that it is a truth that p just in case it is a fact that p, or a speaker says that p just in case their utterance has the content that p, and so on. It is hard to see how it could be that internalism is true about talk about facts while externalism is true about talk about truths while these connections hold. Second, the reasons in favor of internalism about facts or propositions carry over to reasons for internalism about meanings, contents, and reasons. If we collectively name all the different things that we pick out with thatclauses the propositional, then we can say that internalism or externalism applies to the propositional more generally. Thus if internalism is true about facts, then by the above remark it is also true about the propositional in general, and thus also about contents, propositions, meanings, reasons, and so on.

130 idealism and the harmony of thought and reality In particular, it follows from internalism that whatever content some mental state might have, it is a content that is effable by us and thus can also be represented by us. Thus whatever contents the aliens can think or say, we can say those as well in principle. That is what internalism about the propositional implies. It does not imply that there are not other relevant differences between the aliens and us. The aliens can have much more advanced technology and much greater knowledge than we do. Internalism is silent on all of this. And the aliens can have a much more efficient language in that they can say things better in at least two ways: First, they might be able to say things in any context that we can only say in certain contexts. This phenomenon of ‘decontextualization’ is paradigmatically displayed when one introduces a name for an object that one before could only refer to with a demonstrative. So, when I say that I call this thing ‘Napoleon’, then afterwards I can say something in any context that I before could only say while facing Napoleon, the just named object. Second, the aliens might have simpler ways to express something that we can only express in complex ways. This ‘lexical addition’ is paradigmatically the result of introducing a new term via definition, as it often happens in mathematics and also elsewhere. Both are an improvement in expressive power, and they make things simpler, but they are not a relevant advance for our concern about what can be expressed in principle; that remains the same. The aliens can be much more advanced than us in these ways to improve what can be expressed, but if internalism is true, then they can’t be more advanced in principle when it comes to representing facts or having states with completely novel contents. If internalism is true, then for any utterance of sounds that the aliens might make, either they are meaningful assertions with contents that are assertible by us as well, at least in principle, or these sounds do not have contents at all, and thus are not meaningful, do not represent facts, and are not apt for being true. The sounds they produce might even have all kinds of causal effects on them, but they can’t represent more facts, or have more contents than our utterances can in principle have. Any speech that is meaningful at all must thus in principle be speech that can be made by us humans as well.⁶ Something analogous is also true for the expressive development of human languages. They can improve in various ways, but they can’t increase their inprinciple expressive strength and representational capacities. Since all facts and contents are already presentable and expressible, no improvement in this regard is possible. But they can improve in what can be expressed in a more contextinsensitive way or in a simpler way or with similar improvements. Nothing and no one can be better at representing facts in principle than us, but we can make sense of there being less expressive power in other representational systems. We can make sense, for example, that there might be simpler creatures ⁶ A similar conclusion is reached via a completely different argument by Donald Davidson in (Davidson, 1984).

harmony and ineffability 131 who have a simpler language and who represent less than what we can represent. We can represent all the facts, so no one can represent more, but that does not mean that everyone can represent all the facts. Not only is it perfectly coherent to hold that dogs can only represent some of the facts, it is also perfectly coherent to hold that other creatures can represent almost all of the facts, but they are missing some key component to make it all the way. Such creatures we might even imagine speak a simplified version of English, and they, too, talk about what they can represent and what they can’t represent. It might seem that such a linguistic community quickly leads to problems for the present view, via a consideration analogous to one discussed above in Chapter 3. In our example, these other creatures will have their own term ‘fact’ and quantifiers like ‘all facts’, which will presumably also be used in its inferential reading. But since they can’t represent everything we can represent, by assumption, when those creatures utter ‘Every fact is representable by us’ we should want to say that this is false, since we can express more. We might even be able to give an explicit example of something we can express that they can’t express. But it is important to note here that when the simpler creatures truly utter ‘we can represent all the facts’ then this does not mean that they can represent all the facts. ‘All the facts’ as uttered by them does not mean all the facts. In particular, it does not mean the same thing in their language as it does in our language. And although what they say is true, they do not truly say that they can represent all the facts. Their quantifier ‘all facts’ inferentially relates to the instances in their language, which are fewer than in our language. Simpler creatures are coherent, but more advanced representers are not, in the proper sense of the term: it can’t be that other creatures can represent facts that we can’t represent even in principle. What would be required for that is more contents or more facts than what we can represent. But that is ruled out. We get all the facts and contents already, and whatever these other creatures can do better, it is not tied to representing facts or expressing contents. All this might suggest that according to internalism we are only special, since it is us who are talking about who is special, and it is thus our quantifier ‘all facts’ that is the relevant one when we wonder who can represent all facts. There is definitely a kernel of truth to this, and we will closely investigate this issue in the following chapters. For now, though, let us continue to get clear about what internalism and complete harmony implies. Similar issues also arise when we consider a possible variation among human languages with regards to representational strength. Internalism first and foremost implies that every fact is representable in the language I presently speak or write in: English. But couldn’t it be that other human languages have different expressive strength? Might it not be that Japanese, say, allows for the representation of some facts that are not representable in English? Internalism again answers this question in the negative. There aren’t any facts not representable in English, and so whatever Japanese can do better than English, it isn’t representing more facts. This again might seem implausible, and one might wonder how such a substantial question

132 idealism and the harmony of thought and reality as whether human languages are all the same expressively could be answered on the grounds that I presented above in Chapter 4. Whether all human languages are expressively the same is indeed a substantial question, and it is not resolved by internalism.⁷ Internalism is compatible with that human languages are not all the same. It only implies that my present language is representationally complete, but it can be that other languages are not. But that there is this difference is implausible, as those who speak more than one language know. Although it is uncontroversially true that in different languages the same thing can be expressed in simpler or more complex ways, it is not clear that there are any contents expressible in any language that are simply beyond what can be said in another one in any way at all. To illustrate, the German word Fahrtwind is a single word that means wind blowing at one while being in a moving vehicle, whereby this wind is the result of the movement of the vehicle. So, if you are sitting in a parked convertible during a hurricane, that’s not Fahrtwind. Only the wind you feel while driving is Fahrtwind. And only if the car’s movement makes you feel that wind. There is no word like this in English, as far as I can tell, but what can be said with that word in German can also be said in English, although not as simply and elegantly. To motivate that there is an in principle difference in what can be said with different languages one would have to rely on a rather fine-grained notion of sameness of meaning. On such fine-grained notions it might well also be true that what can be said on Tuesdays differs from what can be said on Wednesdays, since only on one of them can one say that today is Wednesday. Different human languages can differ in all the ways that the same language can differ over time: how context insensitive a fact can be represented, how simple such a representation can be, and so on. But if internalism is true, then no language can represent more facts than we can represent in English. This could either be because all human languages agree on what can be said in them, or because they disagree, but whatever other languages can do differently, it is not representing more facts. Here advanced aliens and other earthly creatures are no different. Consider the advanced aliens once more: they visit earth with their incredible space ships, and they learn English, and communicate with us. But they also speak amongst each other in ways that we cannot understand: they produce sounds that causally affect a hearer, but we cannot find sentences in English that we can take as proper translations for those sounds. Furthermore, the aliens might well tell us in English that those sounds can’t be translated into English. All this is clearly coherent as a thought experiment, and so far philosophically neutral. But ⁷ See (von Fintel and Matthewson, 2008) for a discussion of the view that all human languages are semantically essentially the same. A broadly Chomskian approach to language, where language arises from a shared biological basis, is congenial to the view that all human languages are expressively the same. If language arises from our shared biology, then no wonder human languages are all essentially the same. See (Chomsky, 1976).

harmony and ineffability 133 the internalist and the externalist will see this situation rather differently. The externalist will hold that contents are not tied to our language and thus there is no obstacle to holding that the aliens can express more contents than we can. Similarly, reality as the totality of facts is not tied to our language, and so there is no obstacle to holding that reality goes beyond what we can represent. The internalist will hold instead that since these extra sounds produced by the aliens do not have content and that in producing them the aliens are not making assertions. They can have causal influence on the aliens, but they do not represent facts and they do not have meaning. Since the sounds do not have content, they are not conceptual representations of facts and thus we have no reason to think from this examples that reality as the totality of facts goes beyond what we can represent.⁸

5.6 Constraining logical space Complete harmony so far concerned simply the match between our thoughts and the facts. It thus is aimed at how reality is, but not at how reality could be or how it must be. Although complete harmony rules out that there are ineffable facts, it so far leaves open whether or not there could be any. Conceptual idealism holds that there is a constraint on what reality is like that comes from our minds: every fact must be representable. And this ‘must’ suggests that such ineffable facts are impossible. Otherwise it would again be a lucky accident that we happen to live in a world where there are no such facts, while in other worlds there might well be some. Unless there is some reason why our minds and the world as it is must match, while our minds and how the world could be do not match, this would lead to the worry that even if the effability thesis is true, and there are thus no ineffable facts, this was merely luck on our part, and does not warrant metaphysical conclusions about our place in the world. But if it has to be this way, then luck seems less likely, and a reason for the connection might be forthcoming. Internalism is supposed to be the reason that things have to be this way, and now is the time to look more closely why that is so. When we wonder whether there could be facts ineffable for us, there are really two separate questions to consider: one about us, and another one about the world. The former focuses on us and how impoverished our representational abilities might be. In this sense there clearly could be ineffable facts. We human beings could be simpler, and not have the ability to represent certain facts, even any facts with a certain structure, while the world remains in general the same. In that case there would be facts which we then, in that world, could not represent. But this is not the relevant sense in which we need to find out whether there could be facts

⁸ See (Hofweber, 2006) and (Hofweber, 2017a) for more on the issues discussed in this section.

134 idealism and the harmony of thought and reality ineffable for us. The relevant question instead is one about how the world might vary while we remain more or less the same. Thus for us the relevant question is whether it could be that the world is different enough so that we, as we are, would be unable to represent some of the facts that would then obtain. Even if the world in fact does not go beyond what we can represent, could the world be different enough so that we could then not represent certain facts? What we need to do is thus to evaluate the statement (41) It is possible that there are ineffable facts. in its intended reading, where we keep us largely the same, but let the world otherwise vary. What we need to see is whether (41) in this reading is true. We briefly discussed how the internal reading of quantifiers interacts with modality in section 4.4. The upshot was that the internal, inferential reading in its simplest, context-insensitive version does not vary from world to world, since there is no domain that varies across worlds from which the instances are drawn. Instead, the instances are language internal, that is, internal to the language in which the quantified modal statement occurs. Thus in this simple version, (41) is equivalent to (42) It is possible that ⋁ the fact that p is ineffable. which is false on the intended reading where we remain the same, since each of the disjuncts in the big disjunction is false, since a disjunct corresponds to an instance in our own language. And that long false disjunction must remain false as long as we keep what we can represent the same. But as already noted in section 4.4, the simple account is too simple and the proper, context-sensitive semantics of the inferential quantifier needs to be employed instead. On the proper semantics, the truth conditions of (41) are: (43) It is possible that ∃x̄ ⋁ the fact that p[x]̄ is ineffable. Since the quantifier ∃x̄ is taken in its domain conditions reading, it ranges over a domain of entities, and which ones those are might well vary from world to world, as we discussed above in section 4.4. But then it seems that which facts are effable can vary after all. If one world contains extra objects, ones that do not exist in our world, then the external quantifier can range over them in that world, and thus there will be values in those worlds that are not available in the actual world. And thus some facts that depend on these novel objects, say that novel object a is an object, are ineffable in this world, but not in the other one. And so it would seem that there can be facts that are ineffable for us as we presently are, since we do not have access to the objects which there could be, but in fact aren’t.

harmony and ineffability 135 The dependence on what can be represented by us on the availability of objects to be referred to was discussed in some detail above. It is clear that if some objects are not available to be referred to, then, in a strict sense, any fact involving that object cannot be represented by us, given this restriction on what we can refer to. But this, I argued above, relies on the wrong, and too strict notion of the ineffable. A restriction on what objects we can refer to does not motivate a mismatch between our minds and reality, only a limitation on our part with regards to what we can refer to. This is no different when it comes to objects far away or objects in other worlds. The relevant notion of the ineffable for our main issue of the relationship between our minds and reality as the totality of facts is an object-permitting one, and with such a notion it is not enough to motivate that there are ineffable facts by pointing to objects that we can’t refer to. Thus with an object-permitting notion of the ineffable, it is also not enough to motivate that there could be ineffable facts since there could be objects which in fact do not exist. We can now note in addition that the variable domain for the external quantifier is the only difference for the range of facts that can obtain. Although which objects there are varies, what propositions they can otherwise figure in does not vary, assuming internalism. Since the instances in the disjunction in (41) are the same inside and outside the scope of a modal operator, there is no other source of ineffability. And using the object-permitting notion of the ineffable, we can then see that (41) is thus false. Using a strict sense of the ineffable, it is true, assuming that there can be things which actually do not exist, but using the objectpermitting notion of the ineffable it is false. Since that is the relevant notion for our debate, we can conclude that not only are there no ineffable facts, there can’t be any. We can think of logical space as the space of what is possible in the widest sense. Each point in the space is a way things can be. Points are connected in various ways as they for example both might include the truth of the same proposition. If all propositions were independent of each other, then logical space could simply be seen as having a separate dimension for each proposition that could be true. But, of course, not all propositions are independent of each other, and so the structure of logical space is more complicated than that. Whatever it is more precisely, we can conclude from our discussion in this section that there is a connection between logical space and our human minds. This connection is analogous to the connection our minds have to reality as the totality of facts. Although we do not construct the totality of facts, since which facts obtain is not determined by us, we constrain the totality of facts, since every fact in that totality must conform to our representations. And the same holds for logical space: although we do not construct logical space, since what is and isn’t possible is not due to us, we constrain logical space, since every possibility must conform to our representations. In both cases, the totality of facts and logical space, this conformity to our representations is to be taken in the object-permitting way.

136 idealism and the harmony of thought and reality What objects there are and what objects there can be is independent of us. But what facts these objects can figure in is constrained by our minds. Although most philosophers who write on this, I think it is fair to say, hold that logical space is not tied to our minds or language at all, and thus is completely independent of us, there are exceptions. Agustín Rayo has recently argued in (Rayo, 2013) that logical space is constructed by us via our acceptance of ‘just is’ statements. Our acceptance that water just is H2 O in part structures and constructs logical space by ruling out that there is a region in which water is not H2 O. I have my doubts about Rayo’s particular proposal, which I have expressed in (Hofweber, 2014) and to which Rayo has replied in (Rayo, 2014), but leaving all that aside for now, I would like to contrast a proposal like Rayo’s with the proposal defended here. On the present proposal logical space is not constructed at all, and in general we are not involved in which propositions are possible and which ones are ruled out of logical space as impossible. This is analogous to that reality is not constructed and that we are not responsible for which propositions are true. Reality does not truth-depend on us, it only range-depends on us. Similarly for logical space. Logical space is not constructed by us, but it is constrained by us. We delimit what the range of propositions is that are a candidate for figuring in logical space, up to which objects might be involved in them, which is independent of us. This constrains logical space, but it does not determine what is or isn’t possible within these constraints. Constraint without construction applies to logical space as well as to reality. And the constraint is the same for both: our conceptual representations. So far then the defense of internalism and its consequences for harmony and idealism. Obviously, reasonable people can disagree on whether internalism is indeed true, and to settle this is a substantial issue in the philosophy of language. I believe, and have argued, that it is true, both here and elsewhere. But any disagreement about internalism can seem to be besides the point. How could it matter in the end what we do when we talk a certain way? Maybe this or that view is correct about our own speech, but the real issue is one in metaphysics, and metaphysics is about reality, not our speech. What our speech and language is actually like should thus not matter for metaphysics at all. And with that the truth or falsity of internalism should be irrelevant in the end. At best it could point to that we need to change our language to be more suitable for metaphysics. Or to put it differently: internalism might help settle the question we in fact asked, as when we asked about the connection between our minds and reality as all the facts. But internalism doesn’t help us settle whether this was the right question to ask. It doesn’t say anything about whether we used the most metaphysically suitable concepts, or the most metaphysically proper conception on how to understand reality, and with it the proper conception of what idealism should be understood to be. Maybe internalism has implications for the totality of facts, but that might

harmony and ineffability 137 simply point to that asking about our place among the totality of facts was our initial mistake. Maybe it is metaphysically more suitable to not ask about the totality of facts, but about something else, say the totality of facts* and our place in it. On page 83 of Chapter 3, where this issue first briefly appeared, I called the question we should ask when we intend to ask about our place in reality the proper question. Maybe internalism gives us an answer to the question we in fact asked. But that is not real metaphysical progress, since it might simply be taken to negatively reflect on the quality of the question we asked. But what question should we have asked instead? What is the proper question to ask here? And how can we find out what concepts are the most suitable to employ in metaphysical questions? These issues do not merely concern an objection to the view defended here, although they certainly do that. They also point to larger issues about the role of our own language in metaphysics, about what questions we should be asking, what concepts we should be employing in these questions, and how to tell that one is making progress. The defense of conceptual idealism can’t be complete without saying more about these issues, which surely are also of great interest on their own right. I will thus aim to make progress on these questions in the following chapters. Only once that has been done can the full picture tied to conceptual idealism properly come into view.

6 Inescapable concepts In Chapter 3 I argued that conceptual idealism follows from internalism, and thus that we can draw substantial metaphysical conclusions from considerations merely about our way of representing the world. This way we were able to answer the Big Question about our place in the world. But this general form of argument begs for a closer look at two issues: first, how it could possibly be that we can draw metaphysical conclusions from our own language, and second, even if we have reason to think that the question we asked was answered this way, the issue remains what reasons we have for thinking that we asked the right question in the first place. The first of these issues was briefly addressed above in Chapter 3, where I outlined an argument in section 3.3 that the language-metaphysics gap can be bridged. The second of these issues was the one that I ended on in the last chapter, where I concluded that we need to investigate what the proper question is, i.e. the question we should have asked all along. Although these two issues might seem quite separate, I will argue in this chapter that they are in fact closely related. To make the case for this I will first discuss in more detail how the language-metaphysics gap can be bridged and what follows from this for the overall methodology of metaphysics. Then we will look more closely at how we can determine what the proper question is, in this case and in metaphysics more generally, and how settling this goes together with how we can bridge the language-metaphysics gap. I hope to make the case by the end of this chapter that the right answers to both of these questions lead to a broadly neoKantian approach to metaphysics, in a sense to be spelled out below. To get there, let’s first look at what role considerations about our own representations of the world might play in metaphysics.

6.1 The language-metaphysics gap It is widely accepted that one cannot draw substantial metaphysical conclusions about the world from reflecting on our language or thought alone. Thinking about our language or thought only tells us how we represent the world, not how the world is. Metaphysics concerns reality, and thus how the world is, not how we represent it to be. Thinking about how we represent the world is interesting and revealing about us, but thinking about the world as a whole is another matter altogether. It seems clear that we cannot hope to find out anything positive about Idealism and the Harmony of Thought and Reality. Thomas Hofweber, Oxford University Press. © Thomas Hofweber 2023. DOI: 10.1093/oso/9780198823636.003.0006

140 idealism and the harmony of thought and reality what the world at large is like by reflecting on our own language and our own ways of representing the world. To be sure, considerations in the philosophy of language can still play some role in metaphysics, but it would have to be a corrective role rather than a constructive role: considerations in the philosophy of language can prevent us from certain errors, but they can’t make a direct positive contribution to finding out what reality is like. Thus thinking about our own language can be useful, but it will only be of secondary significance. Pointing to scope ambiguities, to differences in how descriptions and names are both about things, and so on, can help us not go down the wrong path, but it doesn’t give us a positive insight into what reality is like. This would be a far cry from anything like the ‘first philosophy’ approach that some had envisioned for the philosophy of language.1 Uncontroversially, considerations about our own language can have a corrective role in metaphysics and anywhere else. What is problematic is a constructive role, one where considerations about our own language lead to substantial positive conclusions. And just this seems to be what is claimed in the argument for idealism given in Chapter 3. This issue does not just affect this particular argument for idealism, it affects the general methodology of metaphysics. If considerations about our own language cannot play a constructive role, then the methodology of metaphysics must move away from language and towards something else. Maybe metaphysics has to move closer to the sciences instead, and read off metaphysical consequences from the largely empirical results of the sciences.2 Or maybe metaphysics has to make decisions on the basis of theoretical virtues alone, focusing on which metaphysical theories are simplest, most parsimonious, and so on.3 Or maybe metaphysics does not answer question of fact at all, but instead merely concerns the construction of models of what is believed to be the facts,⁴ or with the selection of useful concepts,⁵ or the repair of flawed concepts⁶, and so on. And just this is what seems to have happened in metaphysics over the last so many decades. Considerations about our language help, but the truly constructive contributions must come from somewhere else. Here this situation concerning the place of the philosophy of language in metaphysics is no different than what the place of epistemology in metaphysics is commonly taken to be. Epistemology is generally taken to only have a corrective role, but not a constructive one. The epistemologist can point out that these conclusions are badly justified, and those theories have little evidence as their support, but this by itself does not lead to a positive, constructive proposal about how things are in metaphysically relevant ways.⁷ Thinking about epistemology 1 See (Burge, 1992) for a historical discussion and many references. 2 See (Ladyman and Ross, 2007). 3 See (Sider, 2013). ⁴ See (Godfrey-Smith, 2006). ⁵ See (Carnap, 1956) and (Thomasson, 2016). ⁶ See (Scharp, 2018). ⁷ Naturally, there are exceptions. Arguments that, say, we wouldn’t be justified in our mathematical beliefs if numbers were abstract objects, and thus numbers are not abstract objects, since we are justified in our mathematical beliefs, can be seen as giving epistemology a constructive role. But such

inescapable concepts 141 and the philosophy of language focuses not on reality, but on the theorist, the metaphysician, who aims to describe reality and to construct well-supported theories about it. That is all well and good, but by itself finding out about the theorist is not a way to find out about what the theorist theorizes about: reality. In this chapter I would like to argue that all this is a mistake: both epistemology and the philosophy of language have a constructive role in metaphysics. And it is indeed possible to draw substantial metaphysical conclusions from reflecting merely on how we represent the world. These conclusions can be drawn in part by reflecting on how our language in fact is and in part by considering a special class of concepts: inescapable concepts. Combining largely empirical features about our representations together with that these representations are inescapable, in a sense to be explained below, allows us to draw conclusions about what reality is like. And this gives rise to a general strategy of finding out about reality by reflecting on our concepts, which naturally leads to an approach to metaphysics that is broadly Kantian in spirit. This approach maintains that a distinguished class of our concepts is a guide to reality, and that we can find out about general features of reality by reflecting on our representations alone, in particular representations tied to that distinguished class. To get to this position we should think a bit more about how we can overcome the language-metaphysics gap, and how to formulate the proper question about our place in the world. Metaphysics concerns reality and what it is like in various general ways. We human beings are part of reality, and thus we can be included in the concern of metaphysics. Our representations of reality are a part of us, and so they, too, could be of concern to metaphysics. But it is hard to see how any substantial results about reality as a whole can be derived from focusing on our representations of reality. Our representations by themselves are merely an attempt to represent reality. By themselves they do not tell us whether these representations are accurate or true, and so by themselves they are merely about our attempts at representing reality. Since there is no guarantee that these attempts are successful, it is hard to see how anything of metaphysical significance follows from our representations alone. We might vastly misrepresent reality after all. This leads to the widely accepted view that there is an unbridgeable gap between considerations about language alone and metaphysics: (44) The language-metaphysics gap: one cannot draw large-scale metaphysical conclusions from considerations about language alone. And although this formulation of the gap is in letter about language, the same, of course, applies to our mental representations and our concepts. No matter considerations could also be seen as having a corrective role, in pointing out that our mathematical beliefs might not be well justified.

142 idealism and the harmony of thought and reality how we represent reality, in language or in thought, the representations alone are not enough to allow us to draw metaphysical conclusions. The key here is “language alone” or “representations alone”. One can, of course, draw all kinds of conclusions from language together with facts about which sentences are true. That is not about language or representations alone, but about language and how language relates to reality. Language and how it relates to reality concerns not just us and our representations but also reality. Uncontroversially, from language plus truth one can draw substantial conclusions about reality. But it is also fairly uncontroversial that from language alone one cannot draw such conclusions. At best one can draw analytically true conclusions, or derive conceptual truths, or draw conclusions about a small part of reality: us human beings. But these are never substantial metaphysical conclusion in the intended sense, since metaphysics concerns synthetic hypotheses, and not just a small part of reality, but all of it. Thus apparently no real and substantial metaphysical question can be settled from considerations about language alone. Plausible as this might be, I would nonetheless like to develop the argument briefly given in Chapter 3 against it, and argue in more detail in the following that this is false. One can indeed derive substantial metaphysical conclusions simply from reflecting on our representations, without assuming anything about whether these representations are true. I will first give an example of an argument that answers a substantial metaphysical question from considerations about language alone, and then explain how and why such an argument is possible in light of what we have seen so far. Consider the question whether natural numbers exist. This is a question about reality, what it contains and what exists, not just a question about us. Nonetheless, I will argue, this question can be answered from considerations about language alone. The key to this is to reflect on the semantics of number words: ‘one’, ‘two’, ‘three’, etc. There is a real and difficult question about the semantics of number words, which is connected to an old puzzle about the occurrence of number words in natural language, one that goes back at least to Frege in (Frege, 1884). On the one hand they appear syntactically like adjectives or determines, as in: (45) Jupiter has four moons. On the other hand, they sometimes appear in a position usually occupied by a proper name, as in: (46) The number of moons of Jupiter is four. How can one and the same expression appear in both of those positions? There are a number of options that would explain this, and all of them are simply options about our language alone. One is that number words are ambiguous: there is one

inescapable concepts 143 that is a name, and another one, spelled and pronounced the same way, that is an adjective. Another option is that number words are essentially the same in both kinds of occurrences, but that there is some explanation or other of why they appear syntactically in a position that is contrary to what it would normally be expected to be in. So, maybe number words really are adjectives, but for some syntactic reason they appear to be like names in (46). Or maybe the really are names, but for some syntactic reason or other they appear like adjectives in (45). Suppose the first uniform account is correct: number words are adjectives or determiners, but they appear syntactically like names for some reason. In particular, they have the semantic function of adjectives even when they misleadingly appear to be names. Adjectives are not referring expressions, they do not have the semantic function of picking out some entity. Adjectives modify a noun, but they don’t refer. All this so far is simply considerations about language alone. Furthermore, suppose that this picture extends to number terms more generally. Even terms like ‘the number four’ are not referring expressions.⁸ They can occupy the same syntactic place as names or other referring expressions, but their semantic function is different. All this is still only about language alone, and about what we do when we use number words or number terms. Number words and number terms more generally are non-referential on this picture. It is important to again bring up the distinction between two kinds of non-referential expressions, as we did above on page 79 in Chapter 3 and on page 93 in Chapter 4: those that are de facto non-referential, and those that are constitutively non-referential. The second class can be seen to be non-referential from considerations about language alone, while the first can’t. A paradigmatic case of being de facto non-referential is an empty name. An empty name aims to refer, but fails, since the world does not cooperate and does not contain the object it aims to refer to. In this sense, being non-referential is not about language alone. It is about the relationship between language and the world. But being constitutively non-referential is about language alone. This is the sense in which an adjective is non-referential. It does something completely different than referring. It is nonreferential, since its semantic function is something other than reference. If the above option about how to understand the different occurrences of number words and number terms were correct, then number words would be non-referential in this second sense. They are not even in the business of reference as they do something different altogether semantically.

⁸ How could this be? One option is that ‘the number four’ is not a description, but ‘the number’ is apposited to the non-referring number word ‘four’, similar to how ‘the philosopher’ is apposited to ‘Aristotle’ in ‘the philosopher Aristotle’. If so, then assuming ‘four’ is not referential it is arguable that ‘the number four’ isn’t either. A detailed defense of the non-referential view of number terms, and of a matching philosophy of arithmetic, is given in (Hofweber, 2016b), chapters 2, 5, and 6. For a discussion of the two uniform ways to treat number words, see (Dummett, 1991).

144 idealism and the harmony of thought and reality Suppose then that by considerations about language alone we have determined that number words and number terms are constitutively non-referential. Then we can answer the question whether or not natural numbers exist. The argument for this is simply the following: consider all the existing objects and pick one of them at random. Could it be the number four? I am now assuming, for the sake of this argument, that number terms in English are constitutively non-referential. This, in particular, applies to my use of number terms just now. Thus when I ask whether this randomly chosen object could be the number four, I use ‘the number four’ non-referentially. It occupies the syntactic position that can also be occupied by referring expressions, but it isn’t one of them, by assumption. But then the sentence ‘this thing is the number four’ can’t be true. Here ‘this’ refers to the object we picked at random, but ‘the number four’ does not refer at all. So, this thing we picked out is not the number four. And since it was picked at random, nothing in the domain of existing things is the number four. The number four thus is not among the existing things and so it does not exist. And the same holds for all the other natural numbers. Thus natural numbers do not exist. We have answered the metaphysical question about the existence of natural numbers simply from considerations about our own representations alone. And this bridges the language-metaphysics gap. But how can an argument like this possibly work in light of what we have seen above, and in light of the motivation for the language-metaphysics gap? What explains that we could bridge the gap after all? The key to seeing how this can be is to focus on the question we asked. When we tried to find out about reality we asked a certain question: do natural numbers exist? This question is one about what reality is like, not one about our representations of reality. Nonetheless, we used our representations to ask this question. We used our concept of a natural number to ask whether there are natural numbers. Focusing on the question we ask and the concepts we employ in that question gives rise to the possibility that reflecting on our concepts allows us to see what the answer to that question as stated must be. And just this seems to be the case in this example. If our representations are as we assumed them to be in this example, then it seems we can answer the question we asked simply by reflecting on the representations we used in asking it. If number terms are non-referring, then the question whether there are natural numbers can be answered in the negative by thinking about our representations alone. That’s how it was possible to bridge the gap. The mistake in motivating that the gap is unbridgeable was to have too simple a picture of what is going on in inquiry. That picture involved two parts: a statement and the world. And here it seems that you can’t just look at the statement by itself to see if it fits the world, and thus is true. But a more proper picture of inquiry involves three parts: a question, a statement as the proposed answer, and the world. Language is closely tied to two of these: the question and the proposed answer. And although thinking about the answer alone can be hard to see as a proper guide to reality, thinking about the relationship between the question that was asked and the

inescapable concepts 145 proposed answer is easier: in some cases we can find out that the proposed answer is a match to the question asked simply by reflecting on the language employed in both. This is the case in our example of concluding that numbers do not exist from considerations about the non-referentiality of number words. It should be noted that even though we answered the question we asked by reflecting on our concepts, the answer is not a conceptual truth in the traditional sense of the term. That there are no natural numbers is not an analytic or conceptual truth, in the sense that simply being competent with the relevant concepts or words gives one some more or less immediate insight into the truth of the statement. Instead we were able to determine this truth through substantial linguistic, and thus broadly empirical, considerations about our own language. Concept possession alone does not push us towards that answer. But nonetheless, we could answer a question of fact about the world, one that is not about us, merely by reflecting on our concepts and our language. In this regard the terminology of ‘reflecting on our own concepts’ can be misleading. On a more strict sense, it could be taken to mean that those who have the concept are in a position just to think about the concept on its own, and directly figure out that it has a certain feature and that a certain thought in which that concept figures is true. This makes it seem like a more traditional conceptual truth. But on a broader sense of ‘reflecting’, one more closely associated with investigating or trying to figure something about the concept out, there is no presumption that this is closely tied to just having the concept. Instead, it can be tied to broadly scientific investigations into our representations, which might not easily be accessible at all to those who have the concepts. What we did above was tied to the broad, but not the narrow, sense of reflecting on our concepts. Obviously, much of this is terminological, but no matter what terminology one picks, there is a substantial difference between the truth of bachelors being unmarried and the truth of there being no natural numbers. The former is a conceptual truth, as I prefer to use the terminology, while the latter is not, although it can be seen to be a truth by investigating our own representations. Now, even if all this is correct, it might seem like a shallow victory for those who hope to find out about reality from considerations about our representations of reality. The worry is analogous to the worry that we did not answer the proper question from Chapter 5, and overcoming this worry is the key to showing that we indeed made real progress in metaphysics. To do so is the main goal of this chapter. Let us consider this worry in a bit more detail for our case, before discussing how to resolve it. We were able to conclude from the argument given above that natural numbers do not exist, whereby natural numbers are simply the number one, the number two, and so on. If number terms are indeed non-referring, then maybe this conclusion can be drawn, but it might seem not much of an insight into what reality is like. Sure, it does not contain natural numbers, under those assumptions, but that tells us little about what it does contain. It might still contain number-like

146 idealism and the harmony of thought and reality things, say, von Neumann ordinals, positions in structures, lots of abstract objects, and so on. These things wouldn’t be numbers, but they would be a lot like what numbers were thought to be by many. Whether there are those number-like things is completely left open even if there are no numbers. And the questions about the existence of various number-like things are perfectly legitimate questions. In fact, maybe they are the better questions to ask than the one about numbers. Maybe the real issue, the one we should be pursuing, is not about numbers, but about those number-like things. Furthermore, one could take the non-referentiality of number terms to point to a flaw in the very question we did ask. If number terms are constitutively non-referential, but nonetheless appear in singular term position and thus in the position often occupied by referring expressions, then this can be seen as a confusing flaw of our natural language. A proper language would not contain such confusing constructions as our natural language does. A language more suitable for science and for metaphysics would avoid such apparently imperfect talk of numbers.⁹ That we were able to answer a question about numbers simply from reflecting on our own language thus can be seen to point more to a flaw in our language than to progress in metaphysics. Nonetheless, we answered the question we asked at the outset: do natural numbers exist? That counts for something, and in fact it counts for quite a bit, since this question has been asked for a long time, and it would indeed be answered by the argument given. But nonetheless, the issue remains whether we asked the right question in the first place. Did we probe reality in the right way when we asked about the existence of numbers, or should we better have asked about something else instead? Maybe we should not have used our concept of a natural number in the question, but some other, more metaphysically suitable concept. It is one thing to answer the question we asked, and quite another thing to ask the right question. The issue thus gets pushed to this: what is the question we should have asked in the first place? What concept should we have used in the question, if not the concept of a natural number? This leads to a second aspect of reflecting on our concepts and what metaphysical conclusions we can draw from them. It concerns not what we can conclude from the concepts we do in fact employ, but what concepts we should employ instead. To properly answer a metaphysical question we should not only find the answer to the question we did ask, but the answer to the question we should have asked. And to determine that, we need to find out what concepts we should employ in the questions we ask, and then answer those questions. Thus simply having answered the questions we asked is not quite enough to declare victory. We need to find out in addition that we asked a question suitable for metaphysics, and not ⁹ Essentially this sentiment is expressed by Frege in (Frege, 1884, §57) in light of various features of talk of number is natural language.

inescapable concepts 147 simply a flawed or defective question. Let us call a (metaphysically) deep result an answer to a question that is not only the correct answer to that question but is also the answer to a proper question to ask in metaphysics: the right question to ask, relative to nearby alternatives. Let us call a (metaphysically) shallow result an answer to a question which simply answers the question as stated, with no evaluation of the question itself, including whether it was a proper question to ask. What we need are deep results, not just shallow ones. And to get those it is not enough to simply bridge the language-metaphysics gap; we also need to find out what the proper concepts are that we should use in the question. A deep result does not have to be an answer to the uniquely best question one could ask. There might be several equally good questions to ask in a particular neighborhood of questions, all of which suitable for metaphysics. But a deep result can’t just be the answer to a flawed question, one that involves a defective concept. But how could we possible determine that we have achieved a deep result, without first finding out what reality is like and thus what concepts are properly suited to describe reality? It seems that reflecting on our concepts alone certainly won’t give us any guidance on which concepts we should have and thus which questions we should ask. But we can make progress here as well, in a way that turns out to be congenial to the project of making progress in metaphysics by reflecting on our own concepts alone.

6.2 The idea of inescapable concepts It is hard to know what concepts we should use in inquiry without knowing what reality is like. Ideally our concepts should be perfectly matched to reality, and so we face a circle: we need to find out what reality is like to determine what concepts we should use, but to find out what reality is like we already need to represent it with matching concepts. It thus seems that we can’t find out what reality is like unless we first find out what concepts we should employ, and also that we can’t find out what concepts we should employ without first finding out what reality is like. We can call this the conceptual circle and it is this circle that we must break free of. In ordinary cases of inquiry the hope is that we can slowly break free of the conceptual circle in a holistic way, by making simultaneous progress on both parts: what reality is like and what concepts we should use. In a step-by-step way, we might get slightly better concepts allowing us to be slightly better at representing reality, which in turn should lead to even better concepts, and so on. Whether this method for progress in the case of inquiry is sufficient is one thing, but it certainly would be insufficient for our project of finding out what reality is like by reflecting on our concepts alone. Here we need somehow find assurances that our concepts are proper ones to employ by reflecting on these concepts alone, without first finding out what reality is like. And it is hard to see how that could be so,

148 idealism and the harmony of thought and reality and with it it is hard to see how the project of finding a proper question to ask by reflecting on our concepts alone. But maybe there is another way out of the conceptual circle. Maybe we can conclude for certain special concepts that they are the right ones to apply to reality, even though we otherwise do not yet know what reality is like. Maybe there is a reason to think that at least a certain concept C is the right one to employ in describing reality and that we can appreciate this reason without knowing much else about reality. This might seem outlandish, but here is one way, in outline, how this might be: Maybe we can appreciate that any attempt to replace that concept C with a different one would always make things worse. Maybe we can recognize that replacing C would make things worse for us representationally, and thus that we should never do it. And if this were so for a particular concept C, then we might have good reason to think that C is just the right concept to employ when describing reality without first having to find out what reality in general is like. When I consider whether a particular concept C should be replaced with an alternative C*, then I consider the reasons for and against such a replacement. Should I employ concept C* where so far I employed concept C? Should I use C* instead of C in situations like the ones where I so far used C? When I consider such questions, I try to assess whether the switch would be a good one to make. After all, I have to decide whether to make the switch, and so I have to consider and evaluate the reasons for and against the switch. I must assess, by my own lights, whether the switch should be made. I am the one who has to weigh the reasons for and against making the switch, to the best of my present abilities. Thus this decision has to be made by me, in my present situation. And in such a case it could in principle be that I can recognize that no matter what concept C* might be, it would never be rational by my own lights to switch from C to C*. In my own present situation I could then never rationally move from C to C*, since it would not be rational by my own lights to do so, no matter what C* might be. And thus I will have to conclude that there is no acceptable improvement on C, and thus that C is perfect when it comes to describing reality: no other concept can do better. This route of establishing that a concept is ideally suited for inquiry, and how it can be combined with the idea of drawing metaphysical conclusions from thinking about our language alone, is the main target of the present chapter. It is a possible route towards establishing that we asked a proper question already when we asked about the existence of numbers or about our place in reality as the totality of facts. If the key concepts in these questions turn out to be impossible to improve upon, then it can be argued that there is no better question to ask that should replace the one we already asked. If so, then we would have asked the right question already, and we can see that this is so without first knowing what reality otherwise is like. Our method of answering these questions via considerations about our own language would then indeed lead to a deep metaphysical result, and would properly answer the proper question, or so the idea of this chapter in outline.

inescapable concepts 149 Let us call a concept inescapable if one cannot rationally, by one’s own lights, replace it with another one for the purpose of inquiry. That is to say, when I reflect on whether I should give up employing a particular concept C in inquiry, and rather employ a different one, C*, instead, then I can see that this would always be irrational for me to do, by my own lights. If this were so, then I would be rationally required to reject such a replacement and thus to stick with the original concept C. This notion is in need of several clarifications, and needs to be spelled out in more detail. I hope to at least begin to do so now, before we can move on to how it can be employed to make progress on our main topic. First, we should distinguish inescapable concepts from what we can call hardwired concepts. The latter are ones that are simply fixed in our minds, and can’t be replaced at all, no matter what methods we might employ. Whether there are any hardwired concepts in our minds, and thus whether whether every concept can be replaced using such methods as brain manipulation, hypnosis, non-rational persuasion, or what have you, is a question about the causal limitations of these and other methods to change our minds. Our issue here is not whether some concepts can’t be replaced at all, but only whether such replacement would ever be rational. The issue for us does not concern the causal limits of changing concepts, but the rational limits for doing so. In particular, even if every concept can be replaced by directly interfering with our brains, the question remains whether we ourselves could always find such a replacement rational by our own lights. This latter question is the question that matters for us here. We are simply concerned with what the rational limits of concept replacement are. Second, the issue is whether a replacement would be rational for the purpose of inquiry, that is, the purpose of representing reality. It might make a lot of sense to replace a concept in exchange for money, say. But that would be giving up on a concept for the wrong kind of reason. The right kind of reason for replacing a concept is instead a reason of the kind that this concept itself is somehow better, on its own, not because it comes with a good amount of money attached. How to make this distinction between the right and wrong kind of reason more precise is notoriously difficult, and I use ‘for the purpose of inquiry’ to make clear that it is supposed to be rational weighing the right kinds of reasons only, which concern how well the concept does in representing reality and figuring in inquiry, not necessarily how well it does in benefiting us in other ways. Third, the issue is whether the replacement would be rational by my own lights: whether I can appreciate, reflecting on and evaluating to the best of my abilities using my best methods, the reasons for or against that C* would be better to employ for the purpose of inquiry than C. Our question is whether we ourselves should replace one of our concepts with another one. To assess that properly, we need to be able to be in a position to appreciate the reasons for and against doing this, and so we need to evaluate those reasons by our own lights. It is not enough for this to rely on a purely external conception of rationality, where how reasons are to be weighed

150 idealism and the harmony of thought and reality is determined by something external, which is not necessarily accessible to us. So, if there is a sense of being rational where it is rational to have the concepts God prefers us to have, without it being accessible to us why those are the right concepts to have, then it would not be rational by our own lights to switch to God’s favorite concepts. Instead we need to determine for ourselves what we should do, and so we should weigh the reasons for or against the switch according to our own best standards, methods, and abilities for doing so. Thus the switch should be rational by our own lights.1⁰ Fourth, the issue is one about replacement of a concept in application, not replacement of a concept in one’s conceptual repertoire. To replace a concept in one’s repertoire involves getting rid of one concept while adding another. But it can be argued that it would always be irrational for me to get rid of a concept. Instead, one should always keep every concept one has around. Getting rid of a concept would irrationally limit one’s representational capacities. That might well be so, but our issue is not getting rid of a concept altogether, but rather whether one should employ concept C or concept C* instead. If one does the latter one can keep concept C around as a backup, but one would not use it any longer in the same situations: one would employ C* in the situations where one otherwise would have employed C. One’s representational capacities are not diminished by this, but they are exercised in a different way.11 Fifth, I would like to also contrast inescapable concepts with what we can call inevitable concepts. A concept is inevitable if every thinker must have it. Thus it is not merely a concept hardwired in this or that mind, something one can’t get out of a particular mind. An inevitable concept is one that must be present in every thinker in order for them to have thoughts at all. Whether there are any inevitable concepts is doubtful, but maybe the nature of thought itself requires certain concepts. But on the face of it, it does seem coherent to lose any particular concept, and thus lose the ability to have certain thoughts, but still have other thoughts, those that do not involve that concept. It seems excessive to think that if one were to lose that particular concepts, through a stroke or injury, say, that then one would lose all ability for thought. Still, one could argue that certain concepts are required for having thoughts at all: maybe the concept of oneself, one’s self

1⁰ What is rational by my own lights should not be confused with what seems to me to be rational. Being rational by one’s own lights is, as a first rough approximation, what is supported by one’s own deepest principles and methods. One can certainly be mistaken about what such principles and methods support, and thus one can mistakenly take something to be rational that isn’t rational by one’s own lights. The notion of being rational by one’s own lights and its significance is discussed in more detail in Chapter 7. 11 The notion of an inescapable concept is related to, but different from, those of a ‘conceptual fixed point’ in the sense of (Eklund, 2015) or of a ‘bedrock concept’ in the sense of (Chalmers, 2011). I will in particular discuss Eklund’s views of the significance of what he calls ‘conceptual fixed points’ below and contrast it with what conclusions I hope to draw from inescapable concepts.

inescapable concepts 151 notion,12 without which one would not be able to think that one thinks that p. And arguably, if one can’t think that one thinks that p, then maybe one would not be able to think that p and thus not have thoughts at all. I won’t try to resolve the question whether there are inevitable concepts, and if so, what might follow from this. Instead I would like to focus on inescapable concepts, whether there are any and what might follow from that. But it is important to contrast inescapable concepts with inevitable ones, and what their different significance can be for various philosophical projects. One crucial difference between them is that every thinker must have all inevitable concepts, if there are any at all, while different thinkers might have different concepts that are inescapable for them: While thinker A has concept CA inescapable for them, thinker B might not have that concept at all, but instead has a different concept CB , which is inescapable for them. We will see this possibility illustrated shortly. This difference is crucial for how inescapable and inevitable concepts might be put to philosophical use, a topic we will discuss below. For now, though, let’s consider whether there might be any inescapable concepts, which ones they might be, and what considerations might support that they are inescapable. Although it might at first seem that no concept will be inescapable, since one might have strong reason of some kind or other in favor of replacing the concept, there are some prima facie good examples of such concepts, which I would like to turn to now. A first example of candidates for being inescapable concepts are logical concepts. The question whether logical concepts are inescapable is closely tied to the question whether logic is rationally revisable. Can it ever be rational for us to give up our own logic in favor of a different one? Suppose I presently reason classically, but someone suggests that this is a mistake and that I should reason with a different logic instead, say intuitionistic logic. Since intuitionistic logic differs from classical logic by whether the inference from ¬¬p to p is valid, the question comes down to whether I should replace my concept of classical negation, for which this inference is valid, with a different one of intuitionistic negation, for which it is not valid. For the sake of the example, I am here assuming a picture according to which such inferences are somehow closely tied to the relevant logical concepts, a picture that is widely, but, of course, not universally accepted. On this picture, a change in logic comes with a change in logical concepts. Should I continue to employ classical negation and accept this inference, or should I give up on classical negation in favor of its intuitionistic alternative? To answer this question I will have to assess the two options and determine which one wins out on balance of the reasons for and against making the switch. And to do this, I will use my best means for evaluating options in general,

12 See (Perry, 1990).

152 idealism and the harmony of thought and reality including reasoning and inference. But these methods for evaluating anything include deductive reasoning, in particular reasoning with classical logic, which at present is still the logic I employ in and accept for my deductive reasoning. So, using the best methods I presently accept, I can easily conclude that I should continue to use double negation elimination, and that adopting intuitionistic logic instead would be a mistake. I can reason simply as follows: suppose ¬¬p is true. Then it must be that ¬¬p. But then, reasoning classically, p must be the case, and thus p is true. So, the truth of ¬¬p guarantees the truth of p, and thus the classical inference form of double negation elimination is valid. And thus I should not switch to intuitionistic negation. If I were to switch, I would lose a valid form of reasoning, which would be a clear mistake. This argument for the validity of double negation elimination is a purely deductive argument from no premises, and it would thus seem to outweigh any argument or reason in favor of switching that I might encounter. Whatever argument someone might have in favor of intuitionistic logic, for example from the theory of meaning, it would never be as strong as a purely deductive argument from no premises. Thus it seems that not only ought I not to switch, given the information I have so far, but I have an argument that I ought never to switch, no matter what. It would therefore not be rational by my own lights to revise my own logic and my basic deductive belief forming mechanism, or so goes the argument in a nutshell. Whether this argument is in the end correct is a tricky question. A first worry is that there seems to be something question-begging about continuing to reason classically to show that classical logic is the best logic. But whether this is questionbegging in a sense that would invalidate these considerations is not at all clear. At most classical reasoning in favor of classical logic is rule-circular, but not premisecircular. After all, I do not rely on a premise like “classical logic is the right logic”, I merely use the classical rules, which are the rules I presently accept as the best ones to draw a conclusion about classical logic.13 To highlight a second concern, consider that the easiest way to show that classical logic is to be preferred is to show that the classical rules are all valid, something we can do using the classical rules together with the truth rules. But as is well known, the classical rules together with the “naive” truth rules are inconsistent, as is illustrated by various paradoxes like Curry’s paradox or the liar paradox. It might be natural to take this to show that something is wrong with the truth rules, but the same issue arises with them as does with the classical rules for the standard connectives: we can show that the truth rules are to be preferred over any alternative in a rule-circular way as well. The truth rules can be shown to be valid just as the rest of the classical rules can be, with the same resources. But how

13 See (Dummett, 1978a) and (Boghossian, 2000) for more on the status of rule-circular reasoning.

inescapable concepts 153 can the argument that the classical rules and the truth rules are valid be any good when we can show that these rules allow us to derive anything? These issues quickly lead to thorny questions about our rational response to the paradoxes, whether it would be rational to revise our own rules in light of inconsistency, or what other option might be preferred. This is not the place to consider these questions. For what it’s worth, I have presented my own view on these issues in (Hofweber, 2022b), where I argue that it is never rational to revise one’s own logic, including the truth rules, even in light of paradoxes. Naturally, many other people have developed rather different views on this matter.1⁴ Whichever view on this matter is correct is obviously a substantial further question, which we can’t hope to resolve here, but the point remains that all those who hold that logic is not rationally revisable for some reason or other in effect hold that logical concepts are inescapable: they cannot rationally be replaced with different ones. And we can see how this might be from the rule-circular arguments for the validity of one’s own rules. Thus logical concepts are a possible example of inescapable concepts, and rule-circular reasoning illustrates how it could be that such concepts are never rationally replaceable by one’s own lights. If it is indeed correct that logical concepts are not rationally replaceable, then this would illustrate a key feature of inescapable concept: it would not be required that everyone has the same ones. It would be irrational for me to switch from classical logic to intuitionistic logic, but it would also be irrational for an intuitionist to switch to classical logic.1⁵ Thus classical negation would be inescapable for me, while intuitionistic negation would be inescapable for an intuitionist. This shows that inescapable concepts do not need to be inevitable. One can be a thinker and have thoughts in general, while not having all those concepts that are inescapable for someone else. This difference is crucial for the overall philosophical significance of inescapable concepts. For example, inevitable concepts can play a role in a transcendental argument: they can be used in an argument that any thinker must have certain features that can be associated with an inevitable concept, since otherwise thought wouldn’t be possible at all. But no such transcendental argument is possible with

1⁴ For an argument that the concept of truth ought to be replaced because of the paradoxes, see (Scharp, 2013). For an argument that logic should be revised instead, see (Field, 2008). Whether rulecircular reasoning is legitimate in a proof of the validity of one’s own rules is naturally controversial, with some authors arguing that such arguments are defeated in light of disagreements about validity. See, for example, Hartry Field in (Field, 2020). I disagree and have discussed Field and others who take this line in (Hofweber, 2022b), where I more generally defend rule-circular arguments of this kind. This issue will reappear in Chapter 7 below. 1⁵ In this example there is a slight asymmetry in that the classical reasoner can prove, using the truth rules, that double negation elimination is valid, while the intuitionistic reasoner cannot prove that it is not valid. Still, they would not have any good reason to switch, and so it would still be irrational for them to do so, although less irrational than for a classical reasoner to switch to intuitionistic logic. This is an artifact of this particular example, and for other choices of logics the situation can be fully symmetric.

154 idealism and the harmony of thought and reality inescapable concepts, since not every thinker must have the same inescapable concepts. Inescapable concepts are merely such that if you have them, then you can’t rationally replace them. Nothing follows about thought or thinkers in general. But, of course, it is presumably pretty hard to show that a particular concept is inevitable, and possibly quite a bit easier to show that a concept is inescapable. To show the later, all one needs is an argument along the lines of the inescapability of logic considered above. To show the former one needs some insights into the nature of thought itself, something that appears to be much more elusive. The issue remains what conclusions of greater philosophical significance one might be able to achieve from the fact that a given concept is inescapable. This is our primary target in this chapter, but before we return to it, we should consider four more potential examples of inescapable concepts: ought, agency, fact, and number. The inescapability of the concept of a fact, in particular, will be significant later. Arguments for inescapability are not restricted to more narrowly understood logical concepts. A not unrelated argument can also be given for normative concepts like ‘ought’. Suppose I consider whether I should replace my concept of ‘ought’ with an alternative ‘ought*’. Suppose further that I can recognize that this new concept would be extensionally different from my old one, in the sense that what I ought to do is different than what I ought* to do for certain cases. Then I seem to be in a position to reject the proposed replacement out of hand, and conclude that it would be irrational for me to give up my concept of ought in favor of ought*. I can simply reason as follow: if I were to switch, then I would afterwards think about what I ought* to do. If my thinking would be effective, then I would do what I ought* to do. But what I ought* to do is not always what I ought to do, and so I might do things that I ought not to do even when I reason perfectly with the concepts that I would then have. Thus switching would lead me astray, it would lead me to doing things I ought not to do, and therefore I should not switch.1⁶ A similar, although more complicated, case can be made for the inescapability of the concept of agency. Could it ever be rational for me to replace the concept of agency with a different one, say shmagency? Here I am reflecting on what I should do: replace the concept or keep it? But doing something is engaging in agency and a conceptual connection between the concepts of doing and of agency make this vivid. I thus can’t just replace the concept of agency and leave the rest alone. I must also replace doing with shmo-ing, whereby shmo-ing relates to doing as shmagency relates to agency. But then I face a conundrum: should I do something that will result in my not thinking about what I should do in the future? After all, I would employ a different concept than doing something, and so I wouldn’t think about what to do any longer. But can it ever be rational to have an outcome of 1⁶ See (Eklund, 2017) and (Eklund, 2015) for a detailed discussion of the case of choosing between alternative normative concepts.

inescapable concepts 155 practical reflection that has the result that I will end all future practical reflection? Can I rationally do something that will preclude my future reflection about what to do? Can the result of rational practical reflection in effect end practical reflection itself? This question deserves serious consideration, but I hope even so far one can get a sense that an argument that this would not be rational might be forthcoming here. At least it would establish that there is a constraint of structural rationality that would prohibit such a concept replacement,1⁷ and if so, then this would be an argument that agency and doing are inescapable concepts: they can’t rationally be given up.1⁸ Considering rational replacement of one concept with another is often not an isolated affair, as the case of agency illustrates: a replacement of that concept would likely rationally require a replacement of other concepts that are conceptually related to this concept. The concept of agency is connected to the concept of doing, and with it also to various other action concepts: reading, singing, etc. A replacement of the concept of agency should thus engender a replacement of numerous other connected ones. Concepts whose replacement would give rise to large-scale follow-up replacements are likely harder to replace rationally, but that by itself does not rule out replacement. But sometimes these conceptual connections go deep, so deep that it is arguable that a wholesale replacement of the whole group of connected concepts would be irrational. The concept of a fact is a candidate for being inescapable for this reason. To outline the idea of why conceptual connections to the concept of a fact make the replacement of this concept problematic, suppose we consider whether we should replace our concept of a fact with some alternative: fact*. Maybe something is suspicious or allegedly flawed about our concept of a fact, and the concept of a fact* is suggested as an improved alternative. Could it be rational for us to make the switch? In particular, could it be rational to make the switch when engaged in, and for the purpose of, inquiry? Here there is an argument tied to conceptual connections that this could never be rational. Inquiry has a constitutive aim: to find the truth. But the truth is conceptually connected to the facts. It is a conceptual connection that if it is true that p, then it is a fact that p, and the other way

1⁷ Whether requirements of structural rationality can conflict with the requirements of substantive rationality is controversial, and if so, then it might be open to respond that although a requirement of structural rationality is violated, it might still be substantially rationally to replace this concepts. See (Worsnip, 2018) for more on this issue. 1⁸ This argument is developed more fully in (Hofweber, 2022a). The issue under discussion here is different than Luca Ferrero’s claim that agency itself is inescapable. Ferrero argues in (Ferrero, 2009) that agency is inescapable and that we cannot help but be agents. Our present question is whether the concept of agency is inescapable. This is an important difference. The former concerns whether we can be something other than agents, the latter concerns whether we could rationally replace our concept of agency with a different concept. While Ferrero’s argument is aimed at David Enoch’s ‘shmagencyobjection’ in (Enoch, 2006) to constitutivist theories about morality, I am only concerned with the latter issue of rational replaceability of one concept with another here. We will discuss constitutivism and shmagency at the very end of this book.

156 idealism and the harmony of thought and reality round. Thus to replace the concept of a fact with that of a fact* will also suggest a replacement of the concept of truth with that of a corresponding concept truth*. But that would still not be enough. We can’t coherently think of ourselves as being engaged in inquiry and trying to find out what is true* and a fact*. This would not respect the constitutive aim of inquiry, which is truth, not truth*. We would need to replace the concept of inquiry itself with its starred alternative: inquiry*. But then, could it ever be rational while being engaged in inquiry to replace the concept of inquiry itself with that of inquiry*, which does not have truth, but truth*, as its constitutive goal, and which does not aim at the facts, but at the facts*? That any such replacement would bring with it a different constitutive aim for the activity associated with the starred replacement concepts might suggest that on reflection one could never rationally make this switch, since it would badly lead one astray in what one hopes to achieve in inquiry, the activity we are presently engaged in and which we conceptualize as such. Thus, arguably, the concept of a fact is inescapable for us. This last example of a candidate for an inescapable concept is obviously of key relevance for our discussion whether we asked the proper question when we asked about our place in reality as the totality of facts. That question involved the concept of a fact as a key constituent, and its most obvious alternative, asking about our place in reality as the totality of all facts* replaces just that concept of a fact with an alternative concept of a fact*. We will investigate this example more closely in section 6.4 below, but hopefully it is already clear that the concept of a fact is a reasonable candidate for being an inescapable concept. Finally, let us consider the concept of number. Our main example for being a candidate for bridging the language-metaphysics gap was the proposed negative answer to the question whether natural numbers exist, which was reached via the assumption of internalism about number talk. But might that concept be flawed so that it should be replaced with an alternative? There is reason to think that the answer is ‘no’. That the concept of number is inescapable can be argued for via conceptual connections that this concept has to other central concepts, in particular quantificational concepts: There is a conceptual connection between something being an F and the number of Fs being larger than 0. To replace the concept of number with an alternative would seem to require replacement of the connected quantificational concepts as well. If the concept of number is flawed somehow, then so should be the concepts conceptually connected to it, in particular in the tight way pointed out. But that would mean that replacing the concept of number would in effect rule out future reflection on whether there is a concept that is to be preferred over a given one. To engage in this kind of reflection is to employ quantificational concepts, ones that would have to be replaced with the concept of number, and thus won’t be replied upon any longer after such a replacement. And arguably, restricting oneself in this way would be irrational and can’t be accepted by one’s own lights. A conceptual change can’t rationally rule out future reflection on which concepts to have, or so one might argue.

inescapable concepts 157 Obviously, all of these examples of candidates for being inescapable concepts deserve much further discussion.1⁹ What we have seen so far can only be considered an outline of an idea about how one might argue for the inescapability of various concepts. Still, these examples highlight that inescapable concepts are a possibility worth taking seriously, and they also highlight how one could argue that a particular concept is inescapable. And the examples given are not the only ones. Next in line would be concepts like belief, meaning, content, truth, reason, and similar ones. The above examples were simply raised to motivate the idea that some concepts might be inescapable, which concepts they might be, and what kinds of arguments one might be able to give that they are indeed inescapable. But these examples also are not completely unrelated, there is a certain suggestive similarity between them. The concepts that were candidates for being inescapable where in some way or other involved in or connected to the question whether we should replace such concepts with other ones. This question is one about what I ought rationally to do, and that involves the concepts of ought and of doing, and thus of agency. And if deductive reasoning is tied to rationality, then indirectly the question also involves logical concepts. Whether this similarity is a guide to other candidates for being inescapable concepts has to be left open for now, suggestive as it might be. Instead, I would in the following like to focus on the significance of inescapable concepts, if there are any, for metaphysics and for inquiry, and then have a closer look at whether the concept of a fact is indeed inescapable. This is closely tied to the question whether we did ask a proper question all along when we asked about our place in reality as the totality of facts, and with it whether we achieved a deep result when we were able to conclude that we are central in this way.

6.3 The significance of inescapable concepts Suppose we have found a concept where we have good reason to think that it is inescapable. That would mean that we have found an example of a concept C and a good argument that C cannot rationally be replaced by a different one C* by one’s own lights for the purpose of inquiry. What of philosophical significance would follow from this? There are essentially two ways to approach this: one more positive and one more negative. On the positive way, we could take this as a reason to think that we have found the right concept to apply to the world, and thus we have found something out about what the world is like: we found a concept that is a perfect match for the world. Furthermore, when we probe reality with a question that centrally involves 1⁹ I discuss the inescapability of logical concepts in (Hofweber, 2022b), that of agency in (Hofweber, 2022a) and that of facts just below. Eklund discusses thin normative concepts like ‘ought’ in (Eklund, 2017) and (Eklund, 2015).

158 idealism and the harmony of thought and reality this concept, then we have asked the right question relative to nearby alternatives. This would be a real insight. The question remains what else follows from the fact that this concept is ideally suited to the world, but on the positive outlook on inescapable concepts, we at least have some assurance that this concept is the right one to use. But there is also a more negative outlook on this situation. On this way of thinking of it, nothing has been shown about how our concepts relate to the world. Instead, something has been shown about our rational situation: we cannot rationally move on from where we are; we are rationally stuck and forced to remain in place. But that might be so whether or not our concepts perfectly match the world. We simply cannot escape from having this concept while being rational. And thus an inescapable concept can be seen in two rather different ways: either we got trapped by the requirements of rationality to remain in place, even if we are wrong, or else we got it right and the requirements of rationality gave us an insight into the world. We can consequently distinguish thinking of inescapable concepts as traps or as insights. Only if they are insights can we hope to draw any significant conclusions from them about the world. If they are traps, then we can conclude something about our own situation and what rationality does or does not permit in it, but little can be concluded about how our minds relate to the world. The question remains what argument we would have that even if there are inescapable concepts, then they are insights, not traps. This is the first of two related issues that we need to make progress on. The second issue concerns how reflecting on our concepts relates to the methodology of metaphysics and philosophy more generally. Here, too, there are two main ways of thinking of it. One is neo-Carnapian, inspired by the views of Rudolf Carnap.2⁰ On this view, philosophy properly done does not ask questions of fact; that is the job of the sciences. Instead, philosophy reflects on the concepts that are used by the sciences when they ask questions of fact, and it contributes to inquiry by making a proposal about what concepts might be improvements to the ones presently employed, or at least by proposing alternative concepts for the sciences to consider. Philosophy is thus closely tied to reflecting on which concepts we should have. The proper method of philosophy concerns conceptual engineering: finding improvements to our concepts and figuring out which concepts we should have. And this ‘should’ here is, of course, tied to rationality: it is tied to which concepts it would be rational for us to employ in inquiry. If we were to encounter any inescapable concepts, then this project would come to a roadblock. Since it would not be rational for us to employ a different concept instead, this would mean that no further philosophical work on this concept is fruitful. And this, again, can be either a good thing or a bad thing: the work has reached its goal and the concept

2⁰ See, for example, (Carnap, 1956).

inescapable concepts 159 is perfect, or somehow we are hindered from doing better. Either way, the work of philosophy ends when inescapable concepts are encountered. There is nothing left to do for philosophy here, assuming the Carnapian outlook on the subject. This is, in essence, how Eklund sees the significance of what he calls ‘conceptual fixed points’: concepts that we are unable to engineer away, for some reason or other. Conceptual fixed points, in Eklund’s sense, include hardwired concepts, inevitable concepts, and presumably also inescapable concepts. And as such they show the limits of what he calls an attractive conception of the methodology of philosophy: improving our concepts.21 Several other philosophers have also recently defended a similar Carnapian conception of the role of reflecting on our concepts and the methodology of philosophy.22 On such a conception of philosophy, it is less ambitious than many have thought. It does not aim to find general facts about the world, or stand next to the sciences in the project of inquiry. Instead, it focuses on our concepts, leaving questions of fact to others. But the more ambitious conception of philosophy as fact-finding can also be closely connected to the limits of conceptual engineering, in particular by focusing on inescapable concepts. This is the more positive project that I hope to pursue. A very different approach to inescapable concepts is a somewhat neo-Kantian one, inspired by the work of Immanuel Kant.23 On this approach, inescapable concepts are taken to be a guide to what reality is like. Reflecting on these concepts can be a crucial part of answering questions of fact about reality. They are not the end of philosophical work, but a key part of a beginning: a starting point for a project that gives us insight into reality by reflecting on our concepts and their inescapability. The key question, of course, is how one could pull this off. On the orthodox Kantian version, which is to say, Kant’s own version and those that follow it closely, it goes in something like this: there is a distinguished class of concepts, the categories, that have a special relationship to reality, or better the part of reality that is the world of appearances or phenomenal world. The concepts have this relationship to the phenomenal world because the world is constructed, somehow, with the use of those concepts. Reflecting on those special concepts allows us to draw substantial conclusions about what the phenomenal world must be like. Thus a special class of concepts is a guide to reality, understood as the phenomenal world.

21 See (Eklund, 2015). Of course, Eklund can maintain that improving our concepts is just a good part of philosophy, and that not all philosophical problems fall under it. Still, that good part of philosophy is done when we hit an inescapable concept. 22 See (Scharp, 2018), (Thomasson, 2016), and others. For a criticism of the view that there are limits to this Carnapian project because there are such things as conceptual fixed points or bedrock concepts, see chapter 18 of (Cappelen, 2018). 23 See (Kant, 1781). This approach is neo-Kantian simply in the sense that it is inspired by Kant, not in the sense that it is part of the neo-Kantian movement of the late nineteenth century, involving such philosophers as Hermann Cohen and Ernst Cassirer, amongst many others. For a survey of the historical neo-Kantians, see (Heis, 2018).

160 idealism and the harmony of thought and reality On a reformed neo-Kantian version of this approach, there is a different class of concepts, the inescapable ones, such that reflecting on them gives us insights into what reality is like. But contrary to the orthodox version, the reformed version does not rely on the distinction of the phenomenal and the noumenal world. It takes inescapable concepts to be a guide to reality full stop, with no modification. How this could possibly work as a strategy for metaphysical and philosophical progress is, of course, left open so far by this outline, but this is the basic idea behind the project and why it is neo-Kantian. This reformed neo-Kantian approach is exactly the kind of project that would get us what we were looking for above: Not only would it show that certain questions of fact can be answered by reflecting on our ways of representing the world, it would show that the questions that we asked in this way are the right questions to ask. Since the relevant concepts employed in the question are inescapable concepts, we could not have asked a better question, and thus this was the right question to ask in the first place. Putting these things together would give us a deep result, an answer to the right question to ask, in the sense spelled out above on page 147. If this could be pulled off, then we would properly answer metaphysical questions simply by thinking about our own representations of the world alone, together with that they are inescapable representations. And this latter part would guarantee that the question we asked can’t be improved upon by our own lights, at least with regards to the aspect that corresponds to the inescapable concept. Thus if we can answer the question at hand we should accept that we answered the right question and achieved a deep result. No other version of the question (again, with respect to that concept), can be rationally accepted by our own lights as a better question. And thus we would have made proper metaphysical progress by reflecting on our concepts alone. This approach will see inescapable concepts as insights, not as traps. And this seems to me to be the right reaction to inescapable concepts. The hard part is to show that this is indeed the correct approach, but I will argue in the following that this indeed is a viable and fruitful program. Before we can apply this idea to our main topic of idealism, let’s briefly dwell on the more general program just outlined. To carry out the reformed neo-Kantian program in its fullness one needs to do two things: first, one needs to find out which concepts are inescapable, and second one needs to show that such inescapable concepts indeed are a guide to reality and where they lead. To carry out the first part, one might have to proceed bottom up and argue on a case by case basis, for one concept at a time, that it is inescapable. Or one might be able to proceed top down and show that all and only the concepts in a certain group are inescapable, for a particular general reason. If a top down account is possible, then one would have not only the complete list of inescapable concepts but also a single reason why all of these concepts are inescapable. And, of course, the truth could be somewhere in the middle: for a certain group of concepts there is a uniform reason why they are inescapable, but other concepts can be inescapable for one or several different

inescapable concepts 161 reasons. However this will turn out, this is the first part of the reformed neoKantian project tied to inescapable concepts. The second part is to show that one can indeed draw substantial metaphysical conclusions from inescapable concepts. Merely because a concept is inescapable by our own lights seems to be quite independent of what reality in general is like. To carry out this second part of the neo-Kantian project one will have to give an argument that such conclusions indeed can be drawn, and this is closely related to showing that inescapable concepts are insights, not traps. Here, too, one might proceed in a top down or bottom up way. One might aim to argue that the inescapable concepts uniformly have some particular relationship to the world, or one aim to argue that some substantial metaphysical conclusion is connected to a particular concept being inescapable, and possibly a completely different conclusion to a different concept being inescapable. Either way, it would be good to see what conclusions one might be able to achieve along those lines. These two parts of the neo-Kantian project naturally correspond to two key parts of Kant’s original project: the metaphysical and transcendental deduction of the categories. What precisely these two parts of Kant’s project are is naturally subject to scholarly debate, but simply put, the categories are a distinguished class of concepts, the metaphysical deduction is supposed to show what the complete list of these concepts is, and the transcendental deduction is supposed to show that these concepts are guaranteed to apply to the world, and that from this substantial metaphysical conclusions about the world can be drawn. Both parts are pretty much top down, on the above classification, in the original Kantian version. And they are naturally taken to proceed via a transcendental argument: either experience or thought would not be possible at all unless these concepts apply to reality, at least on one particular understanding of reality as the phenomenal world. It is natural to take Kant to hold that the distinguished concepts, the categories, are inevitable in our sense: every thinker must have them.2⁴ And because they are inevitable one can aim for the strong conclusion that experience or thought are only possible because of the proper relationship between these concepts and reality. But on the present approach, inescapable concepts are not necessarily inevitable. Different thinkers can have different inescapable concepts, as is illustrated by the case discussed above of thinkers with different logics. This makes clear that a traditional transcendental argument is not a possible strategy for the present neo-Kantian project: no conclusion can be drawn from particular inescapable concepts about what is required for thought or experience itself, since different thinkers can have very different inescapable concepts. Traditional transcendental arguments are thus out. The question remains what kinds of arguments could possibly allow one to draw conclusions about reality from inescapable concepts. 2⁴ This, of course, is also subject to scholarly debate, but a natural way to think of the relation of the categories to thought itself. See (Filcheva, 2019) for more on this debate.

162 idealism and the harmony of thought and reality If different thinkers can have different inescapable concepts, how can we draw conclusions about what reality is like from certain concepts being inescapable? Here it is worth distinguishing two aspects, or sub-questions, of this problem: First, how can we draw conclusions at all about the world from our concepts and their being inescapable? This is the problem we have mostly been occupied with so far, relating to the language-metaphysics gap and the difference between deep and shallow results. Second, how can any such conclusions be truly valid, and not just something we are forced to affirm, in particular in light of the fact that other people can have different concepts which are inescapable for them? This second question will be the topic of the next chapter, so I will leave it largely aside until then. For now, let us continue our discussion of the first question instead. And here I hope it won’t be a surprise what kind of an argument might show that reflection on inescapable concepts lead to substantial insights into reality: just the kind of argument given above, namely the argument from internalism to conceptual idealism. It is an argument that proceeds from thinking about our own ways of representing the world to a conclusion about our place in reality as the totality of facts. Thinking about the concept of a fact and our talk about facts leads us to conclude that we limit the totality of facts. The totality of facts must conform to our conceptual representations. We were able to draw this conclusion without making any assumptions about what reality in general is like, we simply investigated what we do when we talk about facts. But so far that is only a shallow result, not a deep one, in the above sense from page 147 so far we do not know if we asked the right question when we asked and answered the question whether we are central to reality as the totality of facts. And thus so far we don’t know whether the result we achieved is deep or shallow. What is left to do is thus to apply the ideas from this chapter and to argue that there is no better question in the neighborhood of the question we in fact asked, and that therefore the result is deep. And that is closely tied to arguing that the concept of a fact is inescapable. If it were, then reflecting on an inescapable concept would lead to a deep metaphysical result about what the world is like. And being able to draw that conclusion in that way would vindicate the neo-Kantian program, at least for that one case. Whether there are other cases where this approach might bear fruit is left open by this so far, but it sure would be surprising if something that at first seems impossible only works in one case after it becomes clear that it is possible after all. Still, to have one case where this works is key for the larger approach, and thus what we need to do next is figure out whether the concept of a fact is indeed inescapable.

6.4 The inescapability of the propositional A concept is inescapable just in case it would always be irrational by our own lights to replace it with an alternative one for the purpose of inquiry. Suppose

inescapable concepts 163 someone were to propose that the concept of a fact should be replaced with an alternative concept fact*. They might argue that, in particular in light of the truth of internalism about talk about facts, the notion of a fact is unsuitable to do proper metaphysics. It is too closely tied to us and our language, and a more languageindependent notion of a fact* should serve us better in asking more appropriate questions in metaphysics, or so the proposal. Now, it should be made clear right away that there is nothing problematic about introducing a notion of a fact*, however that might go more precisely, and asking questions concerning it. The issue at hand is not whether we can ask meaningful questions about facts*, but only whether we ask better questions than asking about facts when we do so. So, the issue is whether asking about our place among the facts* is asking a better question than asking about our place among the facts. In particular, is it rational by our own lights that asking about the facts* is asking a better question than asking about the facts? To make progress on this issue it is useful to think a bit more about both facts and facts*. Starting with facts, we can note that there are a series of close, even conceptual, connections between the concept of a fact and other concepts. Most clearly, there is a connection between facts and truths: (47) It is a fact that p just in case it is a truth that p. Similarly, there is a connection between facts and true propositions: (48) It is a fact that p just in case the proposition that p is true. As noted before, this leaves open whether facts are true propositions, and whether facts are truths, but the connection between them obtains either way. And the concept of a fact is also connected to several other ones, either directly or via its connection to truth or propositions. To mention just a few obvious candidates for such a connection: (49) a. b. c. d.

Belief aims at the truth. My reason to Φ is the fact that p. The goal of inquiry is truth, and thus to find out which facts obtain. A belief is true just in case the proposition which is its content is true.

and so on and so forth. Of course, each one of these candidates for connections among the concepts of fact, truth, content, reason, and so on can be disputed, but if they are disputed, then usually the question simply is which other connection of this kind holds, not that there is no connection at all. What this points to is that there are close connections between the notions that concern what I called the propositional above: all the things we talk about with that-clauses. These concepts

164 idealism and the harmony of thought and reality effectively form a network that concerns the various aspects of the propositional: facts, truth, content, meaning, reason, and so on. And this suggests that it will be hard to replace just one of them with an alternative. A replacement of just one concept tied to the propositional might break these connections and leave our new network of concepts fragmented in a problematic way. How much damage such a replacement of just a single concept would do depends on what the original concept gets replaced with. If it is very close to the original one, then maybe little overall fragmentation and damage will be done to our network of concepts concerning the propositional. If it is rather different than the original concept, then either we have to accept fragmentation or we need to replace other concepts as well. We might even need to replace the whole family of concepts that concern the propositional if we replace one of them. We can say that the propositional is inescapable just in case the following holds: any rationally acceptable replacement of a single concept tied to the propositional will itself be tied to the propositional. In other words, we cannot rationally escape that that-clauses are tied to these concepts. If I were to replace the concept of believing that p with some alternative, shmelieving, then for this replacement to be rationally acceptable, it would require that shmelieving also takes that-clauses as complements: shmelieving that p. And thus shmelieving also belongs to the propositional. Any dimension of rationally acceptable improvement thus does not concern the that-clause. This would be especially relevant for our discussion here, since internalism concerns the thatclause aspect of the propositional: whether the relevant complements are referential, and whether quantification over what we believe, hope, etc., is used in its inferential reading. To see whether the propositional as a whole is inescapable, let’s look at the example of how a replacement of the concept of a fact with that of a fact* might go, and how different these two concepts would have to be. The motivation for asking a better question by replacing the concept of a fact in light of the truth of internalism was roughly this: if internalism is indeed true, then facts are closely tied to our representations of the world. Thus even if we managed to answer the question whether we are central to reality as the totality of facts, this reflected more on the inadequacy of the question than on the quality of the answer. To formulate the proper question, the question we should have asked instead, we need to state it in different terms, ones more suitable for metaphysics and less tied to our own representations. Instead of asking about the facts we should thus be asking about something else: the facts*. This proposal makes sense, although the question remains whether we would be asking a better question if we asked about facts* instead. To evaluate this, let’s think a bit more about what facts* might be. Although facts* can be introduced in many ways, it is useful to highlight some ways in which facts* will not be like facts. For example, if facts* are intended to be language independent and not tied to our representations, then it is most natural to think of them as constituting some domain of entities that exist independently of our representations of them.

inescapable concepts 165 They might be some worldly things, say sets of possible worlds, or ordered tuples of some kind, or sets of sentences, or what have you. Talk about facts* would consequently be in the spirit of an externalist view: Talk about particular facts* would be referential, and quantification over facts* would be used in its domain conditions reading. It is important to note that then none of the facts* are a fact. Facts are not to be found in the domain of entities, by the same argument from page 142 that showed that natural numbers do not exist. To quickly repeat it, consider the question whether a particular entity e might be the fact that snow is white. If internalism is true, as we are assuming for now, then fact-terms are constitutively non-referential. Thus ‘the fact that snow is white’ as I just used it two sentences ago is non-referential, and in particular does not refer to e. So e, whatever it might be, is not the fact that snow is white. And since e was chosen at random this means that no entity is the fact that snow is white. And similarly for all other facts: no fact is an entity, assuming internalism. Thus facts are not in the domain of external quantification. But facts* do exist and are in that domain, by assumption, so facts* are never facts. They are simply something completely different, despite having a similar name. In particular, the facts* don’t extend the facts in some way that makes them more independent of our representations. Instead they are completely different. And so switching from talk about facts to talk about facts* is a rather radical and fundamental change. Such a change would also recommend many other changes. Facts correspond to truths, but do facts* correspond to truths? Since facts* do not correspond to facts, but truths exactly correspond to facts, the answer must be: no. We then have a choice to make: should we break the connection between fact and truth, and move in two different directions with fact* on the one hand, and truth on the other, or should we keep the connection and replace both truth and fact with truth* and fact*? If we take the first option, then it is not clear that we made any progress. Although we won’t be asking any longer about our place among the facts, we are left with the question about our place among the truths, which is rather similar to the question we wanted to get away from. If one wants to hold that asking about our place among the facts was to ask the wrong question, and to ask about the facts* is the better question, that person should say the same about our place among the truths, and thus recommend that we move towards asking about truths* instead. To really make progress it won’t be enough to replace just one concept with an alternative that is not tied to our representations. We also need to go further and replace closely connected concepts as well. Thus if we are taken by the motivation to replace fact with fact*, we should on similar grounds also replace truth with truth*. Truth* and fact* can be related to each other just as fact and truth were, and thus their connection will be preserved. But we will have to move to the starred versions of both concepts together. But things don’t end here. Belief aims at the truth, let’s grant, but does it aim at truth*? Since truth and truth* are rather different, just as fact and fact* are,

166 idealism and the harmony of thought and reality the answer must be: no. Believe doesn’t aim at truth*, but its starred equivalent, belief*, can aim at the truth*. Inquiry itself also aims at the truth, but it does not aim at truth* for the same reason. Only inquiry* would aim at truth*. And we can see that similar considerations motivate a more wholesale replacement of our notions tied to the propositional with their starred equivalents. If we suspect that our present concepts do not allow us to properly state the questions that should be asked in metaphysics, then we should blame our conceptual attachment to the propositional as a whole, not merely one single member in this family of concepts. The proposal should consequently be that we should ask about reality* as the totality of facts*, about truth*, reasons*, content*, inquiry*, what we should believe*, and so on. The question remains how we should evaluate this large-scale and wholesale replacement of our family of concepts tied to the propositional with their starred equivalents. Would it be rational for us, by our own lights, to replace the family of concepts tied to the propositional with their starred equivalents? In particular, would it be rational for us for the purpose of inquiry to do so? This is a question we must consider, having both sets of concepts before us as candidates to be evaluated for which one is better. We need to address this issue before making the switch, and we need to think about it using our best present concepts. So far we are engaged in inquiry and we still employ the concepts attached to the propositional: truth, fact, reality, etc. The question is whether we should give those up, and that question is to be thought about using the concepts we presently have, including those tied to the propositional. Thus any such proposed change in our concepts must be appreciated as being rational from our present standpoint, using our present concepts when we think about how we should evaluate this proposal. This asymmetry does not settle the issue by itself in favor of the old concepts. It is often the case that even using a present concept when thinking about a switch, one can appreciate that the switch would be a good and rational one. One can see that a particular concept is flawed even when one employs that concept in those reflections on its being flawed. But in this case things are different. One cannot rationally accept that one should replace all the notions tied to the propositional with their starred equivalents. When I consider, in my present position, whether I should replace the concepts tied to the propositional with their starred equivalents, then I need to consider the reasons for and against doing this. Those reasons are closely tied to whether such a switch would improve my position with respect to what I try to do: find the truth, make progress in inquiry, determine what the facts are, and so on. If I made the switch, then I would afterwards no longer think about what I should believe, what is true, and what reasons I have for accepting this or that proposition, since I would no longer employ the concepts of belief, truth, reason, and so on. I might get other benefits from the switch, but I would severely hinder myself in making progress in inquiry, since I would in the future no longer reflect on what to believe or what I have reason to accept. For a switch in concepts to be rational it needs to be

inescapable concepts 167 supported by the right kinds of reasons, those that support the suitability of these concepts for the purpose of inquiry and for the purpose of representing the world. But just this would be undermined by such a switch. Inquiry aims at the truth and the facts, and giving up these very concepts, including that of inquiry itself, won’t help me get better at inquiry. I would be a fool to replace my concepts of truth, fact, and inquiry with their starred equivalents in order to get better at inquiry. Giving up the concepts that I need to conceptualize the very activity I engage in, inquiry, and its goal, truth, would be quite irrational. Maybe then we should not evaluate the starred concepts ‘for the purpose of inquiry’. Maybe we should not give inquiry this special standing so that alternative concepts are evaluated by how well they contribute to it. We can grant that the starred concepts are better suited for inquiry*, but that is not the project we are engaged in when we evaluate alternative concepts. Should we switch to inquiry* then? Should we give up on aiming for the truth and the facts while engaged in inquiry, and instead aim for truth* while engaged in inquiry*? The issue again is whether we can rationally accept, in our present position, that this is the right thing to do. And the problem for this is that we are presently engaged in inquiry, aiming at the truth. What reason could we have for giving up this whole project and switch to such a radical alternative? And how could we appreciate that reason such that it would be rational for us, on those grounds, to give up the very aim of truth? A right kind of reason can never deliver such a result, although a reason* might support this. But reasons* have no pull on us as we are at present. Thus from our present position it is never rational to abandon the aim of truth, and with it to give up on inquiry and replace it with something else. Thus no matter how one approaches this issue, there is no rational escape from the concepts tied to the propositional. We cannot rationally replace them with alternative concepts that do not themselves belong to the propositional. If such concepts can be improved in some form or other, they can’t be improved by moving them away from the propositional towards something else. They can’t rationally be improved by replacing them with concepts that do not take thatclauses as complements. To illustrate, the above argument does not rule out that the concept of belief, say, can be improved by replacing it with belief+ , whereby belief+ is just like belief, except with the further requirement that one has a strong disposition to verbalize what one believes. Such a suggested replacement of the concept of a belief would still be part of the propositional and still aim at the truth. But any radical replacement, like that with belief*, which aims at something nonpropositional like truth*, can be ruled out as irrational by our own lights. There are other good reasons to reject a replacement of the concept of belief with that of belief+ , to be sure, but the argument given above focuses only on replacements that matter for our discussion now, namely those that hope to escape the propositional. This conclusion is slightly more general than what we need for what matters for now: the inescapability of the concept of a fact. Instead of focusing just on one

168 idealism and the harmony of thought and reality concept tied to the propositional, the above argument concerns the whole family of concepts tied to the propositional. But our main target remains the concept of a fact, since it is the relevant one for the evaluation of the significance of the proposed argument for idealism. The concept of a fact is, of course, included in the family of concepts that concern the propositional, and so we can conclude that the concept of a fact is inescapable, at least in the sense that matters for us now: it cannot rationally be replaced with a different one that does not itself belong to the propositional. The concept of a fact is inescapable for us, but it is not inescapable for some isolated reason. It is inescapable, since it belongs to a larger family of inescapable concepts: those tied to the propositional. Although it can in principle be possible that concepts tied to the propositional can be improved in some isolated way, they can’t be improved by moving them out of the propositional. Thus any worries about a concept tied to the propositional being flawed, in particular in light of internalism about the propositional, are misguided. Whether the concept of a fact can be improved upon at all, even when the suggested improvement remains within the propositional, is doubtful, but can strictly speaking be left aside for now. Maybe our ordinary concept of a fact is minimalist, in that it always supports the equivalence between p and it being a fact that p. And maybe an improved concept only tolerates this equivalence when ‘p’ is a descriptive statement, as discussed above on page 53. None of this matters for our purposes here, doubtful as it might otherwise be, since we are concerned with the significance of the consequences of internalism about talk about facts for metaphysics. Internalism applies equally to other notions within the propositional, and thus the consequences would be the same for such a proposed improvement that remains within the propositional. The key for us is that a radical replacement of the concept of a fact with a concept of fact*, which is outside of the propositional, would never be rational. And with that we can now properly evaluate the question we asked when we asked about our place in reality as the totality of facts.

6.5 Conceptual idealism as a deep result Conceptual idealism holds that our human minds are metaphysically central to reality as the totality of facts, since they constrain the range of the facts that can obtain. I argued for this view above by considering what we do when we talk about facts, in particular all the facts. The semantics of the inferential reading of quantifiers over facts guarantees this. But even if this is perfectly correct, by itself it is only a shallow result: it answers a question without any consideration of whether that was the right question to ask. To turn it into deep result, in our above sense on page 147, we need to also find out whether we asked the right question. This in turn is closely connected to the issue whether or not the concept of a fact is inescapable

inescapable concepts 169 for us. If it is inescapable, then we cannot accept that a nearby alternative to the question about our place in reality as the totality of facts is a better question to ask. In particular, with a proposed replacement of the concept of a fact with fact* which is supposed to avoid internalism and thus the propositional, the question of our place among the facts* cannot rationally be accepted, by our own lights, as a better question. Thus the question we asked originally was the right question to ask, relative to close alternatives, and thus the result we achieved was a deep result: it not only answers the question we asked, but it also was the right question to ask. To clarify, the claim is not that asking about our place in reality as the totality of facts is the best question that can be asked full stop. There might be many completely different questions that are better questions to ask: Why is God hidden? or what have you. The issue is instead that there is no better nearby alternative question to the question about our place in reality as the totality of facts. The concept of a fact is just one of several concepts that appears in this question, but it is the prime candidate for a possible replacement or improvement. The concept of reality is simply understood as the totality of facts, and the notion of a totality seems to be pretty immune to improvement. If the concept of a fact is inescapable, then this question is reasonably taken to be the best question in this neighborhood. And so we did ask, and answer, the right question: a question that is a proper question to ask in metaphysics. None of this should really be possible, and to make that vivid once more, let’s remind ourselves what we did so far. Not only did we answer a substantial metaphysical question about own own place in reality simply by thinking about our own representations alone—which already should be impossible—but furthermore we were able to conclude, again by thinking about our representations alone, that we not merely answered some flawed question we happened to ask, but the right and proper question for metaphysics. This must seem impossible, but it isn’t. Once we focus on the question we ask, and not just the answer we proposed, and on the rational requirements that come with employing a particular concept as your own concept and engaging in a particular activity like inquiry, we can see that this indeed can happen. But there is one important issue that remains: Even if all of the above is correct, there is a nagging feeling that the inescapability of a concept is merely something that we are rationally compelled to accept, but nothing much follows from our rational compulsion about the world. Other creatures can have different concepts that are inescapable for them, and they will be compelled to accept conclusions tied to their inescapable concepts. But in light of this, how can we claim that reflecting on our own inescapable concepts leads to a deep result about the world, or even to a result at all, given that we must recognize that which concepts are inescapable is something that can differ from person to person? Wouldn’t the more proper reaction be to hold that although we must think that these “results” are “deep results”, others must think the same about other propositions, and nothing really

170 idealism and the harmony of thought and reality follows about the world from all this. All that we can conclude is that having certain concepts comes with certain constraints on how to swap them out, but not that the world must conform to these concepts. Although we can’t help but accept such claims, we are simply rationally trapped into having to think this. In other words, inescapable concepts are traps, not insights. And these “deep results” are more results about what we have to think than what the world is like, or so this worry. We now need to address these issues head on, which is the task of the next chapter.

7 The immanent stance 7.1 Rational traps One of our original worries about achieving results in metaphysics via reflection on inescapable concepts was the question whether such concepts lead to traps or insights. If rationality requires me to think that this concept is the best one to apply to the world, might this be because rationality has trapped me into a delusion? Could it be that rationally requires me to think this, even though it is false? Or does being rationally compelled to think this lead to an insight into what the world is like? Am I simply forced by rationality to stick with a certain concept when I have it, even if this only shows something about me and my own situation, rather than something about what the world is like? In this chapter I hope to make progress on this issue. It is an important one for the present project, since without addressing it, a worry must remain that in the end we never properly reached the world, but remained stuck with results that are only about ourselves, our concepts, and the requirements of rationality as they apply to us. To reach the world we need to make clear that these rational requirements point us to what the world is like, not merely to what our minds internally should be like. One key concern about reaching from our minds to the world is that different people can have different inescapable concepts, and thus different requirements on their minds. How can we draw conclusions about the world from inescapable concepts in light of this variation among inescapable concepts? This thought makes it tempting to hold that we get trapped by our inescapable concepts: we are restrained from moving on to new concepts, no matter what the merits of these new concepts might be. To illustrate the idea of a trap in this context, we can distinguish causal traps from rational traps. A causal trap is one where causal forces do not allow one to move to a different place. Here movement and place are taken rather literally at first, as when one falls into a hole, or it can be taken a bit more metaphorically, as when causal forces, internal or external, don’t allow one to change careers or relationships. A rational trap, by contrast, is one where rational forces do not allow one to move to a different place. The constraints of rationality restrict me where I can go from here, and that is not because I am at the right place to be, but because I am stuck in a hole, metaphorically this time, of course. The question is whether we can makes sense of the forces of rationality trapping us, just as we can make sense of causal forces trapping us. It is natural to think that the notion of a rational trap is incoherent. The forces of rationality only Idealism and the Harmony of Thought and Reality. Thomas Hofweber, Oxford University Press. © Thomas Hofweber 2023. DOI: 10.1093/oso/9780198823636.003.0007

172 idealism and the harmony of thought and reality prevent me from going where I should not go, and that can hardly be seen as being trapped. To be trapped requires that one is stuck in a bad place, which could be taken as a place where one shouldn’t be. In the case of a rational trap this would mean that one is rationally required to be in a place one should not be in. But how could we make sense of this ‘should’ if not itself tied to rationality? Clearly one can’t coherently hold that one could rationally be required to believe something that one rationally should not believe. But could it be that one is rationally required to belief something that is false? Here the thought is not just that one might have evidence that supports a false belief. Clearly that can happen, since one can have misleading evidence. But that by itself is hardly a trap, only an obstacle. I can easily overcome it by gathering new evidence that points towards the truth. To be properly trapped one would have to be in a position where no rationally permitted move gets one away from a falsehood. Thus to be properly rationally trapped one would have to be rationally required to believe a falsehood, and one would have no rationally permissible ways available to overcome one’s own unfortunate situation and come to believe the truth. Could we ever be trapped this way, in particular by our own inescapable concepts? Let’s consider an example of a possible case of a rational trap in more detail. We have seen a case of a possible clash of different inescapable concepts above, namely that of logical concepts. A key contrast for us here is the one between inescapable concepts and inevitable concepts, which was also briefly discussed above on page 150. Every thinker must have the same inevitable concepts, if there are such concepts at all, but different thinkers can have different concepts that are inescapable for them. We saw not just the case of different logical concepts in Chapter 6 as an illustration of a likely case of this, but also the starred vs. unstarred families of concepts tied to the propositional—fact, truth, inquiry, etc.—are likely another example of inescapable concepts that are not inevitable. If one starts out with intuitionistic logic, it would be irrational by one’s own lights to switch to classical logic. But if one starts with classical logic, it would be irrational to switch to intuitionistic logic. If one starts with the concept of a fact, it would be irrational to switch to fact*. And if one starts out with starred concepts like fact* and inquiry*, it would likely also be irrational by one’s own lights to switch to the unstarred ones. This seems to support that inescapable concepts cannot lead to insights, but must be seen as traps. After all, different people can have different, even conflicting, inescapable concepts, and they can’t all be insights into one and the same reality. But neither one can rationally by their own lights replace their inescapable concept with a different one, and so at least one of them seems to be trapped. The worry thus is that nothing follows about the world from these concepts being inescapable. Whatever follows must be about the thinkers that have these different concepts, but not otherwise about the world. Everyone will think that

the immanent stance 173 their inescapable concepts are special, just as we do, but the world is indifferent to this. Each one of us can’t help but think that their own inescapable concepts fit the world, but most will be wrong, since we have lots of different and conflicting inescapable concepts, but just one world. We are thus trapped by rational compulsion to think the world is a certain way, whether or not it really is that way. This argument for inescapable concepts leading to traps needs to be addressed, since I want to maintain that they lead to insights instead. And the examples given seem to illustrate the possibility of rational traps: those with alternative logical or propositional concepts are trapped, and we can see that this is so. And with this we can recognize that we ourselves might be just as trapped as they are. To be clear, the issue is not that one person will be rationally required to believe that snow is white while another one will be required to hold that snow is not white, due to their having different concepts which are inescapable for them. The differences are not that radical. But there will be a difference between one person being rationally required to hold that the concept of classical negation, or of a fact, is the right one to employ, while another person will be required to hold that it is the concept of intuitionistic negation, or that of a fact*. This is still a notable difference, and one that is crucial not so much for achieving mere results, but especially for deep results in the sense of Chapter 6. Our main issue will thus be how the variation of which concepts are inescapable for different thinkers affects this move from mere results to deep results. To evaluate this situation more carefully, let us consider the case of disagreement about logic in some more detail. Suppose I accept classical logic and reason in accordance with it, while Iris accepts intuitionistic logic and reasons in accordance with it. Let’s assume that we are both fully behind our own logics: we do our best to reason in accordance with them, and we reflectively fully endorse that this is the proper way to reason. In this case we both accept the same rules of deductive inference, except for double negation elimination, which I accept, but Iris rejects. Since we both fully accept upon reflection that we are reasoning properly when we reason according to our own rules, we each reason correctly by our own lights when we reason in accordance with classical or intuitionistic logic, respectively. Now, I can prove, in the rule-circular way discussed in Chapter 6, that double negation elimination is valid, and thus that I have the right logic, while Iris is missing out, since she is hesitant in making double negation elimination inferences. This proof proceeds from no premises using only rules that I accept as valid rules of deductive reasoning.1 A proof of this kind gives me the highest degree of certainty that I accept: something that can be shown from no premises with only deductively valid rules. And so I conclude, and must conclude, that I am right in

1 To repeat it for the case of double negation elimination: suppose ¬¬p is true. Then ¬¬p, by truth elimination. But then p, by double negation elimination, and so p is true, by truth introduction. So, the truth of ¬¬p guarantees the truth of p, and thus this inference is truth-preserving and therefore valid.

174 idealism and the harmony of thought and reality sticking with classical logic, while Iris is wrong in sticking with intuitionistic logic, since she is missing out on drawing valid inferences.2 Of course, Iris will think something similar about me. She will think that I overreach in my acceptance of rules that are not valid, and that she is right in sticking with intuitionistic logic, while I am wrong with my endorsement of classical logic. Thus the situation seems essentially symmetric.3 We each stick to our guns, and insist that the other one is wrong. This might naturally be seen as supporting two conclusions: first, it illustrates how someone can be rationally trapped; second, it substantiates the worry that I myself might be trapped right now. I can see that Iris does nothing wrong internally, and that she reasons perfectly by her own lights. On my assessment, she has the wrong logic, but she is unable to see this. The forces of rationally acting on her, determining what she should believe by her own lights, do not allow her to see that her own logic is the wrong one. And why not now apply this to myself as well? If I can see that Iris is trapped, although Iris can’t appreciate this, maybe I am trapped as well in just the same way, and I can’t appreciate it? This possibility that I might be trapped as well can naturally be taken as properly undermining my confidence in my own logical proofs, in particular the proofs of the validity of my own rules. It seems then that I should accept the serious possibility that I am being misled by my own deductive reasoning, and thus that I should not conclude that I got it right after all. But drawing the conclusion that I myself am trapped would be a mistake. It is true that the fact that Iris also thinks that she is correct should give me some pause that I might not be correct in my assessment of my own situation. This carries some weight in my consideration about whether I got it right. I need to properly consider this fact in my overall deliberation. But if I am fully rational, then I will weigh all the reasons properly, and thus I will weigh the symmetry consideration and Iris’s disagreement against the reasons in favor of my getting it right: my deductive proof of the validity of my own rules. And once I consider all the reasons, there can only be one result: My proof that my own rules are valid 2 As acknowledged in Chapter 6 already, this example is slightly more complex in that both classical logic as well as intuitionistic logic together with the “naive” truth rules give rise to paradoxes, and with it the issue about what the rational reaction to the paradoxes is. Several philosophers have suggested that the paradoxes require us to give up on those truth rules or alternatively give up on one of the other logical rules. See (Field, 2008) for both a survey of this debate and a defense of the latter option, and (Scharp, 2013) for the former one. I have defended the position that the rational reaction to the paradoxes is not to give up either the truth rules nor the rules of classical logic in (Hofweber, 2008) and (Hofweber, 2022b). Others have argued for similar conclusions in a different way, for example (Cobreros et al., 2013) or (Ripley, 2013). What the proper reaction to the paradoxes is is beyond the scope of this book, and I am sidelining this issue here for the sake of the example used here to illustrate the significance of a variation of inescapable concepts for different reasoners. 3 As already pointed out in Chapter 6, the examples chosen does not make the situation fully symmetric, since I can prove that the rule I accept, but Iris rejects, is valid, while Iris can’t prove that the rule she rejects, but I accept, is not valid. Still, this difference is an artifact of choosing simple examples of different logics, and not essential for our discussion here.

the immanent stance 175 wins out quite clearly. I can prove deductively from no premises that I got it right. This easily outweighs the reasons for doubting that I got it right that come via Iris and the near enough symmetry of our situations. Deductive proof from no premises establishes conclusively that I got it right, and the doubts coming my way via Iris are nowhere near strong enough to outweigh this. If I am rational, I will stick to my proof, even though the symmetry considerations tempt me to think otherwise. Unless, of course, I am somehow no longer entitled to rely on this proof when weighing all the reasons. In particular, in light of Iris’s disagreement and of my recognition of the parallels between her situation and my own, I might no longer be entitled to count my deductive proof as a reason in favor of classical logic. This is a very legitimate concern that we need to look at more carefully. Although it seems clear that a deductively valid proof from no premises provides stronger reasons than Iris’s disagreement with me, the question remains whether I am entitled to continue to rely on this deductive proof, in particular in light of the fact that the validity of my own rules is at issue. To continue to insist on the validity of this proof in light of one of the rules used in it being contested does seem to be question-begging in some way. Not in the most straightforward way, in which I am simply relying on the conclusion as one of my premises, but in the way that I am continuing to use one of my rules when the validity of that very rule is in question. On the other hand, what rules would I use other than the ones that at present I take to be the best ones? Still, there is a worry that somehow the continued use of these rules is problematic and illegitimate. Some philosophers have tried to make a case for this in more detail. For example, Hartry Field has recently argued in (Field, 2020) that in the face of a disagreement over which rules of logic are valid, my own contested rules get defeated. I am no longer entitled to any conclusions I reach with contested rules, Field argues, since those rules are defeated in light of the challenge to them. I believe that this is mistaken, for the reasons given momentarily. First, let’s frame the issue with some commonly used terminology. Following (Pollock, 1974) we can distinguish rebutting from undercutting defeat. These notions are primarily motivated by examples like the following: Rebutting defeat occurs when Pete tells me that p, but later Sue tells me that not p, and Sue is much more reliable on this matter. Undercutting defeat occurs when Pete tells me that p, but later Sue tells me that Pete is notorious for making things up. As a first approximation, these notions can be spelled out this way: Rebutting defeat of (a reason for) a belief that p is a stronger reason to believe not p, whereas undercutting defeat of (a reason for) a belief that p is reason for holding that the reason on which one based one’s belief that p isn’t a good reason after all. It is hard to see how the argument for the validity of our own rules can be defeated in the sense of rebutting defeat. The argument for their validity proceeded from no premises by employing only deductive rules, ones we reflectively endorse

176 idealism and the harmony of thought and reality as the right ones. Deductive proof from no premises is the strongest form of an argument there is, or so I would evaluate the comparative strengths of reasons for believing something. No other consideration can be stronger, and thus no other consideration can defeat, in the rebutting sense, a conclusions I can reach this way. The issue thus comes down to whether my entitlement to rely on a double negation inference is defeated in the undercutting sense by a challenge to it. In the aforementioned article, (Field, 2020), Field argued that reasonable challenges to one’s own logic lead to undercutting defeat of the rules in question. Let me illustrates this situation with an analogous scenarios of our perceptual beliefs being challenged. Consider the debate about composition: are there ordinary objects like tables, or are there only simples arranged table-wise, but no tables? It is tempting to simply rely on perception to rule out the view that there are no tables. After all, I can see a table right in front of me. But the nihilist challenger, who holds that composition never occurs and that there are no composite objects like tables, will insist that perception does no longer entitle you to believe in objects after nihilism has been pointed out as an alternative. After all, things would look just the same even if composition did not occur.⁴ In light of the challenge by the nihilist our entitlement to our perceptual beliefs in objects gets defeated. It might have been fine before, but now it is no longer, and thus relying on such beliefs to refute the nihilist is illegitimate, or so the argument. And similarly in the debate about logic. Once intuitionism has been suggested as an alternative logic, my reasoning with double negation elimination has been defeated and thus cannot be relied upon any more in the refutation of intuitionism and in my argument that double negation elimination is valid. What brings about this defeat? There are in principle several options one has to support this, but Field endorses what seems like an especially radical option: assuming the alternative meets some minimal threshold for being reasonable, the proposal of such an alternative alone leads to defeat, even if no evidence or reasons are presented in favor of that alternative. To quote Field: “It needs to be emphasized that the old observational practice isn’t defeated by new evidence, it is defeated by the suggestion of a new theory” ((Field, 2020, 6), emphasis in the original).⁵ A less radical option would hold that defeat only happens once one has sufficient reasons to take one’s present belief formation to be flawed. It is unclear if either the nihilist challenge to our perceptual beliefs about objects or

⁴ Field in (Field, 2020) uses a different, and, I think, somewhat more complex example to illustrate this point: the debate about the heliocentric vs. geocentric worldview and rejecting the former on the basis of seeing objects fall in a straight line. Essentially the same lesson will apply to both examples. ⁵ See also (Field, 2020, 14) for the qualification that not any crazy alternative will work, and that which alternatives should be taken seriously can vary. I take it, though, that all the alternatives under discussion here will meet the threshold of being reasonable in the sense that they are worthy of being taken seriously.

the immanent stance 177 the intuitionist’s challenge to our use of double negation elimination live up to this stronger standard for defeat, but they clearly do live up to the weaker standard of being a suggestion of an alternative that is worthy of being considered seriously. And if such a suggestion alone defeats the rules that are being challenged, then I will have to retreat to neutral ground: I will have to reason using only rules that have not been challenged. I would then have to face the challenge the intuitionist puts forward without relying on double negation elimination and thus without my proof of its validity, which uses it. Those rules would have been defeated in the undercutting sense, or so Field.⁶ Although this might seem like a reasonable position to take at first, I think we can see that it is mistaken. In particular, to hold that we face defeat from the mere suggestion of a (minimally reasonable) alternative strikes me as the wrong reaction. It is true, of course, that the mentioning of an alternative that one had not considered has some legitimate epistemic effect. One should at least consider it, and maybe consider it very seriously. But to take it to simply defeat one’s perceptual beliefs or one’s conclusions of deductive reasoning is a complete overreaction. To take undercutting defeat to have occurred means that the perceptions or deductive arguments do no longer support these conclusions at all. If the mere proposal of an alternative worth taking seriously were enough for defeat, it would be easy to defeat most of our beliefs, no matter what they are based on. In fact, most would already have been defeated, given all the alternatives that have been mentioned in philosophical discussions: simulations, unreliable memories, hallucinations, etc. What would defeat a particular belief or belief forming mechanism is that one has good reason to think that the alternative obtains, and thus that the belief might not be true or the mechanism might not be reliable. That is what happens in the classical example of me relying on the testimony of someone, and then learning from someone else that the first person is a notorious liar. But mentioning an alternative alone, even a reasonable one, is not enough for defeat.⁷ But if my reliance on classical reasoning is not defeated by Iris’s challenge and her arguments that I am mistaken, then I continue to be entitled to use classical reasoning in my deliberations about whether to switch to intuitionistic logic.

⁶ Field is not alone in endorsing such a radical picture of defeat in philosophical debates. An endorsement of this epistemological stance in the debate about composition is found, amongst others in (Merricks, 2001), (Rosen and Dorr, 2002), and (Sider, 2013, 260). In particular, Sider endorses that merely the suggestion of nihilism defeats our entitlement to our perceptual beliefs in objects. I have defended the opposite view, which endorses that our entitlement to our perceptual beliefs in there being tables is not defeated by bringing up the nihilist alternative, in chapter 7 of (Hofweber, 2016b) and in (Hofweber, 2019a). ⁷ The issues from the last several paragraphs are discussed in more detail in (Hofweber, 2022b), where I also discuss whether there is partial defeat of our own logical rules in light of challenges. I argue that there is not, for the simply reason that I am already, preemptively, in a position to conclusively reject any attempt to defeat my own deductive rules, fully or partially, since I have available to me reasons that are conclusive by my own lights that my own rules are valid, and thus that the attempt to challenge them is misguided. This is related to the “dogmatism puzzle”, see (Lasonen-Aarnio, 2014).

178 idealism and the harmony of thought and reality When I try to settle for myself which concepts I should use, then I must attend to all the reasons available to me and weigh them properly according to the best standards I have for doing so. I must settle rationally, by my own lights, which concepts to employ, and so understood there is no path away from classical logic. One of the reasons available to me is that I can prove purely deductively from no premises that double negation elimination is valid. That reason will easily outweigh whatever considerations Iris might bring forward in favor of switching, and thus overall it is rationally required for me not to switch. I must stick to my own logic in light of a challenge. There is no rational path to go somewhere else. Of course, Iris will also think the same about her situation. She will also insist, correctly, that it is rationally required, by her lights, for her to think that she is right and I am wrong. And I should recognize that she is correct as far as it goes: by her own lights she indeed must stick to intuitionistic logic. Her own best standards for weighing the reasons available to her make it clear that intuitionistic logic is still the right one to employ, despite whatever is true about me. This situation can be seen as indicating that my own reliance on my deductive rules is getting defeated after all, in particular at a meta-level, so to speak, since both Iris and I can continue to insist on our own rules and we are each perfectly rational by our own lights to do so. That fact might in turn be seen as defeating our entitlement to those rules. But here, too, I must conclude that no such defeat happens. Although it is true that both Iris and I reason flawlessly by our own lights, and thus in this regard our situations are symmetric, I nonetheless should not rationally take this symmetry to defeat my entitlement to my own logic. I am in a position to reason conclusively that even though Iris is internally coherent, she is nonetheless mistaken in her insistence to use intuitionistic negation. I can conclusively prove this, and thus I must insist that Iris is mistaken. Therefore I should not conclude that neither one of us is entitled to their own deductive reasoning. Moving things to a higher level just reintroduces the same issues, and invites the same responses. Whether these responses were the right ones to begin with can be debated, of course, but moving one level higher does neither add nor subtract from them. We should accept that our own rules are not defeated in light of other reasoners that employ different ones. Iris and I are in a symmetric situation to an extent, but that fact does not undercut my entitlement to my own rules, and with it my entitlement to hold that I am right in employing classical negation and Iris is wrong in employing intuitionistic negation instead. To be clear, this rejection of Iris and me being complete equals relies on that I have a proof that I am right. This argument given so far is based on that I can establish conclusively that I am right. It wouldn’t necessarily work if I merely had some reason to think I was right, since then we have to weigh the strength of the reason for my thinking that I am right with the strength of the reason for thinking that I am not right, due to the apparent symmetry with Iris. In particular, the present position does not require that one takes a ‘steadfast’ approach to peer disagreement in general, only in cases where one establishes conclusively by one’s

the immanent stance 179 own lights that one’s present view is correct.⁸ But when I can prove that I am right, deductively and from no premises, then this is guaranteed not to be outweighed by other considerations. Overall, then, our entitlement to reason in accordance to our own logic is not defeated in light of other thinkers reasoning with a different logic. Thus for each of us it is rational by our own lights to accept the conclusions we can achieve this way. And since deduction from no premises establishes a conclusion decisively, this means that by my own lights I can establish the validity of my own logical rules conclusively. Thus by my own lights it is rationally required for me to think that Iris is wrong and I am right. And if I am right, then I am not trapped, but guided towards the truth by the requirements of rationality. But all this can be taken to show less about being trapped and more about what is rationally required for me to accept about being trapped. In general we need to distinguish between what is true and what is rationally required to believe to be true. The above considerations seem to conflate these. And, it can further be objected, they focus too much on what is rationally required to believe by my own lights. This notion of what is rational can be contrasted with a more external conception of rationality, where my own initial position plays less of a role. Maybe all this points to is that we used the wrong conception of what is rational to begin with. These two worries are not unrelated, and we will now look at them more carefully.

7.2 Immanent rationality and the asymmetry thesis Let us return to the question whether anyone can be rationally trapped, in particular whether Iris is trapped. In light of what we have seen above, the answer seems to be clear: Iris can’t rationally accept, by her own lights, that she has the wrong logic, even though she does, or so I must conclude. It is not possible for Iris, given her situation, to accept the validity of double negation elimination. Iris is trapped. And thus it is possible after all for someone to be rationally trapped. Iris can’t rationally get out of a bad place. But could it be that I am also rationally trapped in just this way? I can see that I might get myself into a trap if I made the mistake of switching to intuitionistic logic. If I were to switch, then I would from that moment on reason in accordance with intuitionistic logic and fully approve of such reasoning on reflection. I would then be in the same situation that Iris is in now, and thus I would then be unable to appreciate the validity of double negation elimination. Switching thus holds a unique threat: not only would I reason with the wrong logic, I would be unable to get myself out of this bad

⁸ See (Frances and Matheson, 2019) for an overview of the general debate about peer disagreement.

180 idealism and the harmony of thought and reality situation in the future. I would trap myself into accepting the wrong logic, with no rational way back to the correct logic. Iris is trapped this way, and so would I be if I switched. Thus being rationally trapped is possible for Iris as well as for myself. We have to recognize that others can be rationally trapped, and we have to accept that we ourselves could be trapped if we were sufficiently different than we are now. If we switched to intuitionistic logic and endorsed this switch, we would trap ourselves. That leaves open the question whether one could be rationally trapped in the situation one is presently in. To consider this question I want to clearly distinguish from the start whether I could ever rationally accept that I am rationally trapped right now, and whether it can be true that I am rationally trapped right now. To consider the former first, there is a sense in which one can accept it, and another one in which one can’t. There is nothing incoherent in the situation where I get misleading evidence that strongly supports a particular proposition p, even though p is false. The requirements of rationality would then put constraints on what I should believe in light of this, and they will have a strong pull on me to keep my false belief that p. But this is not a real trap in our sense. I could easily get out of it by getting even stronger non-misleading evidence that not p, and the requirements of rationality would then lead me towards the truth of not p. I can search out new evidence regarding p and finding the proper, non-misleading evidence will set me free. A true trap in our sense only occurs when I am rationally required to believe that p, and there is no rationally permitted way out of this requirement. Uncontroversially, rational forces can hinder me from getting at the truth, as in the case of misleading evidence. And I can rationally accept that this might be so in my present situation. But if the proposition that p can be established conclusively by my own lights, then this is different. To establish it conclusively is to establish it in a way that settles the issue. To prove something with purely deductive reasoning from no premises fits that bill. There can be no further question about whether this is indeed correct. I accept purely deductive reasoning from no premises as the strongest way to establish a conclusion: if it alone establishes the conclusion, then the issue is settled. Such reasons are stronger than even those from mathematical proofs, since they rely on deductive reasoning as well as some axioms as premises. If I can establish a conclusion merely on the basis of deductive reasoning alone, without relying on premises, then this is even more reason to accept it. And what’s good enough for mathematics should be good enough for philosophy as well. This means, in particular, that I cannot rationally accept that such reasoning might mislead me.⁹ ⁹ I should reiterate here that I am assuming, for the sake of the argument, a starting point where the thinker is fully behind their own deductive reasoning and logical rules: they accept them as the right ones on reflection, and they give the proper comparative strength to deductive reasoning over perception, conjecture, and so on. If someone reasons deductively, but does not fully trust their own reasoning, or weighs perception as more trustworthy than deduction, then this would, of course, be different.

the immanent stance 181 To accept that reasoning which conclusively settles a question might mislead me would lead to an internal incoherence. It would combine that, on the one hand, the issue is settled conclusively, but on the other hand, it remains open whether this is indeed so. And so I cannot accept that I am trapped when I reason this way. I cannot rationally accept that my own strongest belief forming mechanisms, which establishing a result conclusively, can lead me astray and trap me. So far then what I can rationally accept about myself being now rationally trapped. But could it be true nonetheless that I am now trapped, even if I can’t rationally accept this truth? The answer must be ‘no’. I can argue conclusively that I am not trapped: Not only am I required to accept that classical negation is the proper concept of negation to employ, but furthermore, it is the right concept of negation to employ. I can argue conclusively that this is the right concept of negation, in the outlined, rule-circular way. And so I conclude as I should: I am rationally required to employ the correct concept of negation, and thus I am not trapped. Therefore, not only am I required to accept that I am not trapped, it is true that I am not trapped. Of course, my reasoning here is rule circular, in that it used classical negation and the rules associated with it to establish that reasoning with classical negation is valid, and thus classical negation is the right concept to use. But as we saw in Chapter 6, rule-circular reasoning is not by itself illegitimate. And as we saw just above, my own deductive reasoning is not defeated in light of others reasoning differently. And so I continue to be entitled to reason just this way, and with this reasoning I can conclude that I am not trapped. This argument must seem suspicious, and we should dwell on it a bit more to properly appreciate it. The above considerations point to a real first-person third-person asymmetry, which we can label the asymmetry thesis: Other people can be in rational traps, but I can’t be in one right now. This thesis has a weak and innocent reading, as well as a strong and radical reading, related to our discussion from just above. I hope to defend it in the strong reading. On the more innocent reading the thesis simply concerns a difference in what is rational for me to believe, or to put it another way, a difference in total evidence: I can have total evidence that supports that Iris is trapped, but I can never have total evidence that supports that I am trapped now. So understood, it seems to point to an asymmetry in what is rational or an asymmetry in available evidence: what evidence is available to different people about their own situation or the situation of others and what is rationally required for them to think about the relevant issue in light of their evidence. On this weak reading of the asymmetry thesis, it merely concerns evidence and epistemology, but not necessarily the truth and metaphysics. On the strong reading it concerns not just what evidence we have, but what the thesis literally says: who is and who isn’t trapped, and thus who has the right concepts or beliefs. After all, one is only trapped if one is required to hold a false belief, not if one is required to hold a true belief. This is the reading I aim to defend.

182 idealism and the harmony of thought and reality My defense of my not being trapped must once more raise the suspicion of a confusion between what is rational to accept with what is true. Maybe one can argue that it is rationally required for me to accept that I am not now trapped, but that’s a long way from it being so that I am now not trapped. And in general that is correct, as acknowledged and discussed above: It is one thing for it being rational for me to believe that p, and another thing for it being true that p. I can have misleading evidence that strongly supports that p, while it is nonetheless false that p. And I can rationally accept that this might be my own situation: I might right now have strong evidence that p, while it is nonetheless false that p. But there is a crucial difference between what is rationally strongly suggested and what is conclusive and rationally required. Here such a gap with what is conclusive and what is true can’t be rationally accepted. If it is conclusive for me that p, then I cannot accept that it might not be true that p. It being conclusive rules out this possibility as one that is still open. Since I can establish by purely deductive argument from no premises that classical logic is valid, I can’t at the same time rationally accept that this might be false. When my reasoning is conclusive, as it is in this case, then there is no more room for thinking that the conclusion might nonetheless be false.1⁰ But now, how do we ever get from what is rationally required to what is true? How do I ever get from what is required of me to accept, or even from what is conclusively established, to what is indeed the case? After all, we know that what is required for other people to accept by their own lights can be false, as in the case of Iris. The key to appreciating this step in my own case is to take seriously the first-personal aspect of this whole enterprise: I wonder whether I am trapped. I raise this issue about myself and I hope to answer it to the best of my abilities. And as the question was asked, the answer is clear: it is rationally required for me to conclude that I am not trapped, and so I draw this conclusion. And since it is me who is writing this, I conclude as I should: I am not trapped, even though Iris is. This is how we get to truth from what is rationally required. It is again me who wonders what is true. I need to draw the conclusion that this is indeed true, and so I conclude that it is true. When I wonder what is required of Iris to hold, then such a transition would not be possible. It is not true in general that if someone is rationally required to believe that p, then p is true. It is not in general true that if someone can establish conclusively by their own lights that p, then p is true. 1⁰ This issue is related to, but significantly different from, the question about fallibilism about knowledge. The fallibilist about knowledge accepts that one can know that p even though one’s evidence doesn’t completely rule out not p. In other words, one can know something even though it isn’t established conclusively. This might well be so, but our issue here is whether something one can establish conclusively might still be false. Here we have to resist and deny it for the first-personal case. Deductive proof from no premises does establish a conclusion conclusively, since it is, by my own lights, not just the strongest and most decisive way to establish a conclusion, but a proper way to establish something conclusively.

the immanent stance 183 Iris is a counterexample to this. She is required to stick with her flawed concept of negation. But it is true for my own case when I wonder what is and isn’t true. If I wonder whether or not p and it is conclusive by my own lights that p, then I am entitled to conclude that p, and thus p. That ends my own inquiry into whether or not p. And it establishes that p, and thus that it is true that p. To properly make sense of this we once more need to focus on both the question and the answer, as we did when we looked at how the language-metaphysics gap can be bridged, in section 6.1 above. But now we need not just consider the concepts employed in the question but also the position of the person who asked it and who has to assess the answer. When I ask a question, I must accept the answer that is required for me to accept, and in this case there is no room for me to allow the possibility of it being false. And thus I can conclude that this is indeed the correct answer. Naturally, Iris would again say the same about herself, but once more, she would be wrong, or so I have conclusive reasons to conclude. All this should not raise doubts about my conclusions, and although it might seem self-centered or dogmatic, it simply accepts the conclusions that I have conclusive reasons to draw. But why would I be so lucky that it happens to be me who can’t be trapped? And why would now be so special that this is not the time when I can be trapped? It is hard to say how I got so lucky, but it is easy to show that I am that lucky. Although the apparent symmetry between Iris and me, as well as between my possible future intuitionist self and my present classical self, is reason to support that I might be trapped now, all things considered it is rationally required for me to hold that I am not trapped. And thus I hold that I am not trapped. Since I can argue conclusively that double negation elimination is valid, these reasons outweigh the ones that undermine my present position, and so overall the proper conclusion to draw is that I am not trapped. Of course, a feeling of dissatisfaction must surely remain by this. But if I weigh the feeling against the reasons I have to be satisfied, we should again side with the reasons in favor of us not being trapped. And since I am not trapped this shows that what I can establish conclusively is an insight: it is a correct conclusion about what the world is like. Naturally, you, the reader, should not simply defer to my position here. For you I might be just like Iris. You need to approach this issue in your own, first-personal way. But then all the same considerations apply to you when you do so. It is the first-personal aspect of this question and approach that matters, but obviously not my taking this first-personal stance. Although our focus in the above example was a disagreement about logic, a case can be made for similar conclusions tied to other inescapable concepts. We can see that those who have the starred concepts fact*, truth*, inquiry*, etc., are trapped, while we are not trapped. When we ask about what the right concepts are that should be employed in inquiry, the answer must be that it is the concepts of truth and fact, not truth* and fact*. When we ask whether we should engage in

184 idealism and the harmony of thought and reality inquiry or inquiry*, and answer must be that we should engage in inquiry. All the reasons we consider for switching point towards sticking with inquiry. When we ask whether we should listen to reasons or reasons*, we must conclude that reasons guide the way. And so on and so forth. Of course, those who start with the starred concepts will think the corresponding thing about their situation and they will be rationally compelled to stick with the starred concepts. But that is their problem: they are trapped, and can’t get out of their bad situation. They foolishly engage in inquiry*, not inquiry, and so they don’t get at the truth and whatever is going in their heads might not even have content, but only content*. The crucial difference here is that I am asking the question: should I switch logics or concepts? This question is implicitly or explicitly first-personal in several ways: it concerns me, and what I should do, it is stated using the concepts I happen to have, and it is not just asked by me but also directed at me. I need to answer this question to the best of my own abilities and determine whether it has been answered with a particular proposed answer. I need to state this answer, again using my own concepts, and the answer will once more concern me. The firstpersonal aspect of the question concerns thus not just its subject matter but also the concepts and the epistemic situation of the person who asked it. When I ask such a question I want an answer supported by reasons, not reasons*. And I aim for an answer which is true, not true*. But I don’t merely aim for the true answer. I hope to reach an answer that is both true and rational by my own lights to accept as the true answer. Only then will the issue be properly settled for me, since it will be settled by my own lights that this is indeed the right answer. And given that the question at hand is the one I asked at the outset, the whole setup is biased towards my concepts. If I would have asked whether I should* switch concepts, and what reasons* I might have for doing so, then I might well not be able to secure my own position as the right one. But that is not the question I asked. Should I have asked that question instead? This question, the one I just asked it in the prior sentence, is also one stated in my terms, asking about what I should do, not what I should* do. And as such, the answer will be that I should not have asked a different question. This was the right question to ask, and the answer that we should not switch concepts or logics was the right answer to give. But when I say that this asymmetry is first-personal, I don’t mean to suggest it is exclusively first-personal. It applies to me, but not just me. It also applies to all those like me: all those who accept classical logic and ask the same questions with the same concepts. And it contrasts all of us with those that are sufficiently different: those who accept a different logic and ask different questions with different concepts. It is rationally required of me, by my own lights, to stick with classical negation: the reasons available to me, properly evaluated by my own best methods and standards, leaves no other option. But the question remains whether evaluating concepts ‘by my own lights’ is the proper way of doing so. Why focus on what is

the immanent stance 185 rational by my own lights, and not just what is rational full stop? Why should it be rational by our own lights which concepts we employ? One might think that there is simply a fact about which concepts are best, and thus we should employ those, even if we have no access to why these concepts are best, and even if our own lights point elsewhere. Maybe it is the concepts God prefers us to use. Maybe there is simply a normative fact which demands that we use certain concepts. These external demands on us might be completely inaccessible to us: we might not know that they are there, and even if someone told us, we might not be in a position to appreciate why those are the proper concepts. Why does God demand those concepts, and not the ones that appear to us, from our present position, to be the best ones? On the external demand based approach to rationality, we should adopt the concepts God prefers even if we can’t appreciate why those concepts are better, and thus it would not be rational by our own lights to do so. We can grant that both notions of which concepts one should have make sense as notions of rationality or of ‘should’: First, there is one where I evaluate any answers to this question by my own lights: here I need to decide based on my best basic methods, properly applied. The picture here includes that some methods are basic, likely deduction, induction, perception, and so on, while other methods are derivative and more specialized, like measuring with a thermometer. The details of how this might go more precisely are complex and complicated, but to simplify I imagine that we employ certain basic methods, that we give some relative weight to them, as when we hold that deduction from no premisses is more certain than perception, and that we reflectively endorse the methods that we at present have, and how they are weighed. Real life examples will surely be more complex and less transparent, but I hope to be able to work with this picture to bring out the contrast. I should stress again that on this picture what is rational by one’s own lights is not the same as what seems to one to be rational. One can be mistaken about what one’s own best methods support in a particular case, as when one makes a mistake in a deductive proof. What is rational by one’s own lights is what is supported by one’s best methods when properly applied. This is the approach I have preferred here. On such a conception of ‘should’, it will make sense to us which concepts we should have and employ: It will be something that we can fully stand behind. Alternatively, and second, one might hold that which concepts I should employ is simply determined by some worldly facts, whether or not I can know about them or appreciate them as being a good reason for using those concepts. On this conception of ‘should’ or ‘rationality’, what concepts we should employ might not make sense to us. It might not make sense to us why we should employ the concepts that rationality so understood recommends. It is simply a fact, maybe a fact that makes sense to other creatures, but for us it would be a fact that we can’t properly appreciate as giving us a good reason to employ those concepts. To mark this distinction we should put labels on these two senses of ‘should’ and of ‘rationality’. It is tempting to use the internal-external metaphor for

186 idealism and the harmony of thought and reality them, but that one is certainly overused, including here in Chapters 3 and 4, and it is especially overused in epistemology. Instead of another internal-external distinction, I would like to call rationality tied to what is rational by my own lights immanent rationality, and rationality tied to independent, external standards transcendent rationality. Relatedly, we can distinguish the immanent ‘should’ from the transcendent ‘should’ as the one tied to immanent and transcendent rationality, respectively. Using an immanent-transcendent distinction rather than an internal-external one will hopefully also get further support when these notions of rationality are tied to other notions related to the immanent-transcendent distinction below. Which ‘should’ should guide us when we try to find out what concepts we should have? This might seem like a silly question, since it will simply depend on how the ‘should’ used in this last sentence is to be understood. If it is the immanent should, then probably that one, if not, then probably the other one. It thus seems the issue is just pushed back to the ‘should’ used in the original question. But whichever one it might be, we can consider the options and what difference it will make in the end. If we are guided by the should of transcendent rationality, then we end up with the concepts that the normative facts or God demand, but we might be alienated from these concepts: we might not be able to appreciate why it is those concepts and not some others that are the right ones to employ. If transcendent rationality effectively guides our concept choices, and thus has a causal effect on which concepts we have, then we might end up with concepts that we can’t appreciate as the right ones, alienating us from our own concepts. This alienating rationality might be good enough if all we hope for is to achieve a certain goal: having the concepts that the world or God demands. But we can hope for more. On the immanent conception of rationality, we won’t be alienated from the concepts we should have. We would be in a position to appreciate why those concepts are the right ones, since it would be rational by our own lights to have just those concepts. The question remains whether immanent rationality would guide us to the proper concepts and the true beliefs. Obviously, everyone aims at the truth and at the best concepts. The question is when we should think that we have gotten there. Since which concepts we employ concerns something we do, the question at least in this case come down to when we should stop doing something in this regard. Since doing something is generally under our direct control, it is hard to see how a transcendent notion of rationality can lead to a stable end to our attempt to do what we should do. If what we should do is alienated from us, then it is hard to see that we got where we should be. It is thus hard to see when we should stop trying to get there. Choosing which concepts to employ involves us doing something, and thus involves our agency. We won’t be done doing this until we have reached a stable place, one where we can appreciate that we have gotten

the immanent stance 187 where we should be. As long as we are actively involved in determining what to do, transcendent rationality alone leads to no stable resting place. To be sure, choosing which concepts to employ directly involves our agency in a way that belief does not: we don’t choose what to believe. But even in the case of belief there is a connection to agency, via the question whether one should investigate an issue further. Even if transcendent rationality is effective in producing beliefs in me that are guided by it, the question will remain how the appearance of these beliefs in my mind affects how I will structure my future inquiry. In particular, if the reasons for these beliefs are alienating to me, if I cannot appreciate them by my own lights as good reasons in its favor, then I will be inclined to pursue the issue further. The only stable resting place is a belief that is immanently rational for me to have. It is useful to highlight this consequence of transcendent rationality by imagining that causal and rational forces coincide. In that case the world is as it should be: I have the beliefs that rationality recommends, I change my beliefs as rationality demands, and so on. But if rationality in this scenario is transcendent rationality, then this would alienate us from our own minds, and it would overall not be a good thing. To illustrate how transcendent rationality leads to alienation, consider a view where rationality puts constraints on our beliefs, independently of what the thinker themselves approves of or endorses.11 On such a view the requirements of rationality prohibit certain combinations of beliefs, like a belief that p as well as a belief that not p, and similar simple combinations that are logically inconsistent. These requirements simply obtain, whether or not the believer endorses, approves, or appreciates them. Such requirements of rationality are closely tied to what the correct logic is: if it is classical logic, then certain combinations of states are forbidden. If it is a different logic, then these combinations of states might well be allowed. When someone is fully behind a non-classical logic, then by their own lights certain combinations of beliefs are perfectly permitted, even though transcendent rationality might rule them out. If such a person were to be guided by transcendent rationality, then they would be alienated from their own beliefs: by there own lights it is perfectly permissible to have a certain combination of belief states, but somehow such combinations are prohibited and ruled out in their minds. But it will make no sense to such a thinker why that is so. The demands of structural rationality will be alienating to them, since they restrict their minds in a way that makes no sense to them. If transcendent rationality were to effectively constrain our minds, then this would lead to our minds not making sense to ourselves. And thus these transcendent rational constraints would alienate us from our own minds, and would turn us into thinkers that we rightly would not want to be. 11 See (Broome, 2013, 91–3), (Worsnip, 2021), and many others for such views.

188 idealism and the harmony of thought and reality Thus as agents we should hope to be guided by immanent rationality. We do, of course, aim at the truth and at the right concepts. But only if the truth is also immanently rational for us, and only if the right concepts are the right ones by our own lights, can we be fully content. The difference will come down to a difference between believing the truth, on the one hand, and making sense of the truth, on the other. If we are successfully guided by what is transcendently rational, then we will end up with the truth and the right concepts, but we won’t be able to make sense of the world that we correctly represent. If we are successfully guided towards the truth by what is immanently rational, then we will not just represent the world correctly, but understand or make sense of it. As responsible epistemic agents, we must thus hope to be guided by what is immanently rational. Only that way can we claim a sense of ownership over our concepts and our beliefs, and only that way can we hope to make sense of the world. This line of thought is not dissimilar to some classic enlightenment ideas about being a proper, autonomous, moral agent. To be such an agent we must take responsibility for our own actions. And to do that we must be able to justify them with reasons that we ourselves have and can appreciate and properly evaluate. A responsible agent will not be guided by external norms that they themselves can’t justify. We cannot blindly rely on divine commands, unless we can appreciate the reasons behind those commands, but they being commanded of us is not enough, even if the demand comes from God. We must be guided by our own lights, by what is immanently morally required. And similarly for responsible epistemic agents: we must be guided by what is immanently rational for us.12 As far as I can tell, the significance of something like the immanent sense of rationality is not widely defended or even discussed in the vast contemporary literature on rationality, with the notable exception of Richard Foley. In particular, in (Foley, 1992) Foley defends what he calls ‘egocentric rationality’ and its connection to what he calls ‘responsible belief ’. I don’t agree with everything in (Foley, 1992), but I am certainly behind Foley’s motivation for the significance of egocentric rationality and the significance of being responsible for one’s own beliefs. Since we won’t be able to dwell on these epistemic issues any longer here, and since I don’t have a more detailed position to offer now, I would like to encourage anyone who would like to see more to consult Foley’s work for a more developed epistemological position which is at least congenial in spirit, although not always in detail. In the end, all this leads to a larger lesson about the significance of one’s own conceptual and epistemic position in philosophical inquiry. And this lesson is closely tied to two approaches to the target for philosophical inquiry which we should look at next to get a better sense of the overall picture.

12 A nice presentation of this motivation for the moral case is in chapter 1 of (Wolff, 1970).

the immanent stance 189

7.3 The immanent and the transcendent stance The issues discussed in the last section point to a larger and important difference in what the proper starting point of philosophical inquiry is, and how we should understand its target or goalpost. On one approach we first and foremost start with a particular question. To get started in inquiry is to ask a question, if only to oneself, as something one would like to find the answer to. And when we probe reality with a question, we are thereby employing our concepts, the ones we happen to have, with all their baggage and benefits. Such a question sets the goalpost for the subsequent inquiry: it determines what needs to be done. This goalpost is thus at first set using our concepts, the ones we happen to have. That is the target of the relevant project of inquiry, and that question, as stated, is what we need to make progress on. When we assess a proposed answer to this question we rely on our best ways of evaluating reasons for and against this answer being the right one. When we settle on an answer, we judge the reasons in favor of this answer to outweigh the ones against it. And if we are rational in doing this, then we weigh these reasons correctly. To be properly satisfied with an answer it must be rational for us by our own lights to accept that answer, which is to say that our proper weighting of the reasons in favor wins out over our weighing of the reasons against it. In the end, whatever answer we accept, it should be rational by our own lights to accept that answer; only then will it be a truly stable resting place for inquiry started with a particular question. It is no help to us to have the world or God demand a particular weighing of reasons, which is one that we can’t appreciate as the proper one. An answer that is satisfactory for us must be one that we can appreciate as the right one. Since which concepts we start with affects which concepts we use in the question in the first place, as well as in any proposed answer, and also what will be rational for us to conclude by our own lights, this dependence highlights the significance of one’s present position, conceptually and epistemically, in asking and answering philosophical and other questions. The goalpost we set with an initial question might not be the best one to set. It might well be that we recognize that one of the concepts used in the initial question was flawed, and with this insight we are able to set a new goalpost with a better question, and then aim to answer that question. Or we might recognize that there is an even better question articulated with different concepts. But such a moving of the goalpost must again be rational by our own lights. We are not blindly stuck with our initial question, but we must be able to appreciate moving the goalpost, by our own lights. This is the approach congenial to the position defended in this book. But not everyone will be happy with this. On an alternative way of thinking of inquiry, in particular in metaphysics, we don’t set the goalpost, the goalpost is already there. Reality itself is the goalpost. We don’t set the goalpost with our concepts, instead our concepts must match the goalpost in order for us to reach it. Reality is simply there and we need to describe it correctly. To do so we need

190 idealism and the harmony of thought and reality to come up with the concepts and the language best suited to do so. Our present concepts might be a starting point for facing reality, but better ones will likely be possible and hopefully be available in the future. There is no special role for our present concepts in setting the goalpost for philosophical inquiry. That goalpost is not a question formulated with our concepts, but reality itself. We thus don’t set the goalpost, it’s already there, waiting for us to reach it. Let us call the immanent stance towards a philosophical question the approach that takes our present position, conceptually and epistemically, seriously: it accepts that philosophy aims to answer certain questions, that these questions are asked with the concepts we now have, and that any proposed revisions to these questions, as well as any proposed answers to them, are properly evaluated by our own lights, using our best methods for doing so. We set the goalpost with a question and we have not reached it until it is immanently rational for us to accept the correct answer. This contrasts with the transcendent stance, which takes the goalpost to be already there: it is reality itself, not something articulated with one of our questions. On this approach, which concepts are the best ones to use is simply settled by the world, not by which ones we can rationally accept as the best ones. The transcendent stance thus evaluates alternative concepts more from the outside, so to speak, from the point of view of the world, not our present standpoint. This stance naturally includes taking transcendent rationality to be what should guide us in our selection of concepts and for our beliefs about what reality is like. The immanent stance, in contrast, is naturally connected to immanent rationality. It might seem unclear whether these two stances are really all that different. Doesn’t everyone aim at reality? And doesn’t everyone probe reality with questions that happen to be formulated with our concepts? And although that is indeed so, there is a real difference in how the two stances approach philosophical projects, and that difference can matter. On the transcendent stance one imagines the world being there, demanding certain concepts and a certain description formulated with those concepts. On the immanent stance we ask about what reality is like. But that way the target is set as reality, not reality* or something else. This difference might seem insignificant, since I above characterized the transcendent stance as trying to meet the demands of the world, and what else is the world than reality? But this is merely a result of my using my own concepts to characterize the transcendent stance. As I need to describe it, using my own concepts, the difference might seem small. But in spirit the difference is quite large. The believer in the transcendent stance takes there to be an external pull on what concepts to employ and what to say with them which incorporates the initial goalpost of our conceptual activities like inquiry.13 The believer in the immanent stance instead holds that we set the goalpost by asking questions, including questions about reality and about what concepts to use. I hope to illustrate this difference, and what effects it has on

13 Paradigm examples of someone taking the transcendent stance are (Sider, 2011) and (Eklund, 2017). I will, in particular, discuss Eklund’s position in the next section.

the immanent stance 191 actual philosophical debates, in more detail in the next section, when we discuss Matti Eklund’s account of choosing normative concepts. But first we should get a little clearer on the difference between the two stances, and how this distinction contrasts with other, more familiar ones. To adopt the transcendent stance is not the same as adopting a ‘God’s eye view’, and the difference between the immanent and the transcendent stance is not the same as the difference of describing the world from within or from nowhere.1⁴ In fact, those two issues are quite independent of each other. To adopt the God’s eye point of view is to hold that the world has a complete description from nowhere, which is to say that the totality of all non-perspectival facts or truths is the totality of all the facts or truths. To deny this is to hold that some truths are only expressible from within, say that the truth that it is now 10 am.1⁵ The difference between adopting the immanent or the transcendent stance is neutral with respect to whether the world has a complete description from nowhere, or only from within, or not at all. Instead, it concerns the goalpost of our own activity tied to inquiry, and whether it is set by us or already there, waiting for us to meet it. Even if that goalpost is “correctly represent reality as the totality of facts” that leaves open whether all the facts are non-perspectival facts, or whether some facts are essentially perspectival and thus a complete description from nowhere is impossible. Talk of a ‘stance’ might be misleading, as it suggests that one is at liberty to adopt one or the other, with both being equally legitimate and a matter of choice. But this is not intended with the terminology of an immanent and a transcendent stance. To speak of two stances is more closely tied to two different ways of how inquiry could be understood, two different general outlooks on inquiry. No terminology is perfect, I assume, and I hope nothing about both stances being equal is read into it. I certainly hope to defend that the two stances are not equal, with the immanent one being the correct one to take. There is a clear difference between the immanent and the transcendent stance regarding the significance of natural language and our present concepts. For the immanent stance, natural language is used for articulating the goalpost, and so the goal of metaphysics itself is expressed in our present concepts. Obviously, the goal can be refined as we go along, but any such proposed revisions must be recognizable by us, according to our own best standards for evaluating revisions, as an improvement: it must be immanently rational to move the goalpost. For the transcendent stance, where the goal is already there, natural language should have no special relation to the goal of metaphysics itself. There is little reason to think that we happen to have the right concepts already to even state the goal of metaphysics, and maybe also of inquiry properly understood. The transcendent stance will thus downplay the significance of investigating our own concepts and

1⁴ On the latter distinction, see (Nagel, 1986). 1⁵ This issue is closely related to our discussion of the subjectivity thesis and of fragmentalism in Chapter 2.

192 idealism and the harmony of thought and reality language, and if necessary retreat to ontologese, in the sense of (Sider, 2009). Here “ontologese” is a presumed language that is recommended by the world itself as more suitable for describing the world. In fact, for the transcendent stance, there is little reason to expect that natural languages as we now have them are the proper languages to reach the goalpost. The goalpost is not set by us, and so not tied to our representations. So why think we can meet it with the representations we presently have, or even with representations we human being can have in principle? The two stances furthermore disagree about the role of metaphysicians in metaphysics. For the immanent stance, we set the goalpost and we thus are intimately involved in what metaphysics is supposed to do, at least in the formulation of the goal itself. Metaphysics is understood as a human activity, something that involves us at the core. For the transcendent stance, the metaphysician is at best a mediator, someone who connects reality with some description of it in some language. But for the transcendent stance our present position is not in any way distinguished for settling disputes about which concepts are best or which positions are best supported. For the immanent stance, we evaluate alternative concepts from our present standpoint, using our present best methods. This difference will be crucial in just a minute, and hopefully better illuminated there as well. Of course, similar issues arise also for other parts of inquiry than metaphysics, but maybe with less urgency, since elsewhere there is little dispute about what the target of that domain is. I believe that the immanent stance is the one we should adopt. Immanent rationality should govern our choice of concepts and our beliefs, as I argued above. Only this way can we take full responsibility for our concepts and our beliefs. To take such responsibility for our concepts, our beliefs, and our projects, we must take inquiry to be governed by a goal that we ourselves have set, be it with a general question like what reality is like, or something more specific. Still, we set the goal, and we need to be in a position to properly appreciate that we have or have not reached it. It must be immanently rational for us to declare that we have reached the goal which we ourselves have set. And it must be immanently rational to modify or move the goalpost we set originally. And so we should adopt the immanent stance. And it is just this stance that is congenial to the position defended in this book, as I hope to spell out more clearly shortly. But first, to further clarify the difference between the immanent and transcendent stances, I would like to look at a particular discussion where this difference comes to the forefront.

7.4 Eklund on ought The difference between the immanent and the transcendent stance can be nicely highlighted by considering a problem Matti Eklund formulates in (Eklund, 2017, 4ff.) and discusses throughout that book. Eklund is concerned with the general

the immanent stance 193 problem of how to choose which concepts we should have, in particular, which thin normative concepts like ought or should we should have. To illustrate the problem, Eklund imagines another person, Bad Guy, who does bad things and also has different thin normative concepts: ought* and should*. Here, by assumption of the example, ought and ought* have the same “normative role”, but a different extensions. So, they play the same role in our respective minds, but what one ought to do and what one ought* to do can be different. Bad Guy reasons properly with his normative concept ought* and thus does what he ought* to do, but not what he ought to do. Obviously, Bad Guy is bad in the sense that he does things he is not supposed to do. But Eklund is concerned with how a realist about normativity is supposed to assess this situation. Would the world disapprove of Bad Guy’s behavior just as we would? As Eklund notes, this situation is quite symmetric: Bad Guy will think of us as bad* people, since we do things we ought* not to do. Both we and Bad Guy can recognize that we do what we ought to do, while Bad Guy does what he ought* to do. But if a proper form of realism about morality or normativity is correct, ardent realism in Eklund’s terminology, then one of us must be wrong somehow. It can’t be that both of us get the approval of the world, even though we both conform to what our respective normative concepts demand. As Eklund notes, this take on the situation gives rise to the threat of an inexpressibility problem: what is it that is at issue between Bad Guy and us? Not who is doing the right thing. Clearly we are. And not who is doing the right* thing. Clearly Bad Guy is. Bad Guy will insist that what is at issue is who is doing the right* thing, while we would insist that what is at issue is who is doing the right thing. This situation is again completely symmetric, and it is hard to justify how the world would pick one of these issues over the other. It seems, prima facie, that if realism is onto something, then there is a substantial issue here, but we seem to have a hard time stating what it is. In the remainder of his book, (Eklund, 2017), Eklund discusses various attempts to spell out what is at issue between us and Bad Guy, and with it what the question of ardent realism comes down to. Is it that some of the normative properties we and Bad Guy pick out are more natural than others, in the sense of Lewis in (Lewis, 1983) or Sider in (Sider, 2011)? Eklund finds this option wanting. Is it that the shared internal role which normative concepts have in the minds of us and of Bad Guy determine what property they express? If so, then the case of Bad Guy would be incoherent, since our normative concepts have the same role, but are supposed to express different properties. If normative role determines extension, then the realist can reject the challenge that Bad Guy seems to pose, and insist that what was at issue is what we ought to do after all, which in this case would be the same thing as what we ought* to do, assuming ought and ought* have the same internal normative role. This is the option Eklund takes to be more promising for a defense of realism, but as he discusses in his book, it faces several problems and leads to substantial further issues.

194 idealism and the harmony of thought and reality Eklund’s approach to this issue is a good example of someone taking the transcendent stance. Eklund hopes to find something about the world, something independent of our description of the situation, that can break the symmetry between us and Bad Guy. Maybe there is a difference in the properties that correspond to the normative concepts? Maybe this difference is a more objective one, like being more natural or glowing in some other way? Clearly Eklund has in mind that we must assess the difference between us and Bad Guy from a neutral point of view. We can’t just use our concepts to state the difference, since our concepts are biased towards us, and Bad Guy can symmetrically use his concepts to favor himself. The symmetry between Bad Guy and us needs to be broken by the world, objectively, and from a neutral point of view. That is paradigmatically what one demands from the transcendent stance. But if one adopts the immanent stance, this issue is resolved fairly quickly. On the immanent stance we take our own concepts and questions more seriously. As such the question is: who has the proper and better concepts, us or Bad Guy? Who is doing what they should do and has the concepts that they should have, us or Bad Guy? For all of those questions the answer is immediate: us, of course. Bad Guy does things he is not supposed to do, because he reasons with concepts he shouldn’t use. The fact that he reasons with concepts that lead him to do things he is not supposed to do only shows how messed up he is. Not only does he do bad things, he has bad concepts that allow him to justify to himself what he does. Bad Guy is stuck in his bad ways. He is like Iris, who is in such a bad position that she can’t get out of accepting the wrong logic. Bad Guy thus is not our equal, and the situation is only symmetric in a limited way. We both do things that our deliberations support, but we have the right concepts and reach the proper conclusions, while he has the wrong concepts that lead him astray. What is at issue in this debate is what concepts we should have. That is the goalpost for this limited part of inquiry. That goalpost is set this way, using those concepts, including the concept of should. And as such the answer is clear: we do have the normative concepts we should have, but Bad Guy doesn’t. Now, Eklund is particularly concerned with a form of realism about the normative, his ‘ardent realism’. What ardent realism comes down to is not completely clear, but it should be clear that the immanent stance by itself is in no conflict with realism. No form of relativism is implied by it, and neither is a form of nonfactualism or anti-realism. In fact, if we reformulate the issue as not being about which normative concepts are the right ones, but which ones are the objectively right ones, then the result will be the same. Not only should I use ‘should’, it is objectively so that I should use ‘should’: using ‘should*’ would mislead me, and objectively so. In light of this it is natural to conclude that there is no inexpressible question that articulates the real issue. The real issue can be stated simply as who has the better concepts and who does the right thing, or something near enough. Although Bad Guy is correct in thinking that we do the wrong* thing,

the immanent stance 195 that conclusion is insignificant, since that is not the issue at hand. The question is who does the right thing and who has the proper concepts, and the answer is that we do and he doesn’t. Of course, Eklund will be dissatisfied with this conclusion, since he will take it to miss what is truly at issue: There is a symmetry between us and Bad Guy and we need to evaluate who is in better shape from a neutral point of view, at least if we want to maintain a form of realism about normativity. But this focus on the symmetry and the insistence on a neutral arbiter of our and Bad Guy’s positions neglects how we got here in the first place. We started out trying to evaluate our own normative concepts and how meta-ethics usually focuses only on the ones we have, but not alternative ones we could have instead. We started out wondering whether we have the best normative concepts, and whether we might not want to have better ones. With that as our goalpost, the issue is clear, but the answer is fairly obvious. What Eklund hopes to motivate is that there is another issue, one about what breaks the symmetry between us and Bad Guy. And, of course, there are many other issues about us and Bad Guy. But why should we care about them? Why would I care about who has the better* concepts? Why should I accept that asking some other question beyond who has the better concepts leads to a better question? My own reaction is quite the opposite than Eklund’s. Instead of deferring to some inexpressible question and breaking the symmetry from neutral ground, we should accept that which concepts we should have was the right question to ask all along. Furthermore, we can see that it is the best question we can ask in this neighborhood of questions. Instead of demanding a neutral arbiter or some worldly feature that makes the difference neutrally put, we should accept that we have already asked the best question there is to ask here, and that we can easily answer it. All this is closely related to the argument discussed above in chapter 6 that the thin normative concepts of should and ought are inescapable concepts. We can’t rationally accept by our own lights that there are better concepts than those in this neighborhood. And we can’t rationally accept that we could be asking a better question than the question about which concepts we should have or what we should do. With that we should not hold that the real issue then must be something else, something not stated using our concepts. But we should accept instead that the real question has been stated and it has been answered. After all, we can’t rationally accept that any other question is a better one, and we answered the question we asked. And with that the issue is settled.

7.5 Immanent metaphysics We all have to start somewhere. We find ourselves having some concepts at the outset, due to various historical accidents about what concepts are the ones

196 idealism and the harmony of thought and reality presently employed, and due to the basic human cognitive setup, neither of which we have chosen or approved. And we find ourselves having certain methods for evaluating propositions expressed with the concepts we happen to have. These methods include how to respond in perception, how to respond to reasons, and also how to evaluate our very own methods and our own concepts. This should be uncontroversial so far, but there is a substantial difference in the neighborhood about what lesson to draw from this fact. We could have started with different concepts, and we could have had different methods of evaluation and different responses to evidence. I take deduction from no premises to give stronger support than perception, but other creatures can have a different initial setup where they favor perception instead. What significance does this possible variation in starting points have? There are two main reactions one could have to this: One that takes this variation to undermine our present position, and one that does not. On the more skeptical, undermining, approach, the fact that we could have had different concepts and different methods raises doubt about our own concepts and methods. On the most extreme side, my entitlement to my concepts and methods is undermined by this, unless I can show that these alternatives are inferior. That is, of course, extremely demanding. After all, how could I possibly show this, unless I am entitled to my concepts and methods in even considering the issue? Maybe this demand is legitimate but also unmeetable, and therefore our epistemic and representational situation is terrible. Weaker undermining positions are naturally more plausible. For example, Hartry Field’s position, discussed above, on what to do when one is confronted with an alternative logic. On that proposal, the presentation of a reasonable alternative undercuts my entitlement in my own contested deductive inferences. To properly earn that entitlement back, we need to retreat to neutral ground and successfully establish it from there, or so Field. I believe all this is a mistake. Our entitlement to our own belief forming mechanisms and methods for evaluation is not undercut by the presence of reasonable alternatives. We are defeasibly entitled to them, simply by having them at the outset. That entitlement can get defeated, but not merely by the suggestion of a reasonable alternative, only by an alternative that we have good reason to think is better or some other more substantial consideration. This stance to our own methods for evaluation and belief formation is nicely captured by the dogmatist position in epistemology, in the sense of (Pryor, 2000). Although Jim Pryor originally formulated dogmatism as a view about entitlement in perception, it quite clearly can be generalized to other basic modes of being entitled to one’s beliefs. Here the idea simply is that I am defeasibly entitled to the results of my basic belief forming mechanisms without having to first earn that entitlement somehow. The entitlement originally comes for free, but it is defeasible. What defeats it is debatable, of course, and it can range from something radical, like merely hearing about alternatives, to something more reasonable, like learning that one has bad eyesight. Reasonable as I am, I lean more towards the reasonable end

the immanent stance 197 of this spectrum of defeaters, and with it our own epistemic position is not that easily undermined, although, of course, it can be undermined. A more or less dogmatist line not only applies to entitlement in perception and deduction but also to the entitlement to employ our own concepts: At first we are entitled to employ the concepts we happen to have, defeasibly. We can learn that there are better concepts, and in light of that we can revise or change our concepts. But simply learning about alternative concepts and alternative conceptual setups or schemes is not enough to undercut our entitlement to our present concepts. Our starting point at first is fine, both conceptually and more narrowly epistemically. That original entitlement is defeasible and revisable, but at first it is there. And simply realizing that other people or other creatures start elsewhere does not defeat our own entitlement that we originally have. Any such defeat must come from an argument or reasons that our methods or our concepts are somehow flawed. That we could have had other ones is not enough for defeat. This broadly dogmatist position in epistemology is, of course, not uncontroversial, and this not the place to try to resolve that controversy.1⁶ Dogmatism is one of the standard options in epistemology, it seems to me to be the right one, and it naturally pairs with the immanent stance and the acceptance of inescapable concepts as insights, both of which are congenial to the defense of idealism pursued in this book. If I am entitled to my concepts and my present ways of evaluating them, and if these are not easily defeated by the mere presence of alternatives, then I am entitled to apply my inescapable concepts and to holding that they are inescapable. Similarly, the fact that other people can be trapped by their inescapable concepts and their belief forming mechanisms does not undermine my own entitlement to my inescapable concepts leading to insights. The asymmetry thesis, discussed above on page 181, is on stable ground in such a situation. Others can be trapped, but I am not trapped now. I am entitled to ask questions with the concepts I presently have, and I need to evaluate proposed alternative questions from my own present standpoint. This supports that questions that essentially involve inescapable concepts are indeed the right questions to ask, and we can see this without first knowing what reality is like in other ways. All of this supports that our answers to such questions are indeed deep results, in the sense of page 147. They not only answered the question we happened to have asked, they answered the right question to ask. And doing that is all one could hope for. All this leads to a more general approach towards making progress in metaphysics. This approach aims to achieve results from thinking about our own concepts and that certain concepts are inescapable. It is based on taking the immanent stance towards metaphysical questions as well as on insisting that any changes to the goalpost we set with our questions must make sense by our own lights.

1⁶ For some criticism, see (White, 2006).

198 idealism and the harmony of thought and reality I will call metaphysics pursued in this spirit immanent metaphysics. Immanent metaphysics works out the insights into what reality must be like that can be achieved from thinking about our own representations and their inescapability. It is a way of achieving results in metaphysics, and as such it is to be understood as a part of metaphysics. Immanent metaphysics in general aims to work out the metaphysical consequences that can be drawn from our inescapable concepts and their being inescapable. Not all of metaphysics is immanent metaphysics, but whatever results can be achieved in immanent metaphysics, it leads to real insights into what reality is and must be like. Immanent metaphysics is based on at least four related ideas: First, that one can sometimes draw conclusions about reality from our representations of reality alone. Second, a broadly dogmatist stance towards our defeasible entitlement to employing our own concepts. Third, the immanent stance towards how the goalpost of metaphysics is set by our own questions, articulated with our own concepts. And, fourth, the insistence that any revisions of our concepts must be immanently rational, that is, rational by our own lights, and thus inescapable concepts are indeed the right ones to have. Before we move on I would like to compare and contrast immanent metaphysics with some other approaches to metaphysics that it might be false associated it with. First, immanent metaphysics is distinctly different from descriptive metaphysics in Strawson’s sense.1⁷ Strawson famously distinguished descriptive from revisionary metaphysics, with the former concerning our present conceptual scheme, while the later concerns proposed improvements to our present conceptual scheme. For both the focus is on the concepts we have or should have, but not otherwise on what reality is like. Reflecting on our present and possibly improved future concepts is useful and important, but the goal of immanent metaphysics is not our concepts, but reality. Immanent metaphysics aims to find out about reality via a special strategy connected to our concepts, but the target is reality, not our concepts. Descriptive and revisionary metaphysics, in Strawson’s terminology, concerns our conceptual scheme: how it is and how it might be changed. Immanent metaphysics should also be contrasted with what is often called ‘natural language metaphysics’, or also ‘natural language ontology’.1⁸ Natural language metaphysics aims to work out the metaphysical commitments contained in our natural languages: what kinds of entities they aim to refer to, what distinctions we make in natural language, like those between objects and events, objects and properties, and so on. Natural language metaphysics is merely about what kind of a metaphysical picture is contained in our conceptual representations. It by itself does not aim to show that this picture is accurate. It merely hopes to work out what that picture is, and thus it is concerned with how we represent the world, not directly with how the world otherwise is. Natural language metaphysics is, of

1⁷ See (Strawson, 1959).

1⁸ See (Bach, 1986) and (Moltmann, 2017).

the immanent stance 199 course, a great and useful project. But it needs to be distinguished from immanent metaphysics. Immanent metaphysics aims to find out about what reality is like, and it does so via the presently neglected route of thinking about our own concepts and their inescapability. Its aim is not our representational system, but reality. Our concepts are a target of immanent metaphysics only since a special subclass of them can be a guide to what reality is like. Natural language metaphysics is compatible with our representational system being deeply inadequate to describe reality. Our representations might contain certain distinctions, but these distinctions might not correspond to anything in reality. Immanent metaphysics, on the other hand, is all about reality and what it is like. It achieves insights into what reality is like via reflection on our concepts and their inescapability, but the target of immanent metaphysics is reality, not merely how we represent it to be. Finally, immanent metaphysics needs to be distinguished from what Kit Fine has labeled ‘naive metaphysics’. For Fine, naive metaphysics concerns simply what is the case, whereas foundational metaphysics concerns what is the case in reality. The ultimate goal of metaphysics is what foundational metaphysics aims for, but to get there we first need to pursue naive metaphysics, at least to some extent. Fine’s distinction between naive and foundational metaphysics is based on his distinction between what is the case and what is the case in reality. That distinction might seem quite problematic, and I think rightly so. After all, what is it supposed to mean that something is the case, but it isn’t the case in reality? Normally the contrast between being the case in reality is being the case in a story or a fantasy, not with merely being the case. Fine holds that the relevant notion of reality that is at work in distinguishing naive from foundational metaphysics is not merely the ordinary notion of reality that contrasts with fiction, but a distinctly metaphysical one. To make that clear I prefer to write it with a capital ‘R’ as ‘Reality’. We might be unable to spell this notion out any further, according to Fine, but we nonetheless need to rely on it in metaphysics, in particular also in the statement of metaphysical thesis like realism in general,1⁹ and realism about tense in the philosophy of time in particular.2⁰ I have argued in more detail elsewhere that we should not accept Fine’s notion of Reality into metaphysics unless we can spell it out,21 and consequently we should not accept Fine’s distinction between naive and foundational metaphysics unless we can spell out the notion of Reality. The problem is that if we accept this notion as a primitive metaphysical notion, then we face the problem of what content and what value the question ‘What is the case in Reality?’ has. If the notion of Reality is a primitive and novel concept, then what content does it have? And maybe more importantly, why should we think that asking about what is the case in Reality is a better question than asking about what is the case? How can we, from our present position, appreciate that this question 1⁹ See (Fine, 2001). 2⁰ See (Fine, 2005). 21 See (Hofweber, 2009) and, in particular, chapter 13 of (Hofweber, 2016b).

200 idealism and the harmony of thought and reality is better than simply asking about what is the case, as we did all along? In earlier work I used the label esoteric metaphysics for approaches to metaphysics that load the questions that are supposed to be distinctly metaphysical questions with novel, primitive metaphysical notions like that of Reality.22 Esoteric metaphysics is to be rejected, I argued, since we can’t justify why these questions are better questions than the ones we asked all along. This rejection of esoteric metaphysics is closely tied to adopting the immanent stance. The questions we ask that define what we should do in metaphysics must be statable by us, using our present concepts. If someone proposes that we should ask a different question instead, say about what is the case in Reality, then that question must be a better question to ask by our own lights. We, in our present position, must be able to appreciate that this question is better. But with Fine’s notion of Reality, I do not see a way to do this. Thus we should continue to accept that metaphysics concerns what is the case, not what is the case in Reality. Of course, metaphysics does not concern everything that is the case, but whatever the questions are that metaphysics is supposed to answer, they concern what is the case. Immanent metaphysics does just that, of course. It tries to find out about reality, not Reality, and with it it tries to find out what is the case, not what is the case in Reality. This, then, is what I propose: we take our own present position seriously, and set the goalpost for our own activity of inquiry with an initial question. We then proceed from there, revising both the goalpost itself as well as the concepts used in articulating the goalpost, in a way that makes sense by our own lights. We thereby aspire to be guided by the immanent rationality of any changes in the goalpost and in our concepts, as well the immanent rationality of accepting that we reached the goalpost we set. Our original entitlement to employing our concepts places this project on solid footing, and the fact that this entitlement is not defeated by the availability of alternative starting points puts it on even more solid footing. In this position we can now recognize that certain concepts are such that it would never be immanently rational to replace them with an alternative. And this then points to that any conclusions we can reach by reflecting on these concepts is secure in a way that other conclusions are not. Not only are these conclusions results, they are deep results in the sense from Chapter 6: they answer not just the question we asked, but the right question to ask here. And thus inescapable concepts lead to insights and aren’t traps. Immanent metaphysics is the project of trying to get all the results that one can get this way. It is one, but only one, way of finding out what reality is and must be like. It not only demarcates the limits of our own rational thought about reality, but gains insight into reality itself. The main argument for idealism was one conclusion of this kind, and other conclusions surely can be reached this way as well, as we will discuss momentarily in Chapter 8.

22 See (Hofweber, 2009) and (Hofweber, 2016b). See also (Dasgupta, 2018).

the immanent stance 201 Immanent metaphysics has several connections to philosophical movements of the past, both directly and in spirit. One is to Kant’s theoretical philosophy, and that connection is worked out in more detail in Chapter 8 just below. Another is to some ideas from the enlightenment and also from traditional existentialism. The enlightenment idea of an autonomous and responsible moral agent, who does not accept any external authority that can’t be justified to themselves by themselves is mirrored in the proposed focus on immanent rationality and how we must take responsibility for our own concepts and our own beliefs. A responsible epistemic agent should aim to be governed by immanent rationality, so that they are not alienated from their own minds and so that they can make sense of themselves and the world. And immanent metaphysics chimes with the existentialist theme that we find ourselves in the world with certain concepts and basic belief forming mechanisms, and that we need to create value and meaning from this starting point. In our case, that means that the world does not come pre-furnished with a project that we need to carry out, and it does not come with a pre-established goalpost we need to meet. We are the source of the goals of our own activity of inquiry with our initial question, and we are responsible for the concepts we employ in setting that goal and in aiming to meet it. The world only determines whether we met the goal that we set for ourselves.

8 Conceptual idealism: the overall picture 8.1 The Big Question revisited We started out by asking the Big Question about the place of human beings in the world, and whether we have a special and distinguished standing in reality. Here we focused on a metaphysically special place for human beings, in particular whether human beings are special, since there is a close connection between our minds and reality itself. This led to the possibility of a third, idealist answer to the Big Question about our place: if a strong form of idealism were true, then we would be special in reality, not via a connection to God, but directly. A defense of conceptual idealism is one way to establish such a strong form of idealism, and with it a positive answer to the Big Question. If conceptual idealism is true, then this would not just answer the Big Question, it would also lead to a new metaphysical picture of what the world is like. And if the defense of conceptual idealism given above is correct, then this would lead to a new way to achieve results in metaphysics. By now we have all the pieces in place to put the overall philosophical picture tied to conceptual idealism together. In this chapter I hope to do just that. In order to properly put these pieces together we will need to revisit a few issues that were postponed earlier, fill in promissory notes, contrast conceptual idealism with other views, and look ahead. This will include articulating not just the overall metaphysical view that comes with conceptual idealism but also the approach to metaphysics that crystalized out of the defense of conceptual idealism. To start with all this, we should return to the beginning and reconsider the Big Question introduced in Chapter 1, and how conceptual idealism answers it. The Big Question about our place in the world concerned whether we are special and significant in the world as a whole. Besides the widely discussed negative naturalistic and positive theistic answers, there was also the possibility of a third, idealist answer. According to this idealist answer, we are significant and metaphysically central in the world, not via a connection to God, but directly, via an intimate connection between our minds and reality itself. Not just any version of idealism would answer the Big Question. Only strong idealism can do that, which is to say, a version of idealism that holds that our human minds are metaphysically central to reality. In the main part of this book I have defended just such a strong version of idealism. This includes formulating a version of idealism with the notion of harmony: our human minds and reality as the totality of facts are a perfect match for each other, since every fact that obtains can be represented in our thought. This Idealism and the Harmony of Thought and Reality. Thomas Hofweber, Oxford University Press. © Thomas Hofweber 2023. DOI: 10.1093/oso/9780198823636.003.0008

204 idealism and the harmony of thought and reality harmony does not just obtain by accident, but it obtains for a reason. Internalism about our own talk about facts shows that harmony must obtain, since it explains why all facts must be representable by us. Thus the totality of facts is limited by our minds: any fact must fit our minds in the sense that we must be able to represent them. This supports strong idealism, since not just any minds, but our human minds, have this special place in reality as the totality of facts. But does this answer the Big Question as it was intended? Here there are two relevant considerations: first, does it show that we are metaphysically central enough to warrant a positive answer to the Big Question, and second, does it properly support that we human beings are central in this way. I believe the answer is ‘yes’ to both, but let us consider them in turn. It is not sufficient for us being significant in reality if we are merely the best or the worst at something. Just as the tallest volcano by itself is not properly significant in reality, even though it is the tallest one, so being the best at music or at football is not sufficient for being significant in reality. What would be sufficient instead is some place in the overall story of reality that concerns its large-scale and most general features. We need to appear in the main text of the story of reality, and not merely get mentioned in a footnote. If there is some close connection between us human beings and some of the key large-scale features of reality, then this would have to appear in the main text, and thus it would secure our place in the world. If conceptual idealism is indeed correct, and harmony between our thoughts and reality indeed has to obtain, then this does concern the large-scale features of the world. It concerns the limits of reality itself: reality can’t go beyond what is representable in our thoughts. This is not a mere accident, but a restriction that obtains for a reason. The story of reality as a whole should not just include the facts that obtain but also how and why the totality of those facts is limited in a certain way. If this is not mentioned in the story of reality, then something of significance is left out. And so we human beings have a significant place in the overall story of reality, since we limit what reality is and can be like. We are special in the world because of an intimate connection between our minds and reality itself. And with this we have a positive answer to the Big Question. We are thus special in reality. But who more precisely is this ‘we’? I took it to be human beings above, but this can be challenged from two directions: from below, in that not all human beings are special, and from above, in that more than just human beings are special. To start with the first, consider an extreme case of arguing that not all of humanity is special, but only I am. To motivate this, consider that when I talk about all facts, I inferentially relate the quantified sentence to the instances. But why would I consider instances other than my own instances, the ones that are available to me, in my present situation and my present language? If so, then why isn’t it me personally, and not human beings in general, who are metaphysically special and who constrain reality? It would seem that the totality of facts is limited by my mind, not the human mind in general.

conceptual idealism: the overall picture 205 To clarify this situation, it is important to point out that being special does not mean being uniquely special. I can be special in reality, but so can be other people. Of course, not everything can be equally special, that would indeed undermine it. But uniqueness is not required for being special, only a certain kind of distinguished-ness. In this sense it can be true that I am special, but so are you, as well as the rest of humanity. This also applies looking upwards: if there are other creatures that are not biologically human beings, but have just the same minds as we have, then they, too, would be special in just the way we are. Human biology does not matter for this, only having a mind like ours. So, human beings might be special, but they might not be uniquely special. However, no other creatures can be more special in just the way we are special. Since our minds exactly limit reality, no other creatures can do better in this regard. We human beings might not be unique in how we are special, but we are as special in this regard as anyone can be. And not everything is special in this way: things that don’t have a mind are not special in this way, and things that have a representationally weaker mind aren’t special in this way either. Our specialness might not be unique, but it is also not universal. And with that we are distinguished in the world. However, another way of motivating that I am personally uniquely special is to focus on which instances of the inferential quantifiers are relevant when one quantifies over all facts. Why not restrict it to only instances of the speakers idiolect? If so, wouldn’t it show that only the speaker is special? If we allow the instances of English, wouldn’t that show that English speakers are special, and not all humans? Here the question of the semantic variability of human languages becomes central again. As we discussed above, starting on page 131, there clearly is a difference in what can be represented in different languages: with a very finegrained notion of sameness of content there will be differences in which contents can be expressed in which language. But such notions are generally too fine grained for our purposes. That human languages have the same expressive strength even with a proper, coarser notion of sameness of content has some empirical support in the fact that no clear counterexamples are generally accepted, but it is not obvious and could be false. But even if it is false, the metaphysical centrality of human beings would not be in question. Not all languages might have reached the full potential of what can be conceptually represented by a human mind. Some languages might still miss a piece or two, but, uncontroversially, all human languages can be learned by all human beings, no matter what their mother tongue or location of birth, and so there is uniformity in what can be represented in the human mind in general. If there is variation among human languages in what can in principle be represented in them, then we can only make sense of this downwards. It is possible that other human languages express less than the one we are presently speaking, English, but it is not possible that other languages express more. Here the situation would be just like in our discussion of powerful aliens in section 5.4. No other creatures can

206 idealism and the harmony of thought and reality express more than we can, since every fact is effable, by us. Other creatures could express less with a particular language if their language is like diminished English from section 5.4. But all human beings have the capacity to represent all the facts, since all of them can learn a language that allows them to represent all of them. The limits of reality thus coincide with the limits of human thought and language. And this vindicates the special place of human beings in the world. Of course, we need to qualify this and acknowledge that not strictly all human beings can represent the same things. Some individual human beings might not be able to have conceptual thoughts at all, due to being in a coma or having brain damage. But this is beside the point. I am not claiming that each human being can represent every fact, but rather that the human mind, and with it humanity, can do this. The focus is on humanity in general, not individual human beings in particular. Of course, humanity is closely tied to individual human beings, but for the human mind having a certain ability does not require everyone who is human to have it. Similarly, if there were other creatures with minds like ours, but who are biologically different and thus not biologically human beings, then they would be just as special as we are. It is not our biology that matters for our metaphysically distinguished status, but our mind. But not just any mind gives one the status that we have: it must be a mind just like ours, with the same conceptual capacities. Thus overall then I conclude that if conceptual idealism is correct, then human beings are metaphysically special in reality. They might not be uniquely special; other creatures can be just as special, if these other creatures have a mind just like ours. And not every human being has to have the same mental capacity for human beings collectively being metaphysically special. It is not merely the similarities among our individual minds, but our collective ability for conceptual thought, something we only achieve together to its full extent, that makes the differences. And thus not only are we sufficiently special in reality overall to justify a positive answer to the Big Question, it is indeed us human beings that are special in this way. Wrapping up with a discussion of the Big Question after Chapter 5 might have been a good place to end this book. At that point conceptual idealism was fully defended, and with it was the positive answer to the Big Question. I am sure some will think I should have stopped there. But even if the arguments for conceptual idealism and the third answer to the Big Question are all correct, some dissatisfaction must have remained at that point. The last third of this book aimed at wrapping up the position into an overall satisfactory package. The source of dissatisfaction is closely tied to the fact that the defense of idealism given essentially came from considerations merely about our own language. But how could we conclude from thinking about our own language alone that our minds constrain the world, and that a harmony between our thought and reality obtains? It seems that we need to go beyond our language and investigate the world as well as

conceptual idealism: the overall picture 207 our language and how the two compare to draw such conclusions. The discussion in Chapters 6 and 7 was not merely aimed at removing this dissatisfaction with the conclusion but also to focus on the positive side and to take the position to the next level. If it is really true that we can draw substantial conclusions like conceptual idealism simply by thinking about our own language, then this surely is of significance by itself. It must be tied to a larger picture of how to make progress in metaphysics, one that goes beyond the question of idealism and of our place in the world. I hope to work this positive vision out more in this final chapter, in particular in section 8.4 below. But first, let’s briefly review how we were able to draw such conclusions merely from thinking about our own language. One way to see how such a result is possible can come into view by focusing on the question instead of the answer. The answer, understood simply as a sentence in natural language, seems isolated by itself. Why should we think it matches the world if we simply think about the semantic function of the words in it, and it isn’t an analytic truth? But by focusing on the question we can see how it sets the target or goalpost for an answer. The question involves our concepts and so it becomes possible that via reflecting on the concepts or phrases used in the question we can see what the answer must be. And that can be so even if the answer itself is not a conceptual truth. It can be that by investigating largely empirical aspects of our language we can see that a certain question must have a certain answer. And this, I have argued, is the case with the question whether or not all facts are effable. All that then shifts the weight onto the question. Sure, the question as asked was answered that way, and so in letter the issue is resolved. But did we resolve the issue also in spirit? To find a truly satisfactory answer we can’t just be content with answering the question we have asked. We should go further and also determine that we answered not just any question, but the right question. And with this being left open one might feel that the Big Question hasn’t been fully answered yet. In Chapters 6 and 7 I have tried to take on this issue and argued that the question we originally asked, about our place in reality, was indeed the right question to ask, and we can see this without otherwise knowing what reality is like. When we reflect on whether we should replace the concept of a fact with an alternative, fact*, then we can see that this would always be irrational for us to do by our own lights. The concept of a fact is inescapable for us, and with it is the concept of reality as the totality of facts. We can’t rationally accept that there is a better question in the neighborhood than the question about our place in reality. This supports that conceptual idealism is a deep result, in the sense that not only is it the answer to the question we happened to have asked, but the answer to the right question to ask. This might also have been a good place to end, but it still leaves one question unanswered: since other thinkers can have different inescapable concepts, why think that our inescapable concepts lead to insight into what the world is like? Even if we are rationally required to never give up on that concept, why should we think that this is anything but a trap of rationality? I argued in Chapter 7 that even though

208 idealism and the harmony of thought and reality it can be that other people are trapped, it is not possible that I myself am trapped by rationality right now. This asymmetry thesis is in a sense the result of taking our present standpoint seriously: we are defeasibly entitled to apply the concepts we have, and this entitlement is not that easily defeated, in particular not by the mere presence of alternatives. Furthermore, we set the goalpost of metaphysical inquiry with an initial question, which sets the target of what such inquiry should achieve. And we need to evaluate any attempt to reach that goalpost, or to move it by asking a different question with different concepts, by our own lights. These three parts combined into a somewhat existentialist approach to metaphysics. It aims for authenticity in taking responsibility for settling questions oneself in ways that oneself can appreciate, while also accepting that we simply find ourselves at a certain starting point, with concepts and methods that we didn’t choose, but which we must use to determine our own goals and projects. Taking our own standpoint seriously this way supports that we are not trapped, but that we managed to answer not just the question we asked, and furthermore the question we should have asked. We can thus see that conceptual idealism is not merely true, but furthermore a deep result. And that settles the issue. This then really and finally wraps up the defense of conceptual idealism and with it the positive answer to the Big Question. To properly pull it off, we need to combine several things: some considerations in the philosophy of language about our own talk about facts, some considerations in epistemology about our entitlement to concepts and the rationality of replacing them, some straightforward philosophical considerations, and some more meta-philosophical ones about the significance of our present standpoint and of questions as setting the goalpost of metaphysics. It is only the combination of all of them which leads to an ultimately satisfactory answer. I thus hope that adding the extra complexity to the story is justified. The defense of idealism itself does not depend on any of the issues discussed in Chapters 6 and 7. Only the significance of this result depends on it. The question about our place as asked was answered by the end of Chapter 5. What remained is how significant and proper that question was. Chapters 6 and 7 hoped to show that the original question was indeed the right question to ask, and that its answer is not just a result, but a deep result. But does the defense of conceptual idealism live up to the standards we set at the beginning? On page 8 we discussed some constraints that any reasonable defense of idealism has to meet. It is now time to revisit these constraints and to see whether conceptual idealism indeed meets them.

8.2 How the constraints are met Any defense of idealism must meet some minimal constraints to be worthy of being taken seriously. In section 1.3 I insisted on three such constraints. If conceptual

conceptual idealism: the overall picture 209 idealism is to be taken seriously, then it must meet these three constraints. First, it must meet the explicitness constraint: the idealist position must be formulated in an explicit way, without novel, primitive notions of dependence. Second, it must meet the support constraint: the idealist position must be supported with some argument or some considerations that speak in its favor. Third, it must meet the compatibility constraint: the idealist position must be compatible with what we generally know to be true. In this section I hope to make clear that, and how, conceptual idealism meets all three of these constraints. Conceptual idealism was the view that the totality of facts range-depends on us. Range-dependence of the facts on our minds occurs when our minds limit which facts can obtain. On the present way of defending range-dependence, which facts can obtain is tied to our minds, since every fact must be representable with our concepts, not just by accident, but for a reason. Every fact must be representable, since the semantics of ‘every fact’ goes nicely with the proper notion of being representable, or alternatively, being effable. None of the notions used in the characterization of conceptual idealism are novel primitive notions, nor are they used in a specially dubious way or with a distinctly different meaning. Similarly, although the notion of range-dependence is a novel notion of dependence, it is not a primitive novel notion. Instead, it was spelled out as that which facts can obtain is tied to us, and this was in particular so since the fact that all facts are effable holds for a reason that involves us. This way the notion is perfectly intelligible. This contrasts with other approaches that hold that language is the transcendental precondition of the world, or the like, where this notion of a transcendental precondition is not spelled out in further, more intelligible terms.1 Conceptual idealism thus meets the explicitness constraint. Along with it, it also solves the dependence problem, that is, the problem of spelling out any notion of dependence used in a defense of idealism. The relevant notion of dependence relied upon in the formulation of conceptual idealism, range-dependence, is spelled out in perfectly accessible terms. It should be at least a little bit puzzling how any version of strong idealism could meet the support constraint. What reason could we possibly give that human beings are metaphysically central to reality? Maybe unperceived matter is incoherent, or objects without a mind interacting with them are impossible. Dubious as such arguments might be, they still wouldn’t be enough. These arguments would only support the centrality of minds more generally, but not that of our human minds in particular. Why would it have to be human minds that do what these arguments demand must be done? But the argument given above for conceptual idealism was of a rather different nature. The key to the defense of conceptual idealism was the defense of internalism about our talk about facts. Internalism

1 For example, (Gaskin, 2020) relies on such a notion without spelling it out more explicitly.

210 idealism and the harmony of thought and reality is a position about our own language. It was defended with considerations of the kind that are generally used to defend claims about language: broadly empirical considerations about the syntax and semantics of quantifiers, fact-terms, and so on. Such considerations are not especially puzzling, and it is no surprise how they can support internalism. The more surprising part was to show how internalism can support idealism. This was carried out with a fairly straightforward argument, which was given in Chapter 3 and developed further in Chapter 5. The semantics of our own quantifiers over facts shows that all facts are effable by us human beings, and thus there are no facts which are ineffable for us. The totality of facts is tied to what human beings can in principle represent conceptually. Thus considerations about our own language allowed us to conclude that our own minds are metaphysically central to reality. This argument was based on broadly empirical considerations: the kind of considerations also employed in linguistics, concerning what our languages are in fact like. These are not purely metaphysical arguments, but involve empirical and contingent considerations about our own language. There is no deep puzzle about how we can find out about our own language, but there is a real question about how any such considerations about language could possibly show anything about what reality is like. This question was the topic of Chapter 6, and I hope that question was answered there satisfactorily. All these considerations together show that the support constraint is met: it is these broadly empirical considerations about our own language, together with the more directly philosophical considerations concerning the metaphysical significance of the findings about our own language that support conceptual idealism. Finally, we need to show how the compatibility constraint is met. This is in a sense the easiest one of the three for the present version of idealism, although for other versions it is often the hardest. All that conceptual idealism requires is that any fact that obtains has to be representable by us. It makes no further claim about which facts obtain, and thus it takes on no substantial commitments about what the world is like which could be in conflict with what we generally know to be true. Obviously, anything we know to be true is representable by us, and thus is in no conflict with conceptual idealism. In a sense, the compatibility constraint comes for free in this case. There is no worry about conceptual idealism being incompatible with there having been rocks before there were humans, since that fact is clearly effable by us. In one sense the compatibility constraint is thus trivial for conceptual idealism to meet, but in another sense the issue can seem much more substantial and difficult. After all, how could it be that we constrain the totality of facts even though we didn’t even exist until recently? Conceptual idealism doesn’t just hold that we constrain the totality of facts now, but that we do so for all times. Even in the distant past there can’t be facts that are ineffable for us. But then, how can we reach back in time to constrain the totality of facts even then? This would be a serious issue

conceptual idealism: the overall picture 211 if constraint were to be understood causally. If we cause the totality of facts to be constrained by our minds, then we would have had to be around to do so. But constraint in the sense relevant for conceptual idealism is not a causal notion. It is not that we somehow causally kick out some ineffable facts from the totality of all facts, leaving only the effable ones in. Instead we constrain the totality of facts noncausally. The explanation why all the facts need to conform to our though is tied to the semantics of ‘all facts’ and how it is connected to what can be represented by us conceptually. This is not a causal explanation, but it nonetheless explains why every fact is effable by us. That same non-causal explanation works for past facts just like present facts. All facts, past, present, and future, are effable. In the end, all this is connected to the propositional not being an independent aspect of reality, but something directly tied to us. How that is to be understood more precisely as a metaphysical view is something we will consider shortly. For now we can conclude that the defense of conceptual idealism given above meets all three constraints. Although these constraints are only minimal constraints for any defense of idealism, conceptual idealism nonetheless does better than the other versions of idealism discussed and rejected in Chapter 2. Those versions succumbed to one or another of these constraints. That conceptual idealism meets these constraints might seem like a low bar, but it is not an easy low bar to meet. That it meets this bar supports that conceptual idealism should be taken seriously by everyone as a real contender in metaphysics. And the arguments given hopefully support that it is more than that. So far the review of the arguments given and why they justify a positive answer to the Big Question. Conceptual idealism answers that question, and answering this question was the main goal of this book. But conceptual idealism does not just answer the Big Question, it also is a proposal in metaphysics. And the way it was defended leads to a position about metaphysics and how it should operate at least in certain cases. The Big Question is not the only question all this is relevant for. But what precisely the lesson of both conceptual idealism and how it was defended is for other questions is not at all clear. We should now look beyond the Big Question and at these other aspects of the view: how to understand conceptual idealism as a metaphysical proposal, and how to understand the meta-metaphysical approach connected to its defense.

8.3 The metaphysics of conceptual idealism The defense of the idealist answer to the Big Question given above involves two interconnected aspects: one is a defense of a particular view in metaphysics, conceptual idealism, the other is a defense of a more meta-philosophical picture, which concerned drawing metaphysical conclusions from our own language, the significance of inescapable concepts, and the adoption of the immanent stance. In

212 idealism and the harmony of thought and reality this section I hope to discuss conceptual idealism as a metaphysical view, in the next section we will focus on the meta-philosophical picture. Conceptual idealism is not merely a view about our place in the world but also one about the world directly. It combines a realist view about reality with an idealist view about reality. Reality as the totality of things might well be completely independent of us, and thus ontological realism might well be true, as it seems to me to be. But reality as the totality of facts is not independent of us, and with it alethic idealism is true. This alethic idealist view supports our special place in the world, but it also supports a picture of reality as the totality of facts which is notably different than more standard views. We should now look at bit more at the ontological significance of conceptual idealism, in particular how it is compatible with ontological realism. Internalism about fact-talk implies that facts do not exist and that facts are not entities. Facts obtain, of course, but they do not exist. The argument for this conclusion is just a modification of the argument given above on page 144, where I argued for the non-existence of numbers from the constitutive non-referentiality of number-terms. In essence, it goes like this: if internalism is true, then fact-terms are non-referential in the sense that they semantically do something other than referring: they are constitutively non-referential. But then, whichever entity we might pick from the domain of entities, it won’t be the fact that snow is white. Since I used that phrase ‘the fact that snow is white’ non-referentially in the last sentence, according to internalism, it didn’t refer to whatever entity from the domain I picked out, and thus that entity is not the fact that snow is white. Thus the ontology of the world does not contain the fact that snow is white, and similar for all other facts. Thus facts do not exist. And the same also holds for all the other aspects of the propositional: propositions, meanings, contents, reasons, etc. do not exist either. Objects, on the other hand, do exist, or so it seems to me and everyone else except a few metaphysical radicals.2 And objects can figure in facts, in the sense that the fact concerns the objects, just as the Eiffel Tower figures in the fact that the Eiffel Tower is tall. Furthermore, objects exist independent of us, or so it is reasonable to think, even though we did not directly discuss this issue in more detail. Facts, on the other hand, are not independent of us: the totality of facts range-depends on us. This ontological realism is compatible with alethic idealism, or so I argued above in Chapter 5. The proper, context-sensitive semantics of inferential quantifiers over facts perfectly goes together with the object-permitting notion of the ineffable to give the result that every fact is effable, even though there is an independent domain of objects that can and do figure in the facts.

2 For our discussion here, it doesn’t matter which objects exist. Thus those that hold that there are only very small objects, like simples, still hold that objects exist. The real radicals are those who hold that there are no objects at all, neither simple nor complex. For a discussion of this radial view, see (Turner, 2011).

conceptual idealism: the overall picture 213 A similar question arises about the relationship between objects and properties. Objects exist, but do properties exist as well? Although we did not spend too much time on properties in this book, the question naturally arises whether internalism or externalism is true for talk about properties. The internalist position about facttalk naturally goes together with an internalist position about property-talk. From a philosophy of language point of view, very similar issues arise for the semantics of property-terms as they do for the semantics of fact-terms. For example, they both can exhibit the substitution failure behavior discussed in Chapter 4. Whether internalism is also correct for property talk is a topic beyond this book, but I have argued for this thesis in detail in (Hofweber, 2016b). If internalism is indeed true for talk about properties as well as talk about propositions, then there are no ineffable properties, just as there are no ineffable propositions. Furthermore, properties do not exist, just as fact to not exist. Things still have properties, but this does not mean that they relate to an entity which is a property. This is similar to facts obtaining, even though facts do not exist. For a fact to obtain does not mean, on this internalist picture, that there is some entity which has the property of obtaining. Things exist, and those things have properties, and they figure in facts, but neither the properties nor the facts exist besides the things. This leads to a limited form of nominalism, at least when focusing on the issue of an ontology of properties. The things that have properties do exist, but the properties they have do not exist. In the debate about universals, this is a form of nominalism about universals. But it is not a form of all out nominalism. The things that exist can be abstract or concrete. Nothing we have seen or discussed in any way precludes abstract objects. Nothing follows for all objects being concrete, only for properties not being among the entities, just as facts and propositions aren’t. All these issues, with a particular emphasis on the problem of universals, are spelled our in more detail in Chapter 11 of (Hofweber, 2016b). And all this is part of the metaphysical picture tied to conceptual idealism. Overall then, here is what the world looks like according to conceptual idealism, in a nutshell: there is, or might well be, an independently existing domain of objects, these objects have properties and figure in facts, but the properties they have and the facts they figure in do not exist and are not independent of our conceptual representations. Both properties and facts are range-dependent on us, although which properties which objects have, and which facts obtain, is not tied to us. This answers some metaphysical questions, but it leaves many more unanswered. Many philosophical problems and questions are closely tied to the propositional: content, reason, truth, etc. Conceptual idealism guarantees that the propositional is not a completely independent aspect of the world, and it has some immediate consequences for what it is like. For example, it guarantees that the world must be intelligible to us, since there can be no ineffable facts. And it guarantees that there is no limit to what we can know that arises from a limit of

214 idealism and the harmony of thought and reality what we can represent, since we are not limited in what we can represent, although we are limited in what we can know. The sources of our ignorance are always properly epistemic, and never representational. Ignorance is always due to a lack of evidence, and never due to us being unable to even represent a certain fact. We are consequently never unable to know a certain fact, since we can’t even think the relevant thought. We can say that deep ignorance obtains when we don’t know a fact since we can’t even represent the fact in thought or language. In this case we are not only unable to know a fact, we are even unable to ask the question whether this fact obtains. Deep ignorance is thus ruled out by internalism, although regular ignorance remains. Still, the world is less elusive for us than one might have feared, and our knowledge does not have to face a potential limitation from what we can’t even represent.3 So far so good, but these consequences of conceptual idealism mostly concern our relationship to the world. And we have seen a few other consequences, like those tied to the existence of objects, but not of the properties they have and the facts they figure in. But what else can we say directly about the world itself? What other, less direct consequences does conceptual idealism have for larger philosophical questions, in particular those tied to the propositional? I am afraid I won’t be able to answer this question here. This is not just because I hope to keep this book reasonably short, it is mostly because I am unsure of what the answer is. I have some ideas, but they are not worked out enough to be presentable here. This is in a sense the flip side of how we were able to meet the compatibility constraint above: conceptual idealism does not affect which facts in general obtain, since which facts obtain is independent of us. The only claim of dependence is that the totality of facts range-depends on our minds: all facts must be representable by us. Thus there is no tension with what we know, since that only concerns facts that both obtain and are representable by us. But then there will also be no immediate implication for which facts obtain that are of larger philosophical significance. At least it is not obvious what else about the propositional follows besides that it is not an independent aspect of the world. But this won’t be the final answer. The propositional is so central for so many philosophical issues, that any substantial claim about the propositional should have significant consequences for several philosophical issues tied to the propositional. I believe, for that reason as well as some others, that much more of metaphysical significance follows from conceptual idealism than what I discussed above, but if I had worked out what it is, I would have said so. I must leave this issue largely for another occasion.

3 For more on deep ignorance and its significance for metaphysics, see (Hofweber, 2016a) and chapter 10 of (Hofweber, 2016b).

conceptual idealism: the overall picture 215

8.4 The meta-metaphysics of conceptual idealism What metaphysical questions are answered by conceptual idealism is one thing, how conceptual idealism itself was established as a metaphysical position is another one. Conceptual idealism is merely a view about the place of our minds in reality, but the argument given for conceptual idealism brings with it a larger strategy for answering at least some metaphysical questions. In section 7.5 I called this approach immanent metaphysics, and in this section I would like to discuss the meta-philosophical side of immanent metaphysics some more. Immanent metaphysics is a general approach towards how one can make progress in metaphysics. It is a general strategy for such progress, but not a distinct kind of metaphysics. Just like many other approaches to metaphysics, immanent metaphysics aims at reality and what is the case. But contrary to many other approaches, immanent metaphysics holds that some results about reality can be achieved via thinking about our own concepts and their inescapability. In section 6.3 I outlined how inescapable concepts could be seen as being significant either in an approach inspired by Carnap, or in an approach inspired by Kant. The neoCarnapian approach sees inescapable concepts as the endpoint for philosophical work, since that work focused on improving our concepts. The neo-Kantian sees it as a starting point, since those concepts allow us to draw conclusions about reality. I take the approach defended in this book to be neo-Kantian in spirit, and I would now like to work out some similarities and some differences to Kant’s own approach. If I had to put a label on my own approach here, I would happily call it neo-Kantian, even though there are substantial differences between Kant’s position, as well as Kant’s general approach, and the one adopted here. I used the terminology of a reformed neo-Kantian approach above, on page 160, which was supposed to contrast with more orthodox neo-Kantian approaches, which follow Kant more closely in letter. Arguably, one strictly can’t call oneself a neoKantian in metaphysics unless one accepts transcendental idealism.⁴ Conceptual idealism is not a version of transcendental idealism, and so the present position does not meet this strict standard. Nonetheless, there are some important general similarities between Kant’s attempt to establish transcendental idealism and the argument for conceptual idealism given here. In the contemporary metaphysical scene these similarities seem to me to put the present approach closer to a Kantian one than other actively debated alternatives, although it is not that close to Kant relative to other Kantian approaches to metaphysics. The label-

⁴ Naturally, all aspects of this are disputed in scholarly circles. There are questions about how much transcendental idealism is really a metaphysical thesis for Kant, and there are questions about whether the historical neo-Kantians like Cassirer or Cohen held such metaphysical views. See (Allison, 2004) on the former, and (Heis, 2018) on the latter.

216 idealism and the harmony of thought and reality ing and associating with historical figures mostly comes down to with which movement one feels comfortable to align oneself, and nothing really hangs on it. So, instead of wasting too much time on who deserves what label, let us look at some key similarities and differences between Kant’s position and the one defended here. Obviously, all aspects of Kant’s view are subject to scholarly debate, and I do not have the scholarly expertise to contribute to this debate. Nonetheless, there is a widespread and maybe even standard view of Kant’s position, which I will adopt here for our compare-and-contrast purposes. It goes something like this: In the Critique of Pure Reason, (Kant, 1781), Kant argued that the empirical world must conform to our concepts, in particular a certain distinguished class of concepts: the categories. This is because those concepts are actively involved in how the empirical world is constructed. The empirical world conforms to our concepts in part because it is constructed by us with the employment of these concepts. The categories therefore also constrain the empirical world. Our concepts constrain the world because the world is constructed with our concepts. Constraint thus follows construction on Kant’s view. This result is limited in that it only applies to the empirical world, but not to all of reality. The noumenal world of things as they are in themselves is not necessarily constrained in the same way, since it is not constructed by us using our concepts. This leads to a key aspect of Kant’s metaphysical position. On the one hand the world depends on our minds, but only the empirical or phenomenal world. On the other hand, the noumenal world is independent of our minds. On the one hand, metaphysics is possible, but only insofar it concerns the empirical world. For example, we can find out that every change has a sufficient cause, and that is a result in metaphysics, at least as the term is used these days. Such metaphysical results are possible, but, on the other hand, much traditional metaphysics is impossible for us. Questions about the existence of God, freedom of the will, or the nature of the noumenal world are beyond us, at least in purely theoretical inquiry that aims at knowledge. Nonetheless the results of metaphysics concerning the phenomenal world can be achieved by reflecting on how our concepts are employed in its construction. Obviously, the metaphysical position defended here is quite a bit different from Kant’s position. However, the general approach of trying to achieve results about reality via reflecting on a special class of our own concepts is shared. This focus on our own concepts as a source of non-deflated insight into general features of the world I take to be distinctly Kantian, even though other philosophers surely had similar ideas. It is because of this, and despite all the differences in the resulting metaphysical views as well as in the resulting version of idealism, that I think of the present approach as broadly neo-Kantian. Still, the differences are quite significant and worth making explicit. First of all, conceptual idealism applies to reality as a whole not merely to a part or aspect of reality, like the empirical world. It is all of reality that must conform to our concepts. Conceptual idealism is thus not a

conceptual idealism: the overall picture 217 form of transcendental idealism.⁵ Second, the reason why our concepts constrain the world is very different from Kant’s reason. Conceptual idealism maintains that constraint happens without construction. Kantian transcendental idealism achieves constraint via construction. Conceptual idealism does not take the world to be constructed at all. It achieves constraint via internalism and what it implies about what the totality of facts must be like. But on the downside, conceptual idealism does not directly lead to any of the results like those concerning causation, persistence, and so on, which for Kant follow from the particular construction story of the phenomenal world. Conceptual idealism leaves largely open what reality is like, other than that it is constrained by our conceptual representations. As mentioned above, more results about reality can surely be drawn from conceptual idealism than I was able to do here. But the way this would work will go rather differently from how Kant drew further conclusions about reality from his defense of transcendental idealism. But there are not just differences between the two idealist positions. There are also crucial differences between the respective overall argument strategies for these different versions of idealism. Kant’s argument for why the world must conform to our concepts is a transcendental argument, while the argument for conceptual idealism is not a transcendental argument. This crucial difference in argument strategy is important, and closely tied to the difference between inescapable concepts and inevitable concepts. For Kant, as I understand him, the distinguished class of concepts, the categories, are inevitable: every thinker must have them in order to be a thinker at all. Thought without these concepts is impossible.⁶ This avoids the threat of a certain kind of subjectivism. The empirical world must conform to these concepts, not just as it is for me but also as it is for you and for any other thinker. Any thinker will have to rely on these concepts, and so however the world is constructed, it must conform to these inevitable concepts not just for some thinkers, but for any thinker. Obviously, how all that is supposed to go is left open by this, but even on a rough way of approaching it, the significance of the categories being inevitable becomes clear: experience of the spatiotemporal world is only possible, since the empirical world is constructed by us with our special concepts. Whether these concepts are indeed inevitable is naturally a further question, and to show that they are should be a difficult undertaking.

⁵ Although different names for various forms of idealism are used differently by different people, there is a use of the term ‘absolute idealism’ as a version of post-Kantian idealism that overcomes various dualisms in Kant’s philosophy, including the one between the phenomenal and noumenal world. See (Guyer, 2017). That would apply here as well, and so in this particular sense, conceptual idealism is a form of absolute idealism. However, many other things associated with absolute idealism do not carry over to the present position, and in general it is probably best to avoid these heavily, but differently, used labels. ⁶ For a discussion of this position in Kant, and what argument he might have had for it, see (Filcheva, 2019).

218 idealism and the harmony of thought and reality In contrast, the argument for conceptual idealism is not a transcendental argument. I never argued that thought or experience or anything else are not possible, unless some special class of concepts applies to the world. And the special class of concepts used to support conceptual idealism—inescapable concepts—do not have to be inevitable. Instead of a transcendental argument concerning the necessary conditions for the possibility of something or other, I hoped to argue that, first, a certain question can be answered by reflecting on our concepts or language, and, second, that the question as answered was the right question to ask, since the key concept in it is an inescapable one. The fact that other thinkers can have other concepts that are inescapable for them is then not taken to undermine our own question, but rather to lead to the considerations about the immanent stance and to us taking our own position seriously. Conceptual idealism is not defended via a transcendental argument, and it doesn’t need to be. However, the fact that we don’t have a transcendental argument for it brings with it the further complexity tied to the immanent stance, which I will discuss a bit more just below. But this difference is a key one, and I believe being able to argue for idealism without requiring a transcendental argument is one of the key advantages of the present approach. It brings with some extra complexity, but it also makes a new kind of argument possible that does not require transcendental idealism, but rather supports idealism for all of reality as the totality of facts. Finally, Kant’s own approach relied on a more systematic characterization of the special class of concepts. Kant starts out early on in his argument with what he takes to be the complete list of the special concepts: the table of categories. There is notoriously little argument in the Critique of Pure Reason concerning why this is supposed to be the complete list.⁷ But it is clear that the Kantian argument proceeds from the complete list of the categories. Kant’s strategy is first to find the complete list of the categories, namely in the metaphysical deduction, and then to show that these categories apply to, or structure, the empirical world, namely in the transcendental deduction. This is not a piecemeal approach, but a holistic one. In contrast, the present argument does not proceed via the complete list of inescapable concepts. Although we discussed several candidates of such concepts, and even entertained some hunches about what a complete list might look like, the argument for idealism simply and exclusively relied on the inescapability of the concept of fact and similar nearby notions tied to the propositional. No systematic account of inescapable concepts was required or given, although such an account would naturally be desirable. If there is an argument that proceeds from the complete list of inescapable concepts to some metaphysical conception of the world, I would be more than eager to see it.

⁷ See (Reich, 1998).

conceptual idealism: the overall picture 219 Although, to the best of my knowledge, Kant never explicitly used the term ‘immanent metaphysics’, he gets pretty close in places,⁸ and there is a widespread use of that term in the Kant literature. According to this use, for Kant immanent metaphysics is the metaphysics of the phenomenal world, as opposed to transcendent metaphysics, which goes beyond the phenomenal world and makes metaphysical claims about noumenal reality. So understood, Kant holds that immanent metaphysics is possible, while transcendent metaphysics is not, at least not as a purely theoretical enterprise. Immanent metaphysics, in Kant’s sense, is possible, since we are in a position to find out about basic features of the phenomenal world, via the role of our concepts in the construction of the phenomenal world. Immanent metaphysics in the sense defended here is not to be understood this way, but the two senses of the term are also not completely unrelated. As I understand ‘immanent metaphysics’, it is a method for making progress in metaphysics by thinking about our own concepts and their inescapability, while adopting the immanent stance. There is no distinction assumed or required between the phenomenal and the noumenal world, and there is no assumption made that the world is constructed by us in any way. Furthermore, and maybe most importantly, no claim is being made that other approaches to metaphysics are impossible and pointless. To the contrary, I think of immanent metaphysics to be just one part of metaphysics, although likely quite an important part. But despite these differences, there are also some similarities between the two senses of immanent metaphysics. For example, on both senses progress is being made via the significance of our concepts for what reality is like. But the key differences spelled out above will hopefully make clear that there is not too much else they have in common. I use this terminology not to mimic the Kant literature, but because it naturally suggests itself after adopting the terminology of immanent rationality and of the immanent stance. It fittingly suggests that immanent metaphysics concerns features of reality that are in a sense already immanent to our conceptual representations and the rational requirements that come with them. Other terminological choices would certainly have lead to other cases of terminological overlap, and I picked what seemed best, all things considered. Whether the present view deserves the label of being neo-Kantian in light of the similarities and despite all the differences is obviously not a substantial question. I find inspiration in Kant, and so I like the label, but nothing hangs on this. But the difference between using transcendental arguments and inevitable concepts, as Kant does, and using inescapable concepts, but no transcendental arguments, leads to a more substantial difference on which a lot hangs. If we rely on inescapable concepts which are not inevitable, as I did above, then we face the problems tied to the possibility of others who have different concepts that are inescapable for them.

⁸ See, for example, (Kant, 1781, B873).

220 idealism and the harmony of thought and reality I put in an effort to deal with this situation above in Chapters 6 and 7. If one wants to rely on inevitable concepts, then one faces the problem of arguing that these concepts are truly required for having thought at all, and furthermore the more serious problem that reality must somehow conform to these concepts. After all, maybe thought in general is unsuitable for reality, and thus even though every thinker must have these concepts, reality wants to have nothing to do with them. This will be a tough bridge to cross; too tough, I think, for inevitable concepts being a fruitful concept in metaphysics. How all this relates to Kant is obviously only of comparative significance. What is more important is what would follow for how different parts of philosophy relate to each other. If immanent metaphysics, in the sense defended here, is viable, then this would have consequences for how different philosophical disciplines interact with each other. In particular, it would support that both the philosophy of language and epistemology have a constructive and not just a corrective role in metaphysics, in the sense discussed above on page 140: these disciplines would make a positive contribution to achieving results in metaphysics, and not merely help the metaphysician to avoid errors. To achieve results within immanent metaphysics one needs to derive some broadly empirical conclusions about our own conceptual representations. We need to draw some conclusions about what we do when we talk or think the way we do, and then use this to bridge the languagemetaphysics gap. And we need to rely on some broadly epistemic considerations that settle which concepts are inescapable, i.e. which concepts can’t rationally be replaced with different ones. These two parts together lead to metaphysical results, including conceptual idealism, or so the basic idea of this book. If that is correct, then the philosophy of language and epistemology would play a constructive role in metaphysics, and with it there would be a close connection between those three disciplines. On the face of it, these three areas of philosophy appear to be about different things. The philosophy of language is about how we represent the world in language, epistemology is about what we are entitled to do or think, but metaphysics is about what reality is like, and those are quite different. But if immanent metaphysics is a fruitful project, then these parts of philosophy are much more closely connected, and they should therefore not be pursued in isolation. To repeat once more: immanent metaphysics is not all of metaphysics. But how does immanent metaphysics relate to the rest of metaphysics? Here there are three natural ways to think of it, one egalitarian, and two that are not. On the egalitarian way, immanent metaphysics is just one of many ways to make progress in metaphysics. It has no special or distinguished status, and it stands next to other methods that try to make progress as well. How this all comes together into one picture of the world will have to be settled on a case by cases basis, depending on who has the better arguments or stronger reasons in particular cases. But we should

conceptual idealism: the overall picture 221 not, on the egalitarian take of it, expect that immanent metaphysics is somehow special in general, either for the better or the worse. It might win out here or there, but at first it is just one among many other methods. On the inegalitarian conception of the relationship between immanent metaphysics and the rest of metaphysics, immanent metaphysics is special. But that leaves open if it is special in a good way or a bad way. On the good way it can be seen as setting the boundaries within which the rest of metaphysics needs to operate. It is special in that it is the most secure and central part of metaphysics, leaving open other things to be filled in. In the bad way, on the contrary, it has only a minor and derivative status, being secondary to other metaphysical methods. Which of these conceptions is the correct one is not easy to see, in part since it is not clear how metaphysics in general makes progress and how its different parts relate to each other. But there is a strong preliminary case for the inegalitarian conception in which immanent metaphysics is special in a good way. That case is based on the epistemic strength of the results of immanent metaphysics compared to those in other parts of metaphysics. Immanent metaphysics achieves results by drawing conclusions from our own representations, which is something that is epistemically tractable. We have good support for finding out about our own language and our own concepts, although clearly much disagreement persists. And if we can show that a particular concept is inescapable, then this is especially well supported, since only then would the concept truly be inescapable. Any result of immanent metaphysics would thus be well supported, if it can be achieved at all. But in general this is not so in metaphysics. Many metaphysical debates are speculative, and positions are supported with fairly weak reasons like theoretical virtues: comparative simplicity, ontological sparseness, and so on. Such results would not have strong standing, and thus would not be in a good position to displace any result of immanent metaphysics. Since these have a distinguished support, they have a special place in all of metaphysics. If so, then immanent metaphysics would draw the boundaries within the rest of metaphysics needs to operate. But whether that is indeed so is far from clear. Maybe insights into our own language are less secure than I made it out to be, and maybe some other parts of metaphysics have an even more special standing, with even stronger support. For example, it could be that some parts of metaphysics are settled in the sciences, and possibly in those sciences that are more secure than anything we might find out about our own concepts or language. All this is an endgame consideration, requiring a much clearer picture of how metaphysics can achieve results than we have today. Still, in the present metaphysical scene immanent metaphysics would do rather well compared to the alternatives, but how it will all turn out in the end must be seen as being largely open.

222 idealism and the harmony of thought and reality

8.5 Onward The main goal of this book was to defend the third, idealist answer to the Big Question about our place in the world. I made my best case for this above. But I am modest enough to accept that it might be deeply mistaken and completely wrong. Maybe you suspect as much, but hopefully you are modest enough to allow for the possibility that it might have been right or at least onto something. If we take that possibility seriously, then the question arises where to go next. What follows from what we have seen, and what else can be established in a similar way? In this final section I would like to briefly reflect on this, in part because it seems to me that if what happened before is onto something, then this is not an isolated argument for an isolated thesis, but rather the beginning of something that is possibly quite fruitful. Let me therefore close this book by taking a peek ahead. There are really two separate things to consider while looking forward: what follows from conceptual idealism itself, and what else can be achieved with the methods that allowed us to defend conceptual idealism. As already acknowledged and discussed just above, in section 8.3, it is not completely clear what the metaphysical consequences of conceptual idealism are, nor what the more precise metaphysical picture is that comes with it. I outlined some connected positions, for example an ontology of objects, but not of properties and facts. But the picture still needs to be filled in, as acknowledged just above. Obviously, this is a first thing to continue to think about. The next thing to consider is not directly tied to conceptual idealism, but to the natural generalization of the position that was relied upon to defend conceptual idealism: internalism about facts. As discussed above as well, internalism will not just apply to talk about facts, if it is true at all, but it will generalize to the propositional: all that which is associated with that-clauses like content, reasons, and so on. Conceptual idealism just relied on internalism about facts; internalism about reasons, say, played no role in its defense. But internalism about the propositional in general is much stronger than internalism merely about facts. The question thus remains what else we can conclude from internalism about the propositional more generally. For example, it follows from considerations like the ones given above that there are not just no ineffable facts, there are also no ineffable reasons. But what the larger philosophical consequences of internalism concerning the propositional as a whole are is so far an open question. But possibly the most promising source of progress is not tied to the consequences of the positions defended here, but to the methods that were employed in defending them. If the defense of conceptual idealism works, then we were able to settle a question about reality simply by reflecting on our own representations about reality, even though the answer is not a conceptual truth in the traditional sense of the term. And what is more, we were able to determine that we did not just answer the question we happened to have asked, we answered the right question to

conceptual idealism: the overall picture 223 ask, again all by simply reflecting on our own representations of reality alone. All this should be impossible, but if it is possible after all, then it would be surprising if it only worked for the defense of conceptual idealism, and not more broadly. What else then can we figure out this way? Immanent metaphysics is the approach to making progress in metaphysics that uses this methodology. Somewhat programmatically, we can call the immanent metaphysics program the systematic attempt to answer all the questions that can be answered this way. If what I said above is onto something, then this is a legitimate program. What is unclear is how significant it might be and what might be in its scope. But there are some promising candidates to consider. First, there are cases where one focuses on conceptual requirements tied to particular concepts, but faces an objection targeting the status of these concepts. A classic example of this are so-called constitutivist approaches to some philosophical problem, and a classic example of that is the case of agency. Here the proposal is that agency has a constitutive goal: the good, self-constitution, selfunderstanding, or what have you. That this is the goal of agency can be taken to be tied to a conceptual truth connecting agency to its constitutive goal. And this gives rise to the strategy of arguing that since we are agents we are subject to various norms tied to the constitutive goal of agency. On ambitious attempts of this it might even include the requirements of morality.⁹ But this faces an objection tied to the status of being an agent. Why not be something other than an agent, a shmagent, and thereby escape the requirements that come with being an agent?1⁰ To insist here that one can’t help but be an agent is only a weak response, since a shmagent can be seen as being just like an agent, except without certain constitutive goals, and it is hard to see why one can’t be one of these, maybe simply by choice. A stronger response is tied to the inescapability of the concept of agency. If that concept is inescapable,11 then it would be irrational for me to replace it with a different one. I am not only an agent, I think of myself as an agent. Thus I apply that concept of agency to myself, not just in how I conceive of myself but also when I deliberate about what to do. If that concept of agency is inescapable, then replacing it with that of shmagency in my thinking would be irrational, and thus not permissible. But then, arguably, so would be the move of trying to escape the requirements tied to the constitutive aim of agency by trying to be a shmagent rather than an agent. The inescapability of the concept of agency thus solidifies the norms tied to the constitutive aim of agency. This might not seem like much of a metaphysical conclusion, but that is merely a terminological point. The fact remains that by reflecting on our concept of agency

⁹ For proposals along these lines, see (Korsgaard, 2009) or (Velleman, 2000). 1⁰ See (Enoch, 2006). 11 I outlined an argument for this above on page 154 and develop it, as well as the thought in this section, in more detail in (Hofweber, 2022a).

224 idealism and the harmony of thought and reality and its inescapability we can draw conclusions about normative requirements that apply to us inescapably. This conclusion could be drawn simply from thinking about our own concept, but it nonetheless is a conclusion about reality. Thus it is in the ballpark of immanent metaphysics, despite the conclusion concerning normative facts. But there is also a crucial difference to the argument for conceptual idealism: there the conclusion about reality was not tied to a conceptual truth, but to broadly linguistic facts about our talk about facts. Still, I think both of these approaches are largely congenial, and it is no accident that both of them can with some justification be labeled neo-Kantian. Constitutivist approaches of all kinds, augmented with the inescapability of the relevant concepts, are thus natural next targets for immanent metaphysics. But not all that can be gotten from our own concepts is a conceptual truth involving this concept. That fact-terms are constitutively non-referential was not a conceptual truth about fact-terms. And it is not a conceptual truth tied to the concept of a fact that facts do not exist, at least in the sense of ‘conceptual truth’ spelled out above on page 145. It is something that we discovered through substantial investigation into our own language, and it is not something that is closely tied to merely having the relevant concepts. I find this way of answering questions to be a much neglected possibility, and I suspect much more can be done in this regard than what has been done already. And if we can augment such results with the inescapability of the relevant concepts, then we would bring the issue in the domain of immanent metaphysics. Here there are some easier ways to make progress and some so far more elusive ones. The easier ones are closely tied to the issues we have discussed above. For example, if a class of expressions is constitutively non-referential, then we can conclude from this that there are no entities of a certain kind and that certain things do not exist. This was spelled out above, with the argument that internalism guarantees the non-existence of the relevant things. This applied to numbers and facts, and it will also apply to any other case.12 But drawing conclusions about reality from reflecting on our own representations does not have to be restricted to ontological conclusions. That every fact is effable by us, for example, is not directly an ontological conclusion, although it is connected to facts not being entities and not existing. To find more such connections, either in a piecemeal or in a large-scale way, will be a key part of making progress in immanent metaphysics. How that can go for other cases is so far elusive, but I see no in principle obstacle for reaching new pastures that way. In summary, then, here is where we are: We all have to start somewhere, in inquiry in general, and in metaphysics in particular. When we start out, we find ourselves with various concepts, that we are defeasibly entitled to employ. We use these concepts to ask questions that we hope to answer, and thereby set the

12 Much more on this is in (Hofweber, 2016b).

conceptual idealism: the overall picture 225 goalpost for what we hope to do. We hope to answer these questions in a way is not just true, but that also makes sense to us, so that by our own best way of weighing the reasons for or against an answer, the answer comes out on top. We can adjust our ways and our concepts, but any such adjustment must make sense to us, and thus be rational by our own lights. This puts some constraints on what alternative concepts we may employ instead, since for certain concepts— inescapable concepts—it would never be rational by our own light to replace them with alternatives. And thus these concepts can never rationally be given up. Furthermore, sometimes it is possible to answer questions of fact simply by reflecting on our own concepts, even though these answers are not conceptual truths. When this concerns an inescapable concept, then this answer is a kind of a fixed point that will remain as we move forward in search for other answers and better concepts. It would be a result of immanent metaphysics, one that can be taken to put some boundaries on what other results we should look for. The immanent metaphysics program hopes to work out these boundaries, and with it all the results about what reality is and must be like that we can get this way. These results are not just ones that we must accept on pain of irrationality, they are really insights into what reality is like. Conceptual idealism and the idealist answer to the Big Question are such results, but they won’t be the only ones. How much of an advance this approach will lead to I can’t predict with any confidence. My own sense is that the immanent metaphysics program is a fruitful one, although it remains to be seen just how fruitful it can be. I hoped to have made the case in this book that it can lead to substantial and significant results, with the defense of conceptual idealism as a proof of concept. We should not expect that all or even most metaphysics is to be carried out this way. Immanent metaphysics is at best a special part of metaphysics. But immanent metaphysics should seem impossible, and thus having one instance that works justifies the general idea of the project, and with it the immanent metaphysics program. Once it is on the table we should pursue it as far as it can take us. Even if we had that program completed, much of metaphysics would likely still be open. But I dare conjecture that we would have made a lot of progress.

Bibliography Adams, R. (1979). Primitive thisness and primitive identity. Journal of Philosophy, 76: 5–26. Adams, R. (2007). Idealism vindicated. In van Inwagen, P. and Zimmerman, D., editors, Persons human and divine, pages 35–54. Clarendon Press. Allison, H. (2004). Kant’s Transcendental Idealism: an interpretation and defense. Yale University Press. Ayer, A. J. (1946). Language, Truth and Logic. Victor Gollancz, London, 2nd edition. Bach, E. (1986). Natural language metaphysics. In Marcus, R. B., Dorn, G., and Weingartner, P., editors, Logic, Methodology, and Philosophy of Science, VII. North-Holland. Bach, K. (1997). Do belief reports report beliefs? Pacific Philosophical Quarterly, 78: 215–241. Bader, R. (2010). The Transcendental Structure of the World. PhD thesis, University of St. Andrews. Beiser, F. (2002). German Idealism: the struggle against subjectivism 1781–1801. Harvard University Press. Berkeley, G. (1710). A Treatise Concerning the Principles of Human Knowledge. Aaron Rhames. Black, M. (1952). The identity of indiscernibles. Mind, 61:153–164. Blackburn, S. (1984). Spreading the Word. Oxford University Press. Blackburn, S. (1993). Essays in Quasi-Realism. Oxford University Press. Boghossian, P. (2000). Knowledge of logic. In Boghossian, P. and Peacocke, C., editors, New Essays on the A Priori, pages 229–254. Oxford University Press. Broome, J. (2013). Rationality through Reasoning. Wiley-Blackwell. Burge, T. (1992). Philosophy of language and mind: 1950–1990. Philosophical Review, 101(2):3–51. Cappelen, H. (2018). Fixing Language: an essay on conceptual engineering. Oxford University Press. Carnap, R. (1956). Empiricism, semantics, and ontology. In Meaning and Necessity, pages 205–221. University of Chicago Press, 2nd edition. Chalmers, D. (2011). Verbal disputes. The Philosophical Review, 120(4):515–566. Chalmers, D. (2019). Idealism and the mind-body problem. In Seager, W., editor, The Routlege Companion to Panpsychism, pages 353–373. Routledge. Chisholm, R. (1948). The problem of empiricism. Journal of Philosophy, 45(19):512–517. Chomsky, N. (1976). On the biological basis of language capacities. In Rieber, R., editor, The Neuropsychology of Language, pages 1–24. Springer. Cobreros, P., Égré, P., Ripley, D., and van Rooij, R. (2013). Reaching transparent truth. Mind, 122(488):841–866. Dasgupta, S. (2018). Realism and the absence of value. Philosophical Review, 127(3): 279–322. Davidson, D. (1984). The very idea of a conceptual scheme. In Inquiries into Truth and Interpretation, pages 183–198. Oxford University Press. Dickler, G. (2011). Berkeley’s Idealism: a critical examination. Oxford University Press.

228 bibliography Dreier, J. (2004). Meta-ethics and the problem of creeping minimalism. Philosophical Perspectives, 18(1):23–44. Dummett, M. (1978a). The justification of deduction. In Truth and other Enigmas, pages 290–318. Harvard University Press. Dummett, M. (1978b). The philosophical basis of intuitionistic logic. In Truth and other Enigmas. Duckworth. Dummett, M. (1978c). Truth and other Enigmas. Duckworth. Dummett, M. (1991). Frege: philosophy of mathematics. Harvard University Press. Dyke, H. (2008). Metaphysics and the Representational Fallacy. Routledge. Einheuser, I. (2003). Conventionalism. PhD thesis, MIT. Einheuser, I. (2006). Counterconventional conditionals. Philosophical Studies, 127(3): 459–482. Eklund, M. (2015). Intuitions, conceptual engineering, and conceptual fixed points. In Daly, C., editor, The Palgrave Handbook of Philosophical Methods, pages 363–385. Palgrave Macmillan. Eklund, M. (2017). Choosing Normative Concepts. Oxford University Press. Enoch, D. (2006). Agency, shmagency: why normativity won’t come from what is constitutive of action. Philosophical Review, 115(2):169–198. Faye, J. (2019). Copenhagen interpretation of quantum mechanics. In Zalta, E. N., editor, The Stanford Encyclopedia of Philosophy. Metaphysics Research Lab, Stanford University, winter 2019 edition. Ferrero, L. (2009). Constitutivism and the inescapability of agency. Oxford Studies in Metaethics, 4:303–333. Field, H. (2008). Saving Truth from Paradox. Oxford University Press. Field, H. (2020). A “naturalistic” take on epistemology. Proceedings and Addresses of the American Philosophical Association. Filcheva, K. (2019). Kant on alien logical forms. The Philosophical Forum, 50(3):287–307. Filcheva, K. (2020). Can there be ineffable propositional structures? Journal of Philosophical Research, 44. Fine, K. (2001). The question of realism. Philosophers’ Imprint, 1(1):1–30. Fine, K. (2005). Tense and reality. In Modality and Tense, pages 261–320. Oxford University Press. Fine, K. (2009). The question of ontology. In Chalmers, D., Manley, D., and Wasserman, R., editors, Metametaphysics: new essays on the foundations of ontology, pages 157–177. Oxford University Press. Foley, R. (1992). Working without a Net: a study of egocentric epistemology. Oxford University Press. Förster, E. (2012). The Twenty-Five Years of Philosophy: a systematic reconstruction. Harvard University Press. Foster, J. (2008). A World for Us: the case for phenomenalistic idealism. Oxford University Press. Frances, B. and Matheson, J. (2019). Disagreement. In Zalta, E. N., editor, The Stanford Encyclopedia of Philosophy. Metaphysics Research Lab, Stanford University, winter 2019 edition. Frege, G. (1884). Die Grundlagen der Arithmetik: eine logisch mathematische Untersuchung über den Begriff der Zahl. W. Koebner. Gaskin, R. (2020). Language and World: a defense of linguistic idealism. Routledge. Godfrey-Smith, P. (2006). Theories and models in metaphysics. Harvard Review of Philosophy, 14:4–19.

bibliography 229 Goff, P., Seager, W., and Allen-Hermanson, S. (2020). Panpsychism. In Zalta, E. N., editor, The Stanford Encyclopedia of Philosophy. Metaphysics Research Lab, Stanford University, summer 2020 edition. Goodman, N. (1978). Ways of Worldmaking. Hackett. Guyer, P. (2017). Absolute idealism and the rejection of Kantian dualism. In Ameriks, K., editor, The Cambridge Companion to German Idealism, pages 43–64. Cambridge University Press. Guyer, P. and Horstmann, R.-P. (2015). Idealism. In Zalta, E. N., editor, The Stanford Encyclopedia of Philosophy. Metaphysics Research Lab, Stanford University, fall 2015 edition. Heis, J. (2018). Neo-Kantianism. In Zalta, E. N., editor, The Stanford Encyclopedia of Philosophy. Metaphysics Research Lab, Stanford University, summer 2018 edition. Hofweber, T. (2005). Supervenience and object-dependent properties. Journal of Philosophy, CII(1):5–32. Hofweber, T. (2006). Inexpressible properties and propositions. In Zimmerman, D., editor, Oxford Studies in Metaphysics, volume 2, pages 155–206. Oxford University Press. Hofweber, T. (2008). Validity, paradox, and the ideal of deductive logic. In Beall, J., editor, Revenge of the Liar: New Essays on the Paradox, pages 145–158. Oxford University Press. Hofweber, T. (2009). Ambitious, yet modest, metaphysics. In Chalmers, D., Manley, D., and Wasserman, R., editors, Metametaphysics: New Essays on the Foundations of Ontology, pages 269–289. Oxford University Press. Hofweber, T. (2014). Rayo’s The Construction of Logical Space. Inquiry, 57(4):442–454. Hofweber, T. (2015). The place of subjects in the metaphysics of material objects. dialectica, 69(4):473–490. Hofweber, T. (2016a). Intellectual humility and the limits of conceptual representation. Res Philosophica, 93(3):553–565. Hofweber, T. (2016b). Ontology and the Ambitions of Metaphysics. Oxford University Press. Hofweber, T. (2017a). Are there ineffable aspects of reality? In Bennett, K. and Zimmerman, D., editors, Oxford Studies in Metaphysics, volume 10, pages 124–170. Oxford University Press. Hofweber, T. (2017b). Conceptual idealism without ontological idealism: why idealism is true after all. In Pearce, K. and Goldschmidt, T., editors, Idealism: New Essays in Metaphysics. Oxford University Press. Hofweber, T. (2018). Replies to Eklund and Uzquiano. Analysis, 78(2):315–344. Hofweber, T. (2019a). Empirical evidence and the metaphysics of ordinary objects. In Cumpa, J. and Brewer, B., editors, The Nature of Ordinary Objects. Cambridge University Press. Hofweber, T. (2019b). Idealism and the harmony of thought and reality. Mind, 128(511): 699–734. Hofweber, T. (2022a). The inescapability of the concept of agency. Unpublished manuscript. Hofweber, T. (2022b). The unrevisability of logic. In Hawthorne, J. and Turner, J., editors, Philosophical Perspectives, volume 35. Wiley-Blackwell. Hofweber, T. and Lange, M. (2017). Fine’s fragmentalist interpretation of special relativity. Noûs, 51(4):871–883. Hofweber, T. and Lange, M. (2020). Fragmentalism and special relativity. Teorema, 39(3): 5–15. Hofweber, T. and Pelletier, J. (2005). Encuneral noun phrases. Unpublished manuscript, available at www.thomashofweber.com. Hume, D. (1739). A Treatise of Human Nature. John Noon.

230 bibliography Jeshion, R. (2010). Singular thought: acquaintance, semantic instrumentalism, and cognitivism. In New Essays on Singular Thought. Oxford University Press. Jonas, S. (2016). Ineffability and its Metaphysics: The Unspeakable in Art, Religion, and Philosophy. Palgrave Macmillan. Kant, I. (1781). Kritik der reinen Vernunft. Johan Friedrich Hartnoch, Riga. Keisler, H. J. (1971). Model Theory for Infinitary Logic. North-Holland. King, J. (2002). Designating propositions. Philosophical Review, 111(3):341–371. Koch, A. F. (1990). Subjektivität in Raum und Zeit. Klostermann, Frankfurt am Main. Koch, A. F. (2006a). The subjectivity thesis and its corollaries. Philosophical Inquiry, 28:9–20. Koch, A. F. (2006b). Versuch über Wahrheit und Zeit. Mentis, Paderborn. Koch, A. F. (2010). Persons as mirroring the world. In O’Shea, J. and Rubinstein, E., editors, Self, Language, and World, pages 232–248. Ridgeview Publishing. Koch, A. F. (2016). Hermeneutischer Realismus. Mohr Siebeck. Korman, D. (2015). Objects: Nothing Out of the Ordinary. Oxford University Press. Korman, D. Z. (2016). Ordinary objects. In Zalta, E. N., editor, The Stanford Encyclopedia of Philosophy. Metaphysics Research Lab, Stanford University, spring 2016 edition. Korsgaard, C. (2009). Self-Constitution: Agency, Identity, and Integrity. Oxford University Press. Künne, W. (2003). Conceptions of Truth. Oxford University Press. Ladyman, J. and Ross, D. (2007). Everything Must Go. Oxford University Press. Lange, M. (2016). Because without Cause: Non-Casual Explanations In Science and Mathematics. Oxford University Press. Langton, R. (1998). Kantian Humility: Our Ignorance of Things in Themselves. Oxford University Press. Lasonen-Aarnio, M. (2014). The dogmatism puzzle. Australasian Journal of Philosophy, 92(3):417–432. Lewis, C. (1946). An Analysis of Knowledge and Valuation. Open Court. Lewis, D. (1983). New work for a theory of universals. Australasian Journal of Philosophy, 61:343–377. Lewis, P. (2016). Quantum Ontology: A Guide to the Metaphysics of Quantum Mechanics. Oxford University Press. Lipman, M. (2015). On Fine’s fragmentalism. Philosophical Studies, 172(12):3119–3133. Lipman, M. (2020). On the fragmentalist interpretation of special relativity. Philosophical Studies, 177:21–37. Mander, W. J. (2011). British Idealism: A History. Oxford University Press. Markosian, N. (2016). Time. In Zalta, E. N., editor, The Stanford Encyclopedia of Philosophy. Metaphysics Research Lab, Stanford University, fall 2016 edition. Maudlin, T. (2019). Philosophy of Physics: Quantum Theory. Princeton University Press. McDowell, J. (1994). Mind and World. Harvard University Press. McTaggart, J. M. E. (1908). The unreality of time. Mind, 17:457–473. McTaggart, J. M. E. (1921). The Nature of Existence. Cambridge University Press. Menzies, P. and Beebee, H. (2020). Counterfactual theories of causation. In Zalta, E. N., editor, The Stanford Encyclopedia of Philosophy. Metaphysics Research Lab, Stanford University, spring 2020 edition. Merricks, T. (2001). Objects and Persons. Oxford University Press. Mill, J. S. (1865). An Examination of Sir William Hamilton’s Philosophy. Longmans, Green, and Co. Miller, A. (2014). Realism. In Zalta, E. N., editor, The Stanford Encyclopedia of Philosophy. Metaphysics Research Lab, Stanford University, winter 2014 edition.

bibliography 231 Moltmann, F. (2003a). Nominalizing quantifiers. Journal of Philosophical Logic, 32(5): 445–481. Moltmann, F. (2003b). Propositional attitudes without propositions. Synthese, 35(1): 77–118. Moltmann, F. (2013). Abstract Objects and the Semantics of Natural Language. Oxford University Press. Moltmann, F. (2017). Natural language ontology. In Oxford Encyclopedia of Linguistics. Oxford University Press. Montague, R. (1974). The proper treatment of quantification in ordinary English. In Thomason, R. H., editor, Formal Philosophy. Yale University Press. Moore, A. (1997). Points of View. Oxford University Press. Moore, A. (2003). Ineffability and nonsense. Proceedings of the Aristotelian Society, 77: 169–193. Mumford, S. (2006). The ungrounded argument. Synthese, 149:471–489. Nagel, T. (1986). The View from Nowhere. Oxford University Press. Neale, S. (1990). Descriptions. MIT Press. Neale, S. (1993). Term limits. Philosophical Perspectives, 7:89–124. Pearce, K. (2017). Mereological idealism. In Pearce, K. and Goldschmidt, T., editors, Idealism: New Essays in Metaphysics. Oxford University Press. Pelczar, M. (2015). Sensorama: A Phenomenalistic Analysis of Spacetime and its Contents. Oxford University Press. Pelczar, M. (2019). Defending phenomenalism. Philosophical Quarterly, 69(276):574–597. Perry, J. (1990). Self-notions. Logos, pages 17–31. Pinkard, T. (2002). German Philosophy 1760–1860: The Legacy of Idealism. Cambridge University Press. Pollock, J. (1974). Knowledge and Justification. Princeton University Press. Priest, G. (2002). Beyond the Limits of Thought. Oxford University Press. Prior, A. (1967). Past, Present and Future. Oxford University Press. Prior, A. (1971). Objects of Thought. Clarendon Press. Pryor, J. (2000). The skeptic and the dogmatist. Noûs, 34(4):517–549. Pryor, J. (2007). Reasons and that-clauses. Philosophical Issues, 17(1):217–244. Rayo, A. (2013). The Construction of Logical Space. Oxford University Press. Rayo, A. (2014). Reply to critics. Inquiry, 57(4):498–534. Reich, K. (1998). Die Vollständigkeit der kantischen Urteilstafel. Felix Meiner Verlag, 3rd edition. Rescher, N. (1973). Conceptual Idealism. Blackwell. Ripley, D. (2013). Revising up: strengthening classical logic in the face of paradox. Philosophers’ Imprint, 13(5):1–13. Robinson, H. (2020). Dualism. In Zalta, E. N., editor, The Stanford Encyclopedia of Philosophy. Metaphysics Research Lab, Stanford University, fall 2020 edition. Rosefeldt, T. (2008). ‘That’-clauses and non-nominal quantification. Philosophical Studies, 137:301–333. Rosen, G. (1994). Objectivity and modern idealism: what is the question? In O’LearyHawthorne, J. and Michael, M., editors, Metaphysics in Mind, pages 277–319. Kluwer. Rosen, G. and Dorr, C. (2002). Composition as a fiction. In Gale, R., editor, The Blackwell Guide to Metaphysics, pages 151–174. Blackwell. Sampson, M. (2018). Solving the problem of creeping minimalism. Canadian Journal of Philosophy, 48(3–4):510–531.

232 bibliography Schaffer, J. (2009). On what grounds what. In Chalmers, D., Manley, D., and Wasserman, R., editors, Metametaphysics: New Essays on the Foundations of Ontology, pages 347–383. Oxford University Press. Scharp, K. (2013). Replacing Truth. Oxford University Press. Scharp, K. (2018). Philosophy as the study of defective concepts. In Burgess, A., Cappelen, H., and Plunkett, D., editors, Conceptual Engineering and Conceptual Ethics, pages 396–416. Oxford University Press. Schiffer, S. (1987). Remnants of Meaning. MIT Press. Schiffer, S. (2003). The Things we Mean. Oxford University Press. Sidelle, A. (1989). Necessity, Essence, and Individuation: A Defense of Conventionalism. Cornell University Press. Sidelle, A. (2010). Modality and objects. Philosophical Quarterly, 60(238):109–125. Sider, T. (2009). Ontological realism. In Chalmers, D., Manley, D., and Wasserman, R., editors, Metametaphysics: New Essays on the Foundations of Ontology, pages 384–423. Oxford University Press. Sider, T. (2011). Writing the Book of the World. Oxford University Press. Sider, T. (2013). Against parthood. In Bennett, K. and Zimmerman, D., editors, Oxford Studies in Metaphysics, volume 8, pages 237–293. Oxford University Press. Simons, P. (2013). Relations and idealism: on some arguments of Hochberg against trope nominalism. dialectica, 68(2):305–315. Smithson, R. (2021). Edenic idealism. Australasian Journal of Philosophy. Stang, N. (2017). Transcendental idealism without tears. In Goldschmidt, T. and Pearce, K., editors, Idealism: New Essays in Metaphysics, pages 82–103. Oxford University Press. Strawson, P. (1959). Individuals: An Essay in Descriptive Metaphysics. Methuen. Thomasson, A. L. (1999). Fiction and Metaphysics. Cambridge University Press. Thomasson, A. L. (2007). Ordinary Objects. Oxford University Press. Thomasson, A. L. (2016). What can we do when we do metaphysics? In D’Oro, G. G. and Overgaard, S., editors, The Cambridge Companion to Philosophical Methodology, pages 435–458. Cambridge University Press. Trueman, R. (2021). Idealism and the identity theory of truth. Mind, 130(519):783–807. Turner, J. (2011). Ontological nihilism. Oxford Studies in Metaphysics, 6:3–54. Uzquiano, G. (2018). Quantification, inference, and ontology. Analysis, 78(2):303–315. van Inwagen, P. (1977). Creatures of fiction. American Philosophical Quarterly, 14(4): 299–308. van Inwagen, P. (1990). Material Beings. Cornell University Press. Velleman, J. D. (2000). The Possibility of Practical Reason. Oxford University Press. Vetter, B. (2015). Potentiality. Oxford University Press. von Fintel, K. and Matthewson, L. (2008). Universals in semantics. Linguistic Review, 25:139–201. White, R. (2006). Problems for dogmatism. Philosophical Studies, 131:525–557. Williamson, T. (2008). The Philosophy of Philosophy. Blackwell. Williamson, T. (2013). Modal Logic as Metaphysics. Oxford University Press. Wilson, J. M. (2014). No work for a theory of grounding. Inquiry, 57(5–6):535–579. Winkler, K. P. (1994). Berkeley: An Interpretation. Oxford University Press. Wittgenstein, L. (1921). Tractatus Logico-Philosophicus. Harcourt. Wolff, R. P. (1970). In Defense of Anarchism. Harper and Row. Worsnip, A. (2018). The conflict of evidence and coherence. Philosophy and Phenomenological Research, 96(1):3–44. Worsnip, A. (2021). Fitting Things Together. Oxford University Press.

bibliography 233 Yetter-Chappell, H. (2018). Idealism without God. In Goldschmidt, T. and Pearce, K., editors, Idealism: New Essays in Metaphysics. Oxford University Press. Young, J. O. (2018). The Coherence Theory of Truth. In Zalta, E. N., editor, The Stanford Encyclopedia of Philosophy. Metaphysics Research Lab, Stanford University, fall 2018 edition. Zalta, E. N. (1999). Principia metaphysica. Available at http://mally.stanford.edu. Zimmerman, D. (2005). The A-theory of time, the B-theory of time, and “taking tense seriously”. dialectica, 59(4):401–457.

Index A-theory of time 42, 43 Adams, Robert 28 n.11 agency 188, 223 alienation 186–7, 201 antirealism 53–4 asymmetry thesis 181 Bad Guy example 193 banana example 23 benevolent communicator example 120–1 Berkeley, George 4 n.1, 24 Big Question 1–3, 61, 87, 203–8 idealist answer 2–3, 203–8 naturalistic answer 2 theistic answer 2 Blackburn, Simon 51, 54 Carnap, Rudolf 33, 70, 158–9, 215 carving 33–6 Cassirer, Ernst 159 n.23 cheese example 24 Chomsky, Noam 132 n.7 cognitive leaps 127–8 Cohen, Herrmann 159 n.23 compatibility constraint, see constraint, compatibility complete effability thesis 123 see also structural effabilty thesis composition 31–33, 176 compositional semantics 95 concepts application of 150 hardwired 149 inescapable 149–57 inevitable 150, 153, 217 logical 151–4 normative 154 of agency 154–5 conceptual circle 147 conceptual engineering 158 conceptual fixed points 159 conceptual repertoire 150 conceptual truth 145, 224 conditionals 35–8 counter-conventional 35

counter-substratum 35 indicative 36 subjunctive 35, 36 constitutivism 155 n.18, 223–4 constraint on reality 75 constraint vs. construction 14, 57, 62, 76, 88, 136, 216, 217 constraints 8–12, 209–11 compatibility 11–12, 209, 210–1 explicitness 9–11, 56 n.38, 209 support 11, 209–10 constructive role 140, 220 content ineffable fact 65, 111–5 context sensitivity 100–5 conventionalism 31–8 corrective role 140, 220 Davidson, Donald 130 n.6 decontextualization 130 deep ignorance 214 deep result 147, 160, 162, 168–9 defeat (epistemic) 153 n.14, 175, 196–7 rebutting 175 undercutting 175 demonstratives 101–3 dependence 10–1, 37–8 carving 35 causal 10 counterfactual 10 range 76, 87, 136, 209 substratum 35 truth 76, 87, 136 dependence problem 11, 38, 56 n.38, 209 dogmatism 196 Dummett, Michael 58 n.40, 59 Dyke, Heather 78 n.13 Einheuser, Iris 33–8, 99 n.8 Eklund, Matti 150 n.11, 157 n.19, 159, 191, 192–5 enlightenment 201 Enoch, David 155 n.18 existentialism 201, 208 explanation causal 29 constitutive 29

236 index explicitness constraint see constraint, explicitness explicitness-compatibility tension 34, 51 expressivism 52–3 externalism 72 fact-terms 68, 92, 93–5 facts 16, 52, 92, 155–6, 163–8 as things 16, 80 existence of 78–80 expressive 53 descriptive 53 incommunicable 120 ineffable 64–7, 85–6, 111, 115–8 mind-dependent 56 mortal 119 perspectival 41–4, 47, 191 representation of 115–6 see also representation structure of 63 facts* 84–5, 155–6, 163–8 Fahrtwind example 132 fallibilism 182 n.10 Ferrero, Luca 155 n.18 Field, Hartry 153 n.14, 175–7, 196 Filcheva, Krasimira 74 n.10 Fine, Kit 16, 44–9, 54 n.36, 199 first-personalism 46–7 Foley, Richard 188 form of thought see thought, form of Foster, John 22 fragmentalism 44–9 Frege, Gottlob 63 n.2, 142, 146 n.9 Gaskin, Richard 56 n.38 goalpost 189 God 2, 5, 21, 24–26, 115, 150, 185 Goodman, Nelson 33 grounding 5, 9–10 of identity and difference 39–44 Guyer, Paul 7 harmony 65–7, 111–5 complete 65–66, 112, 124–6 structural 65–66, 112 hereness 41–44 Horstmann, Rolf-Peter 7 human beings 117, 205–6 humanity 117, 204–6 Hume, David 51 n.31 idealism absolute 217 n.5 alethic 16, 112

broad 5–6 characterization of 4–8 classical 4 conceptual 76 constraints on 8–12 linguistic 56 n.38 main options 12–7 minimal theistic 21, 25, 52 ontological 16, 112 panpsychic 21 strong 6 weak 6 ignorance 214 immanent metaphysics see metaphysics, immanent immanent metaphysics program 223 immanent rationality see rationality, immanent immanent stance 190–2, 194 incommunicability test 120 ineffable fact see fact, ineffable ineffability content 65, 111–5 object-permitting 122 structural 65, 73–5, 85–6, 111–5 inescapable concepts see also concepts, inescapable as insights 158 as traps 158 inquiry vs. inquiry* 85, 156 internalism 72, 91–5, 108–9 intersubjectivity 23–4 Jupiter’s moons example 142 Kant, Immanuel 9 n.7, 15 n.9, 159, 161, 201, 215 Koch, Anton Friedrich 38–44, 47, 49 Lange, Marc 29 n.14, 49 n.30 language-metaphysics gap 78, 81, 88, 139, 141, 183 Lewis, David 193 lexical addition 130 logic classical 151–3, 173–9 disagreement about 173 intuitionistic 151–3, 173–9 manifest image 54 maple syrup example 30 match of form and structure 63 materialism 4, 24, 26 matter 4 n.1 metaphysical centrality 13–14, 87 metaphysical deduction 161, 218

index 237 metaphysics as human activity 192 descriptive 198 esoteric 200 foundational 199 immanent 198–201, 215–21 naive 199 natural language 198 preliminary characterization of 4 revisionary 198 transcendent 219 Mill, John Stuart 27–28 Miller, Alexander 7 mind-dependence see dependence, mind minimal constraints 8–12, 21–2 see also constraints modality 106–8 Moltmann, Friederike 96 n.7 Montague, Richard 95 Nagel, Thomas 66 n.5 natural language ontology 198 neo-Carnapian positions 158, 215 neo-Kantian positions 159–61, 215–20 orthodox 159, 215 reformed 160–1, 215 non-referential terms 79, 93, 143 Nothing, the 79–80 numbers 142–4 objects as bundles of phenomena 23, 24 elusive 121–2 unobserved 23, 24 ontologese 192 ontology 16 paradoxes 152–3, 174 n.2 Pearce, Kenneth 33 n.20 peer disagreement 178–9 Pelczar, Michael 22, 28, 29 n.13 perception 22, 25 persistence 31–33 perspectivalism 43–4, 47 perspective 41–4 phenomenalism 22–31 categorical 27 hypothetical 27–31 analytical hypothetical 27 reductive hypothetical 27–28 potentials 28–31 polysemy 98 projection 51–2, 54

proper question 83, 86, 145, 156 properties 39–44, 213 incompatible 43, 44 indexical 41–2 mind-dependent 50–5 natural 193 object-dependent 40 qualitative 40–1 response-dependent 54–5 tensed 42 proposition 55–6, 60, 92 proposition-terms 92 propositional, the 83, 129–30, 162–8, 213–4, 222 Pryor, Jim 196 qualities primary vs. secondary 51 quantifiers see quantification quantification 68–73, 96–100 domain conditions reading 69, 96 external reading 70, 96 inferential reading 69–72, 97–100 internal reading 70, 97–100 Meinongean 105 quantum mechanics 20–1 quasi-realism 51–3 range-dependence see dependence, range rationality alienating 186 by one’s own lights 149–50, 184–5 egocentric 188 external 149–50, 185–6 immanent 186–92 transcendent 186–7 Rayo, Agustín 136 realism 52 anti-idealist 52 anti-antirealist 52 ardent 193 ontological 50, 87, 212 reality 14–17, 53–4, 216 as it is in itself 15 as the totality of facts 15–7, 53–4, 82–3 as the totality of things 15–7, 82–3 as world of appearances 15 vs. Reality 199 reasoning premise-circular 152 rule-circular 152 reference 68–9, 93, 95 referential picture of language 93 relativism 86

238 index representation aboutness sense 115 conceptual vs. non-conceptual 116 mirroring sense 115 Rescher, Nicholas 76 n.12 Rosen, Gideon 55 n.37 Sapir-Whorf hypothesis 33 n.19 semantic underspecification 98 semantic value 95 shallow result 147, 162 shmagency 154–5, 223 Sidelle, Alan 33 Sider, Ted 177 n.6, 193 Simons, Peter 7 solipsism light 44 special theory of relativity 49 n.30 subjectivity thesis 38–44, 49 substitution argument 68–9, 94–5 substratum 33–6 support constraint see constraint, support Strawson, Peter 198 structural effability thesis 74, 111 structural ineffability thesis 74 structurally ineffable fact 65, 73–5, 85–6, 111–5 structure of facts see fact, structure of table tennis example 24, 48 tense 44–9

terms 68–9, 93 de facto non–referential 93 constitutively non-referential 93 thereness 41–2 thought form of 63–6 singular 121 n.4 transcendent rationality see rationality, transcendent transcendent stance 190, 194 transcendental arguments 153, 161, 217–8 transcendental deduction 161, 218 traps causal 171 rational 171–9 truth analytic 142, 145 coherence theory of 59 n.42 conceptual 142, 145 epistemic theory of 57–60 identity theory of 60 n.42 mind-dependent 56 minimalist theory of 52 vs. truth* 85, 156, 165–6 truth-dependence see dependence, truth truth-preservation 152, 173 n.1 über-reality 45–9 Uzquiano, Gabriel 105 n.12 Williamson, Timothy 121 Wittgenstein, Ludwig 15–16