Understanding the Nature of the Body Model Underlying Position Sense [1 ed.] 9783832592332, 9783832544607

Citation preview

MPI Series in Biological Cybernetics No. 49, March 2017

Aurelie Saulton

Understanding the nature of the body model underlying position sense

Understanding the nature of the body model underlying position sense

Dissertation

zur Erlangung des Grades eines Doktors der Naturwissenschaften

der Mathematisch-Naturwissenschaftlichen Fakultät und der Medizinischen Fakultät der Eberhard-Karls-Universität Tübingen

vorgelegt von

Aurelie Saulton

Aus Moutiers-au-Perche, France Oktober 2016

1

Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available on the Internet at http://dnb.d-nb.de .

c

Copyright Logos Verlag Berlin GmbH 2017 All rights reserved. ISBN 978-3-8325-4460-7

Logos Verlag Berlin GmbH Comeniushof, Gubener Str. 47, 10243 Berlin Tel.: +49 (0)30 42 85 10 90 Fax: +49 (0)30 42 85 10 92 INTERNET: http://www.logos-verlag.de

Tag der mündlichen Prüfung: 30.01.2017 Dekan der Math.-Nat. Fakultät: Prof. Dr. W. Rosenstiel Dekan der Medizinischen Fakultät: Prof. Dr. I. B. Autenrieth 1. Berichterstatter: Prof. Dr. Heinrich H. Bülthoff 2. Berichterstatter: Prof. Dr. Hong Yu Wong Prüfungskommission: Prof. Dr. Heinrich H. Bülthoff Prof. Dr. Hong Yu Wong Dr. Tobias Meillinger Dr. Marc Himmelbach

3

I hereby declare that I have produced the work entitled “Understanding the nature of the body model underlying position sense”, submitted for the award of a doctorate, on my own (without external help), have used only the sources and aids indicated and have marked passages included from other works, whether verbatim or in content, as such. I swear upon oath that these statements are true and that I have not concealed anything. I am aware that making a false declaration under oath is punishable by a term of imprisonment of up to three years or by a fine.

Aurelie Saulton, den 13.10.2016

4

Acknowledgments First and foremost I would like to thank my supervisor Stephan de la Rosa for the supervision of this work and great support all along the duration of the thesis. I am very grateful to Heinrich Bülthoff for giving me the opportunity to do my Ph.D. at the Max Planck Institute for Biological Cybernetics. Professor Bülthoff managed to create a very nice research environment which I will take as an example in the future. I also would like to thank Dr. Hong Yu Wong, Dr. Betty Mohler, Dr. Matthew Longo, Dr. Tobias Meilinger and Dr. Marc Himmelbach for their help, constructive criticisms and collaboration during my Ph.D. I appreciated all of our fruitful discussions. I am also grateful to my colleagues and friends who contributed to make my time at the MPI memorable. I especially want to thank Trevor Dodds for his constant encouragement, love and support during my Ph.D.

5

6

Abstract To perform accurate actions such as object grasping, one needs to localize an object of interest in external space but also to know the configuration (posture and structure) of the effector one wants to act with. One hypothesis emitted by Longo and colleagues (Longo & Haggard, 2010; Longo, Long, & Haggard, 2012) is that proprioception must rely on a stored body model of the body’s metric properties. In other words, localization of the body in external space requires that immediate sensory signals concerning the body posture be combined with a stored representation of the body size and shape. The general aim of this thesis was to identify the specificity and characteristics associated to the body model underlying position sense. Chapter 1 investigated whether hand distortions were specific to the body model or shared characteristics with objects including a rake, a rectangular post-it pad and a square CD box. In spite of numerous similarities between hand and object representations, especially hand-like items, the magnitude of localization task distortions was greater for the hand than for objects. While the similarities between hand and object distortions suggest the presence of general mechanisms involved in localization task distortions (e.g. memory), the measurement of greater hand distortion suggests the presence of body specific principles in hand localization judgments. Chapter 2 aimed at understanding why distortions measured in Chapter 1 would be more similar across a rake and a hand than across a hand and other geometrical objects. Factors such as hand similarity, memory and somatosensation were tested. While those factors seem to contribute to hand distortions, they could not fully explain the larger magnitude of hand distortions measured in study 1. Chapter 3 found out that the greater magnitude of hand distortion could be explained by the inaccurate conceptualization/memory of the hand’s knuckle location. Interestingly, localization judgments associated to better conceptualized hand landmarks (e.g. the bottom of the finger, see Saulton et al., submitted) showed significantly similar distortions to objects. This result has challenged the view that hand distortions are specific to the hand when measured in localization tasks.

7

8

Table of Contents I.

Introduction ................................................................................................................. 11 I.1

The role of proprioception in position sense ....................................................... 11

I.2

A body representation underlying position sense ................................................ 11

I.3

Measuring the body model................................................................................... 12

I.4

The relation between the body model and somatosensory distortions ................ 12

I.5

Dissociation between body image and body model ............................................. 13

I.6

Thesis Overview: Aim and Structure of the Thesis ............................................. 14

I.6.1 Comparing hand and object distortions ............................................................ 14 I.6.2 Comparing implicit vs. explicit judgments of an item shape index ................. 15 I.7

Chapter organization. ........................................................................................... 16

I.7.1 Chapter 1: Objects exhibit body model like shape distortions ......................... 16 I.7.2 Chapter 2: The role of visual similarity and memory in body model distortions. 18 I.7.3 Chapter 3: Conceptual biases explain distortion differences between hand and object in localization tasks .......................................................................................... 20 II. General discussion ...................................................................................................... 21 II.1

What drives hand and object distortions in localization tasks? ........................... 25

II.1.1

Somatosensory distortions ............................................................................ 25

II.1.2

General mechanisms in hand distortions....................................................... 26

II.1.3

Same distortions, different causes ................................................................. 27

II.2

A multimodal version of the body model mediating position sense ................... 31

II.3

How many body representations do we need? ..................................................... 32

III.

Conclusion ............................................................................................................... 33

III.1 Supplementary material ....................................................................................... 34 IV.

Declaration of the contribution of the candidate ..................................................... 35

V. Chapter 1: Objects exhibit body model like shape distortions ................................... 36 VI.

Chapter 2: The role of visual similarity and memory in body model distortions ... 53

VII. Chapter 3: Conceptual biases explain distortion differences between hand and objects in localization tasks ............................................................................................... 77 VIII. Bibliography ............................................................................................................ 93

9

10

I.

Introduction I.1

The role of proprioception in position sense

Knowing where body parts are localized in space is fundamental for interacting with the environment e.g. grabbing objects, as well as for self-oriented actions e.g. scratching one’s nose. The information used to evaluate the position and movements of our body parts is in large parts due to sensation arising in proprioceptors such as muscle spindles, golgi tendons, skin and joint receptors (for complete review on proprioception see Proske & Gandevia, 2012). However, none of those receptors deliver purely positional information (Feldman & Latash, 1982). For instance, signals from muscles spindles, which are considered as one of the major kinesthetic sensors in the human body, only provide information about changes in joint angle (Goodwin, McCloskey, & Matthews, 1972; Proske & Gandevia, 2012). More precisely, when a muscle is stretched, the Central Nervous System (CNS) interprets the increase in spindle activity as a longer muscle and thus as a more flexed or extended joint. The most important evidence in favor of this claim is probably the illusions of forearm movement and changed position felt by subjects following vibrations of their elbow muscles (Goodwin et al., 1972). Although other senses like vision or touch (for vision, see: van Beers, Sittig, & Gon, 1999; van Beers, Wolpert, & Haggard, 2002; for touch see: Gordon & Soechting, 1995; Paillard & Brouchon, 1974) as well as additional central processing are known to contribute to limb position sense (e.g. Motor commands; Gandevia, Smith, Crawford, Proske, & Taylor, 2006; Walsh, Gandevia, & Taylor, 2010; Walsh, Proske, Allen, & Gandevia, 2013), it is not clear yet how the CNS integrates all this information to create an accurate estimate of our limb position in space.

I.2

A body representation underlying position sense

Body representations are known to play a fundamental role in perception and action. At the cognitive level, a fundamental distinction is generally made between two different higher order body representations: the body schema and the body image (de Vignemont, 2010; Paillard, 1999). Generally, the body schema indicates an ongoing, mainly unconscious integration of successive multisensory input (e.g. proprioceptive signals) used to update the body posture in contrast to the body image which generally refers to a conscious visual representation of the way the body appears from the outside, e.g. looking at the physical configuration of one’s own body in a mirror (de Vignemont, 2010; Gallagher, 1986; Haggard & Wolpert, 2005; Holmes & Spence, 2004). Recent work seems to indicate that there would be an additional body representation: the body model underlying position sense (Longo & Haggard, 2010). According to Longo and Haggard (2010), the essential contribution of this body model to the human position sense is the specification of the relative locations of body parts. Although the necessity for such a body model is debated (see discussion in Pagano & Turvey, 1995; Pagano & Turvey, 1992), this hypothesis provides an interesting solution onto how the CNS can 11

specify the location of an extremity in space (in line with this argument see: Gandevia & Phegan, 1999; Gandevia, Refshauge, & Collins, 2002).

I.3

Measuring the body model

Longo and colleagues (2010) suggested that localizing body parts in external space requires the combination of online afferent signals from proprioception (e.g. joint angles) with a stored body representation or body model of the body parts’ metric properties. In order to measure and identify the properties of this stored body model, Longo and Haggard (2010) used a localization task in which participants had to estimate in external space the localization of 10 landmarks on their occluded hand (the tips and knuckles of each finger). The authors then compared the relative distance between the judged locations of the hand landmarks (e.g. relative distance between the judged tip and knuckle of the middle finger) with the actual distance between those landmarks on the participant’s hand. Using this method, they created maps of perceived hand size/shape. Those maps were highly distorted, showing underestimation of finger length and overestimation of the spacing between knuckles. Those observations generalized across both hands, and across different hand orientations suggesting that distortions were not caused by a general foreshortening of the perspective or a motor bias (Longo & Haggard, 2010).

I.4

The relation between the body model and somatosensory distortions

This pattern of hand distortion (length underestimation relative to width overestimation) was often interpreted as a reflection of somatosensory anisotropies characterizing the hand in the cortical sensory homunculus (Longo, Azañón, & Haggard, 2010; Longo & Haggard, 2010, 2011). In favor of this interpretation, three major characteristic distortions of the hand were put forward by the authors as evidence for a relation between localization task distortions and somatosensation: 1) Length underestimation relative to width. This pattern has often been related to the decreased tactile acuity found on the dorsum of the hand along the length compared to hand width. In that sense, distortions underlying position sense would be similar to those underlying tactile perception of hand size (Longo, Mancini, & Haggard, 2015; Margolis & Longo, 2015; Mattioni & Longo, 2014). 2) Gradient in finger underestimation. Hand distortions measured in localization tasks have often been characterized by a significant decrease in finger length underestimation from the little finger to the thumb. It was often suggested that this gradient in finger size mirrors similar gradients of decreasing tactile acuity and somatosensory cortical territory from the radial to the ulnar side of the hand (Mattioni & Longo, 2014; Longo & Haggard, 2010). 3) Clustering of finger. Longo and colleagues showed a specific clustering pattern associated with the finger. There were significant similarities in distortions for the middle and the index finger (first cluster) as well as for the ring and the little finger (second 12

cluster) and finally the Thumb. This clustering of finger representation was associated with the segmental distribution of the cutenaous nerves in the limbs (Longo & Haggard, 2010) which is also represented under three cervical dermatomes1 (see C7, C8 and C6 on Fig.3, from Keegan & Garrett, 1948).

Fig.3. The dermatomes of the trunk and the upper extremity. Note the three cervical dermatomes C6, C7 and C8 crossing the hand. From Keegan and Garrett (1948).

I.5

Dissociation between body image and body model

Research suggests that the body model underlying position sense differs from the body image (Longo & Haggard, 2010, 2012). In contrast to the large distortions measured in the localization task, people were highly accurate in recognizing the exact shape and size of their actual hand through a large array of images depicting hands of various dimensions (template matching tasks) (Gandevia & Phegan, 1999; Longo & Haggard, 2010). Such an accurate recognition would not be expected if the body image and the body model were sharing the same distorted representation. Hence, larger distortions in the localization task than in the template matching have been interpreted in favor of a dissociation of the implicit body model from the body image (Longo & Haggard, 2010, 2012). On this view, the body model underlying somatosensation is distinct from broader aspects of cognition such as visually recognizing one’s own hand shape. Nevertheless, the dissociation measured between the body model and the body image requires further investigation. Indeed, some metric measures used to assess the body image have shown similar distortions to the one obtained in the localization task (see the line length task in Longo & Haggard, 2012). The line length task used by Longo et al (2012) constitutes an alternative measure to depictive methods used to assess the body image (e.g. template matching tasks). In this task, participants have to compare parts of their hand, e.g. finger length, with the size of lines displayed on a screen. As for the localization tasks, results showed that hand distortions were characterized by an A dermatome is an area of skin supplied by sensory neurons that arise from a single spinal nerve.

1

13

underestimation of length relative to width (mean length estimate < mean width) with a significant gradient in finger length underestimation from the thumb to the little finger. Although distortions were present in smaller magnitude than in the localization task, this result shows that the dissociation previously found between the body model and the body image might not be that categorical (Longo & Haggard, 2012).

I.6

Thesis Overview: Aim and Structure of the Thesis

It is important to note that the relation established between hand distortions in localization tasks and somatosensory distortions is only analogical. Indeed, no direct evidences were furnished by the proponents of the body model regarding a direct link between hand distortions and somatosensation in localization tasks. Hence other mechanisms than somatosensation could be responsible for hand distortions measured in localization tasks. In order to better understand the nature and implication of this new body representation we investigated further the specificity and characteristics associated to the body model underlying position sense. Two major approaches were used to achieve our goal: 1) comparing hand distortions with objects and 2) comparing implicit judgments performed in localization tasks with results from other tasks referring to more explicit judgments about an item’s shape (see Fig.2 p.13). Those two approaches are exposed in the next paragraphs and used in the three studies of this thesis. I.6.1 Comparing hand and object distortions In order to better comprehend the characteristics and specificity of hand distortions, we systematically compared hand distortions with a large variety of non-corporeal objects including a rake, a post-it pad, a CD box (Chapter 1) a rubber hand (Chapter 2) and simple L and T shapes (Chapter 3). If localization task distortions measured on the hand are body specific they should not be observed in objects void of somatosensory influences. Analysis used in the thesis. In order to compare the shape representation of each stimulus with one another, we quantified each item’s shape using its width to length ratio, referred to as its Shape Index (SI = 100 x width/length). The Shape Index is assumed to reflect the overall aspect ratio of an item (see “Methods,” Longo and Haggard 2012). We then normalized the Shape Index for each stimulus (NSI=Perceived Shape Index/Actual Shape Index). In Fig. 1, the width of an item is marked with a yellow line, and the length is marked with a red line. More details about analysis and statistical tests used in this thesis can be found in the method and analysis section of each chapter.

14

Fig.1. from left to right: hand, rake, post-it pad, square CD box, T shape and L shape. Item-centered width dimensions are indicated in yellow and item-centered length dimension in red.

I.6.2 Comparing implicit vs. explicit judgments of an item shape index Evidence shows that hand distortions measured in the localization task can also be found in more explicit measurement of hand shape (see line length task : Longo et al., 2012). These results suggest that localization task distortions are more widespread than initially expected. To investigate this aspect further, we contrasted results obtained in the localization task with three other tasks including a distance task, a conceptual task and a template matching task. All tasks are described below and represented in Fig.2. In contrast to the localization task, which constitutes an implicit measure of an item shape (the task does not explicitly demand for size information about the item, e.g. subjects are asked to point towards individual landmarks of an item. Size information is calculated afterwards by combining information form the individual trials), the three other tasks target more explicit information about the item’s shape including the memory of relative distance between landmarks (distance task) and semantic knowledge about its structural configuration (conceptual task). If localization tasks distortions were unique to a body model underlying somatosensation, they should not be measured in tasks targeting more explicit conceptual knowledge or visual memory of the items shape.

15

Localization task: Participants had to place a mouse pointer on a screen directly above where they felt (for the hand) or remembered (for objects) the location of specific landmarks presented on the occluded hand or the occluded object. This task is inspired from Longo et al (2010). Details about differences in methodology with Longo et al 2010 can be found in the method and discussion section of Chapter 1. Distance task: Participants had to indicate on a coordinate system, the remembered relative distances between landmarks located along the length (e.g. knuckle to tip of the middle finger) or the width of an item (e.g. knuckle from little to index finger). This task is inspired from a line length task used by Longo et al (2012). Details and methodological variations used in the distance task can be found in the methodological section of experiment 2 and 3 of Chapter 2. Template matching task: Participants saw pictures of an item (e.g. hand or object) and had to judge whether the item was more slender or wider than the original item. This task is inspired from Gandevia et al (1999) and Longo et al (2012). Details about this task can be found in the method section of Chapter 1. Conceptual task: Participants had to indicate the remembered location of the tips and knuckles of their finger on a black silhouette of their own hand. This task follows similar procedures to the one used in Margolis and colleagues (2015). Details about this task can be found in the methodological section of Chapter 3.

I.7

Chapter organization.

The first chapter of this thesis investigated whether shape index distortions measured in localization and template matching tasks were specific to the body model or shared characteristics with objects including a rake, a rectangular post-it pad and a square CD box. While larger distortions were measured for the hand in the localization task, surprising similarities were found between the rake and the hand distortions compared to other items. The goal of Chapter 2 was to understand why distortions measured in Chapter 1 would be more similar across a rake and a hand than across a hand and other geometrical objects. Factors such as hand similarity, memory and somatosensory feeling of hand shape were tested. Finally, Chapter 3 investigated whether larger hand distortions measured in localization tasks could be due to misconceptualized hand landmarks and whether hand-objects distortion difference could vanish when referring to better conceptualized hand landmarks. We will now summarize and discuss the findings of each chapter in more detail. I.7.1 Chapter 1: Objects exhibit body model like shape distortions Research question: In Chapter 1, our aim was to understand whether the dissociation found on the hand between localization and template matching tasks (Longo et al., 2010) could also be measured in non-corporeal objects. Because the larger distortions in the localization task than in the template matching task is a defining property for the body model underlying somatosensation, it is important to ensure that this task-specific 16

distortion is unique to the body. To address this point, we compared localization performance of the participant’s left hand and objects including a rake, a rectangular post-it pad and a square box with results of a template matching task involving more explicit recognition judgments of the same items (see Longo et al., 2012). Main results for Chapter 1 Interestingly, the hand, rake, post-it and box showed the tendency to be more distorted in localization tasks compared to template matching tasks. While all objects measured in localization tasks were characterized by larger length underestimation relative to width, only the rake item showed clear similarities with hand distortions. We will now report the most significant finding of this study with a strong emphasis on hand and rake similarities. Typical hand distortions found on the rake in localization tasks. Interestingly the rake item shared two major characteristic distortions associated with hand shape: (1) an overall overestimation of rake width and (2) an overall underestimation of branch length. As previously demonstrated for the hand (Longo et al 2010), rake distortions were preserved across orientation and significantly correlated when the rake was rotated 90° relative to the body suggesting the presence of item-centered distortions. Finally, both rake and hand localization task distortions were significantly larger than for other objects. Although hand distortions were still greater than the one measured on the rake, the similar direction measured in hand and rake distortions suggest that other factors than somatosensation e.g. memory or visual similarity could contribute to distortions in localization tasks. Dissociation between localization and template matching tasks results for the rake. In stark contrast to the large distortions characterizing rake representation in localization tasks, judgments in the template matching tasks were approximately veridical. Similar results were reported for the hand between localization and template matching tasks. Observing such distinct results for the rake item, suggests that other factors than somatosensory information might contribute to the differential distortions observed in hand shape between template matching task and localization tasks. For instance, two independent representations may be formed when participants are learning a spatial layout visually (McNamara 2003): one involved in localization tasks which encodes spatial relations between landmarks (distance, direction) and one supporting (holistic) scene recognition by means of visual memory (Shelton & McNamara, 2004). Hence, the dissociation obtained between body model and body image might simply be part of a more general dissociative process in spatial memory. Alternatively, differences in task function (recognition vs. localization) and complexity (easy vs. difficult) could contribute to the differential results obtained between the two tasks. Orientation sensitivity in hand magnitude distortion. Interestingly, we found out that the magnitude of hand distortions measured in the localization task was sensitive to hand orientation (upright vs. 90 degrees rotated posture). This was especially the case of 17

hand width overestimation which was not correlated across orientation and significantly diminished in the rotated posture. In contrast, finger length underestimation was rather well preserved across orientation. This result suggested that hand distortions were partly influenced by viewer-centered biases and not only driven by item-centered somatosensory processing of hand shape. Those results were also found for the right hand in Chapter 2 (for direct comparison between left and right hand see supplementary material of Chapter 2). Summary of Chapter 1: Overall, we have shown that distortions analogous to those measured on the hand i.e. larger length underestimation compared to width, could be found when participants had to indicate the remembered locations of different landmarks on non-corporeal objects (rake, post-it and CD box). Despite differences in the magnitude of distortions, the rake and hand items shared more distinct characteristics compared to other objects, e.g. larger magnitude distortions. One factor susceptible to explain the greater similarities between hand and rake distortions in localization tasks is their structural likeness (e.g. five fingers/ five tines). We therefore hypothesized that greater visual similarity between an item and a real hand could potentially contribute to an increase in localization task distortions. We investigated this aspect in Chapter 2. I.7.2 Chapter 2: The role of visual similarity and memory in body model distortions. Research question: In Chapter 2 our aim was to understand why distortions measured in Chapter 1 would be more similar across a rake and a hand than across a hand and other geometrical objects. This study comprises of three experiments. In Experiment 1, we investigated whether the similarity between hand and rake distortions could be due to body-related factors i.e. the visual similarity between the stimulus and a real hand. To answer this question we compared localization tasks results between a rake (rated as least similar to the real hand), a rubber hand (rated as most similar to the real hand) and the participants’ right hand. In Experiment 2, we investigated whether the similarity between hand and rake distortions could be due to non-body-related factors i.e. a memory effect, and compared hand, rake and a rubber hand distortion in a distance memory task. Finally, in Experiment 3, we controlled whether the distortions measured on the participant’s hand in the distance memory task could be behaviorally dissociated from distortions coming from the somatosensory feeling associated with one’s own hand. Main results of Chapter 2 Increasing visual similarity does not increase the magnitude of localization task distortions. Results of Experiment 1 showed no significant effect of increased visual similarity on the magnitude of localization task distortions: hand distortions were again, greater than those found on a rake and a rubber hand (note that rake and rubber hand distortions were not significantly different). It might be that the rubber hand used in our experiment is too different from a real hand (despite being rated as more similar to the 18

hand than the rake) to detect an effect of increased visual similarity on our results. Using a more realistic hand shape as in Longo et al., 2015, could increase the chance of finding closer distortions between the participant’s own hand and the stimulus. Similar magnitude of distortions for hand and object in a distance memory task. While localization judgments of individual landmarks revealed different distortions between hand and objects, memorizing relative distances between two landmarks (Experiment 2) generated a similar magnitude of distortions across the hand, the rubber hand and the rake. Although distortions were generally smaller than in the localization task, they were characterized by a similar pattern of significant length underestimation relative to width. Hence, memorizing relative distances between landmarks on hand and objects was associated with qualitatively similar distortions to localizing individual landmarks on the same items. Nevertheless, the presence of larger hand distortion in the localization task suggests that different mechanisms to memorizing distances on hand and objects are used in the localization task. Feeling the hand can increase the magnitude of hand distortions. In experiment 3 of this chapter, we identified one potential factor susceptible to explain larger hand distortions in the localization task: using the somatosensory feeling associated with one’s hand. We demonstrated that changing instruction from memorizing to feeling one’s hand increased the magnitude of hand distortions measured in the distance task. This finding could potentially generalize to the localization task results (for which similar difference in instructions were used between hand and objects) and contribute to explain why greater distortions were found on the hand compared to objects. However, hand distortions measured in localization task remained larger than hand distortions measured in the feeling condition of the distance task. Hence, additional factors other than instructions (feeling hand landmarks vs. memorizing objects landmarks) might explain hand and object distortion differences measured in localization tasks. Summary of Chapter 2: Chapter 2 aimed at understanding why distortions measured in Chapter 1 would be more similar across a rake and a hand than across a hand and other geometrical objects. Factors such as hand similarity (Experiment 1), memory (Experiment 2) and instructions (feeling vs. memorizing hand parts in experiment 3) were tested. Experiment 1 showed that increasing the level of visual similarity between a hand and a rake via a rubber hand stimulus, did not contribute to an increase in localization task distortions. While memory could contribute to the pattern of hand and object distortions measured in the localization task (large length underestimation relative to width), we showed it could not explain the extent of hand distortions measured in the localization task (Experiment 2). Finally, Experiment 3 showed that the somatosensory feeling of one’s hand could contribute to increasing hand distortions. Felt distances on the hand (distance task) were nevertheless significantly different from localizing felt landmarks on the hand (localization task). This could suggest that individual location judgments of hand landmarks (localization task) are processed differently from relative 19

distance judgments about the same landmarks (distance task). In direct line with this idea, recent pointing tasks have shown that absolute location judgments of hand knuckles were inaccurate (Longo et al., 2015). When indicating the location of the knuckles (via a baton) directly on their palm or another person’s palm, subjects systematically pointed towards the crease of the finger instead of the actual knuckle location. This systematic bias was interpreted as reflecting a general false belief about the structural representation of the hand. If similar misunderstanding about the knuckle location occurs in our localization task, it could explain why we regularly observed larger hand distortions in localization tasks. We tested this aspect in Chapter 3. I.7.3 Chapter 3: Conceptual biases explain distortion differences between hand and object in localization tasks Research question: In Chapter 3, we investigated whether hand-objects distortions differences measured in localization tasks of Chapter 1 and 2 could be due to the presence of a general false belief about hand knuckle locations (referred to later as “conceptual bias”). To demonstrate the presence of a conceptual bias in the hand knuckle location, we used a conceptual task in which participants had to indicate the remembered location of the tips and knuckles of their finger on a black silhouette of their hand (Experiment 1). In Experiment 2, we examined whether the conceptual bias could – in theory – explain distortion differences between hands and objects. To do so, we used the subjects’ own conceptual configuration of their hand landmarks (from the conceptual task) to analyze their hand localization task data. We then compared those results with previous work from Chapter 1 and 2 showing rake distortions in localization tasks. This allowed us to see whether a numerical correction for the conceptual bias would affect the differences between distortions in hands and objects. Finally, in Experiment 3, we directly measured the influence of conceptual biases on localization performance differences between hands and objects (T and L shapes). To manipulate the amount of conceptual bias on the hand, participants pointed to different types of hand landmarks known for being related to a different amount of conceptual error i.e. the knuckles (large conceptual biases) or bottom of the finger (small conceptual biases; see supplementary material of Chapter 3). We predicted the observation of similar distortions between hands and objects for hand landmarks with small conceptual biases (bottom of the finger) and different distortions for hand landmarks with large conceptual biases (knuckles). Main results of Chapter 3 Larger hand distortions induced by misconceptualized knuckles. Results of Experiment 1 confirmed that participants’ conceptual knowledge about the spatial location of the hand’s knuckles was inaccurate: knuckles were all significantly biased towards the bottom of the finger. In Experiment 2, we mathematically corrected for these hand conceptual biases in the localization task. When comparing this corrected estimate of 20

hand shape with rake estimates obtained in similar localization tasks (data from Chapter 1 and 2) we found no statistically significant differences between the hand and the rake. In fact, calculating the Bayes Factor provided evidence in favor of the null hypothesis (i.e. hand and rake distortions are similar). This result therefore suggests that conceptual biases of the hand could account for localization differences between the hand and the rake. Similar hand and object distortions associated with well conceptualized hand landmarks. In experiment 3, we found that hand landmarks associated with small conceptual biases (i.e. bottom of the finger) abolished distortion differences between objects and hands in localization tasks. In this condition, length but also width magnitude estimates were significantly correlated across participants between hand and objects (L and T shapes). In contrast, landmarks associated with large conceptual biases (i.e. knuckles) preserved hand –object distortions differences found in previous work. These results speak strongly in favor of a conceptual bias in hand landmarks to explain larger distortions in hand compared to object in localization tasks. Summary of Chapter 3: The goal of Chapter 3 was to determine whether inaccurately conceptualized hand landmarks (e.g. knuckles) could be responsible for the larger distortions measured on the hand in localization tasks. We demonstrated that comparing objects (hand-like items e.g. rake or non-hand-like items e.g. L and T shapes) with hand shape distortions on the basis of localization judgments performed on wellconceptualized hand landmarks (bottom of fingers) led to a similar magnitude of distortions in localization tasks. More precisely length but also width estimates were significantly correlated across participants for the Hand Bottom condition and simple geometrical forms. This result suggests the presence of similar perceptual/ cognitive processes underlying hand and object localization judgments. This last chapter therefore furnished direct evidences in favor of non-specific hand shape distortions measured in localization task.

II.

General discussion The general aim of this thesis was to identify the specificity and characteristics associated to the body model underlying position sense. Previous research initially suggested that hand map distortions measured in a localization task could retain distortions characteristic of the hand in early somatosensory maps (Longo & Haggard, 2010). According to this view, the relative proportions of this hand model are distorted in a way that reflects the relative sensitivity of the different skin regions (Longo et al., 2010; Longo, Long, & Haggard, 2012; Longo & Haggard, 2011). Importantly, this distorted representation is dissociated from higher level cognitive processes involved in our subjective understanding of our body dimensions (body image: Longo & Haggard, 2010, 2012). For those reasons, previous work considered the body model to be an implicit, purely perceptual body representation which is dissociated from wider aspects of 21

cognition (see: Longo & Haggard, 2010; Longo, Mattioni, & Ganea, 2015). In contrast to this idea, this thesis provided evidences suggesting that hand distortions are more widespread than initially thought (e.g. hand distortions found in cognitive tasks: Chapter 2 and 3), unlikely to be hand specific (e.g. similar distortions found on objects: Chapter 1, 2 and 3) and probably driven by multiple causes (e.g. evidence for memory and conceptual biases in hand and object size representation: Chapter 2 and 3). First, as mentioned above, we found distortions that were qualitatively similar to those found in the localization task using tasks assumed to be void of somatosensory influences. For instance, participants underestimated the length of their fingers and overestimated the distances between their knuckles both in the distance memory task (Chapter 2: task which targets memorized distances between hand and object landmarks) and the conceptual task (Chapter 3: task which targets the conceptual spatial knowledge of landmark location on a hand silhouette). Similar conclusions can be drawn for other typical hand characteristics e.g. gradient in finger size underestimation, also measured in non-proprioceptive tasks assessing hand size representation (e.g. line length task in: Longo & Haggard, 2012; silhouette task in: Longo et al., 2015 and Saulton et al., Chapter 3; and verbal localization tasks in: Mattioni & Longo, 2014). Hence, typical hand distortions are not limited to proprioceptive localization tasks. A second important aspect of this thesis is the finding that hand distortions measured in localization tasks are not specific to the body. Similar biases were indeed measured in localization tasks for hands and objects, especially a rake item whose distortions persisted in different orientations and were characterized by an overestimation of width and an underestimation of length (Chapter 1 and 2). Interestingly, those rake distortions were observed in a similar magnitude to the hand in the visual template matching task and the distance memory task (see Fig.4). Altogether, these results already suggest that similar processes might be involved in certain types of size judgments associated to hands and objects.

22

Fig.4. Summary graph comparing hand and rake distortions across tasks. Normalized Shape Index for the rake and hand in the localization task (Chapter 1), distance memory task (Chapter 2) and templated matching task (Chapter 1). A normalized Shape Index of 1 corresponds to the actual size of the item. A normalized shape index greater or smaller than 1 means the item is distorted. The rake and hand depictions above the figure represent from right to left the increase in distortion from the actual item size to its represented distorted size in the other tasks. Error bars indicate +/- 1 SE from the mean.

Nevertheless, differences between the magnitude of distortion found on the rake and the hand were found significant in the localization task (Chapter 1, see Fig.4 above) suggesting that some of the processes involved in hand localization judgments might be body-specific or driven by other cognitive biases. In Chapter 2 and 3 of this thesis, we therefore investigated potential factors susceptible to explain hand vs. rake distortions differences in the localization task including visual similarity, memory, change in instructions and a conceptual bias associated to knuckle positioning. The last factor i.e. conceptual bias (Chapter 3), provided the most promising explanation for hand vs. object distortion differences within the localization task. In particular, similar distortions were found between hands and objects (rake and L/ T shapes) after correcting for the conceptual bias (knuckle misrepresentation) associated with the hand (see localization correction on Fig.5). This result favors the idea that distortions measured in localization tasks reflect a general cognitive/perceptual bias that can be measured both with hands and objects rather than specific somatosensory distortions.

23

Fig.5. Summary graph comparing hand and rake distortions across tasks after correcting for the hand conceptual bias in the localization task. Normalized Shape Index for the rake and hand in the corrected localization task (Experiment 1, Chapter 3), distance memory task (Chapter 2) and templated matching task (Chapter 1). A normalized Shape Index of 1 corresponds to the actual size of the item. A normalized shape index greater or smaller than 1 means the item is distorted. The rake and hand depictions above the figure represent from right to left the increase in distortion from the actual item size to its represented distorted size in the other tasks. Error bars indicate +/- 1 SE from the mean.

Finally, a critical review of findings (from this thesis and other studies) suggests that hand distortions are unlikely to be driven by a single cause and instead, may be influenced by multiple cognitive and perceptual factors. For instance, results showed that the magnitude of hand distortions measured in localization tasks was sensitive to a change in hand orientation (upright vs. rotated orientation; see Chapter 1 and 2 or p.17 of the introduction), the conceptual representation of the hand, i.e. the amount of conceptual bias associated to different hands’ landmarks (Chapter 3), finger positioning (finger spayed vs. pressed together; Longo, 2015), cueing of localization judgments (whether landmarks are cued using touch or verbal labels see: Mattioni & Longo, 2014) and visual feedback (localization tasks performed sighted vs. blindfolded; Longo, 2014). Attributing the magnitude of the distortions observed on the hand only to somatosensation might therefore be critical, mostly when factors such as hand posture and visual feedback show interferences with length and width estimates of the hand. Hence, the body model underlying position sense might be a more complex and integrated multisensory body representation than previously expected (see section on p.32: A multimodal version of the body model mediating position sense). Although our results tend to favor the interpretation of a general cognitive/perceptual bias underlying hand and object distortions in localization tasks, we do not reject the existence of a distorted body model underlying position sense. Rather 24

our findings suggest that the localization task distortions are not specific to the body and consequently localization tasks results cannot be reliably linked to somatosensation. Hence, our thesis point to a methodological challenge in the interpretation of current localization task results. In the next section, we will discuss both body-specific and general factors susceptible to explain differences and similarities between hand and object distortions measured in localization tasks. We will start by demonstrating the interests and limitations of three arguments advanced in favor of hand distortions as being driven by somatosensory processes (those arguments were also mentioned in the introduction p.10) and successively present more general alternative explanations to localization task distortions.

II.1 What drives hand and object distortions in localization tasks? II.1.1 Somatosensory distortions To better understand the nature of hand distortions measured in localization tasks, Longo and haggard (2010) investigated the relation between length underestimation of the 5 finger and found significant similarities between fingers grouped into three different clusters: index and middle finger, ring and little finger and the thumb. This clustering of finger representation was interpreted as mirroring the segmental distribution of the cutaneous nerves in the arm (see Fig.3, p.11). Although this idea is interesting, it has never been replicated. We tried to investigate this aspect on our hand localization data (we conducted a Principal Component Analysis on finger length underestimation of the 5 digits for 66 subjects, see supplementary material on p.36) and found that the first component explained 76 % of the variance. Adding further components did not decrease the variance substantially (see scree plot on p.36). This suggests that the variance in the data can be well accounted for by a single component. More importantly, the finger loadings were all high with this first component suggesting that all fingers contribute about equally to explaining the variation. In other words, we did not find a clustering pattern of fingers across several principal components, as it has been reported previously (Longo & Haggard, 2010). Hence, there is inconclusive evidence regarding the relation between sensory afferents of the hand and finger underestimation. It is important to remember that correlation is not causation. Hence, those groups of fingers could have emerged for other reasons e.g. the visual configuration or angles associated to each group of finger in one specific localization task. Hand distortions measured in localization tasks have often been described by a significant decrease in finger length underestimation from the little finger to the thumb. This pattern was interpreted as mirroring gradients of decreasing tactile acuity and somatosensory cortical territory from the radial to the ulnar side of the hand (Longo et al., 2010; Longo & Haggard, 2010, 2011). However, this specific finger length pattern was also measured in tasks assumed to be relatively void of somatosensory influences i.e. the conceptual silhouette task used in Chapter 3 (for other cognitive tasks reporting this bias see Longo & Haggard, 2012; Longo et al., 2015) and was not always detected in 25

proprioceptive localization tasks (see localization task results of experiment 1 in Chapter 3 of this thesis; for another example see Mattioni & Longo, 2014, tactile condition). Hence this gradient in finger size underestimation may well be related to cognitive or conceptual processes associated to the structural configuration of the hand. Another characteristic pattern of hand distortions is the relative length underestimation compared to hand width. This feature of hand shape representation has often been related to biases found in tactile perception (Longo et al., 2010, 2012, Longo & Haggard, 2010, 2011) especially the observation that perceived tactile distances are smaller when orientated along the length of the body than across its width (Longo & Haggard, 2011; Weber, 1996). Once more, this characteristic was found on all objects measured in the localization task. Although the magnitude of this pattern often remained greater for hands than objects, it is important to verify that hand distortions are not better predicted/explained by other more general mechanisms e.g. spatial biases in memory or visual perception. Verifying this claim becomes especially important since the differences between hand and object distortions were shown to vanish when using correctly perceived hand landmarks (bottom of fingers instead of knuckles, see Chapter 3). In this case, we wish to remind the reader that hand and object distortions were correlated or observed in similar magnitude (see results of Chapter 3). From this result, it becomes extremely difficult to conclude that distortions are specific to the hand. II.1.2 General mechanisms in hand distortions One possible interpretation for length underestimation relative to width in hand and object refers to visual distortions associated with bisecting segments. Studies investigating the vertical-horizontal and bisecting line illusions report length underestimation relative to width estimates in the case of objects and figures depicting vertical segments crossing horizontal ones (Chapanis & Mankin, 1967; Finger & Spelt, 1947; Hamburger & Hansen, 2010). In some sense, the L and T-shape stimuli used in Chapter 3 can be considered as representative of such illusions. Although the origin of the bisecting line illusion is still debated (see Wolfe, Maloney, & Tam, 2005), it has often been attributed to two factors. The first factor is a length bisection bias (Finger & Spelt, 1947) and the second one is the anisotropy between vertical and horizontal segments due to the oval shape of our visual field (framing effect of the visual field, see verticalhorizontal illusions in Künnapas, 1955). Those visual biases could potentially also take place in our localization task for the hand (fingers bisecting the palm) and objects (e.g. rake: branches bisecting the rake handle) and explain why we measured a significant length underestimation relative to width. Another possible mechanism that could be common to object and hand distortions in localization tasks is memory. Spatial biases associated to the categorical perception of memorized landmarks are known to depend on whether the judged landmarks belong to different or similar categories of landmarks (Huttenlocher, Hedges, Corrigan, & Crawford, 2004; Huttenlocher, Hedges, & Duncan, 1991; Huttenlocher & Lourenco, 26

2007). For instance, distances stored in memory between entities of the same categories (e.g. known cities on map) are perceived relatively smaller compared to distances between entities of different categories (unknown cities, Tversky, 1992). Semantically, fingers constitute a separate body part category (Enfield, Majid, & Van Staden, 2006). Moreover, in all our studies, the tips and knuckles of each finger were labelled as being part of the same finger while the knuckles of each finger were labelled as being part of different fingers. Similar labelling strategies were used for landmarks localized on the rubber hand and the rake item in Chapter 2. Accordingly, one would expect that hand and object landmarks belonging to the same category (e.g. knuckle and tip of one finger for the hand; bottom and top of a single branch for the rake) would be perceptually attracted, resulting in finger/branch length underestimation while landmarks belonging to a different category (e.g. knuckles of adjacent fingers for the hand or bottoms of adjacent branches for the rake) would be perceptually repelled, leading to a significant overestimation of the inter-knuckle/inter-branch distances. While this hypothesis requires more investigations to be confirmed, it is in line with the results observed both in the distance memory task and the localization task. II.1.3 Same distortions, different causes Observing similar distortions between hands and objects does not necessarily mean that hand and object distortions result from the same causes. For instance, it has been shown that the vertical horizontal illusion could be found both in the tactile and visual modality (Gentaz & Hatwell, 2004; Heller & Joyner, 1993). Researchers found out that processes involved in the distortions were partly due to specificity associated to each modality (e.g. anisotropy due to the shape of the visual field and anisotropy of the tactile receptive fields) and partly due to similar processes (the role of bisection see: Gentaz & Hatwell, 2004.; Hamburger & Hansen, 2010; Heller & Joyner, 1993). Similar principles could occur when comparing hand and object distortions in localization tasks. In addition to the influence of common/general mechanisms, the significant length underestimation relative to width might reflect specific characteristic of our somatosensory system in the case of the hand and specific features of visual memory in the case of objects. However, in the current experimental context, it seems difficult to disentangle the two hypotheses. Hence different methodologies might be necessary to understand the specificity of hand distortions measured in localization tasks. If not different methodologies, there is evidence to think that hand and objects biases measured in localization tasks would be more clearly differentiated if calculated on the basis of the error between the actual and estimated locations of each landmark (Gross, Webb, & Melzack, 1974) rather than analyzed using the relative estimated distance between landmarks (Longo & Haggard, 2010). Calculating the relative distance between estimated landmarks can hide important information about the location judgments associated to each landmark. For instance, it does not take into account the fact that the actual estimated position of each landmark might vary with time (Gross et al., 1974). Multiple studies have shown that arms or hands lying in upright posture present a 27

systematic drift in absolute landmark location error (Brown, Rosenbaum, & Sainburg, 2003; Desmurget, Vindras, Gréa, Viviani, & Grafton, 2000; Gross et al., 1974; Wann & Ibrahim, 1992). For instance, the arm is systematically perceived as closer from the body (Gross et al., 1974). It has been suggested that the drift in limb position is partly due to muscle receptor adaptation (Tsay, Giummarra, Allen, & Proske, 2016). When our limbs remain static, muscle spindles remain active and dependent on the contraction history of the muscles. This property is referred to as muscle thixotropy (Proske, Tsay, & Allen, 2014). So far, the relation between the drift in hand position and finger length underestimation measured in localization tasks remains unexplored. This is problematic as both aspects might reflect different interpretations of the same underlying mechanisms. In other words, finger length underestimation might be related to the declining afferent discharge in muscle proprioceptors.

Fig.6. Schematic representation of the bias in hand and object location judgments in upright orientation. Hands and objects present similar length underestimation (see red lines) but are perceived at different absolute locations. In line with the literature, the hand in upright position is perceived closer to the body.

This idea is important for interpreting hand vs. object localization task distortion differences. If the drift in hand perceived position is body specific, then we should not measure it with objects. Imagine a case for which the relative estimated distance between the top and bottom of an object’s length dimension is underestimated in a similar amount to the estimated distance between the tip and knuckle of the hand fingers (see example of represented estimated length on Fig.6). Despite similar relative distortion, there could nevertheless be a significant difference in absolute location between landmarks located along the length of the hand and the object. In line with the literature, we would especially expect to observe a significant shift in the hand’s absolute perceived position. If this shift is body specific, then it should not be observed for objects. We tested this hypothesis by reanalyzing the data from all our studies, using the absolute localization error associated with each of the item’s landmarks, i.e. the distance between the actual and estimated locations (see Gross et al., 1974). Interestingly, calculating the absolute positioning error of hand landmarks along the y axis (item28

centered y axis, Fig.7) showed significant differences for all comparisons of hand and objects measured in our thesis (rake, box, Post it pad, L and T shapes). We especially found a large drift in hand position which was not present for objects (see representation Fig.8). This drift in perceived hand position was observed in a similar manner for the left and right hand and was independent of orientation. For both types of hand posture (upright or rotated), all hand landmarks drifted towards the wrist. This drift increased as a function of the duration that the hand remained concealed from vision (see Fig.9). As suggested previously, this drift in hand location might be due to the fact that information available from proprioception degrades over time. This interpretation highlights the online nature of position sense and could challenge the arguments for an anatomically based stored model of the body. Further investigations are nevertheless needed before strict conclusions can be drawn regarding the relation between the drift in absolute perceived hand location and the distorted body model. Importantly, the conclusions drawn earlier in this thesis are still valid: other aspects of cognition and perception like visual and conceptual biases might also contribute to the biases measured in localization tasks2. In line with this idea, Gross et al. (See Ph.D. Thesis (1973): Study IV: Effects of Loss of Sight and of Manipulation of Visual Attention on Body Perception) demonstrated that drifts in perceived hand position were partly constituted by information from other modalities (e.g. vision). For instance, the authors showed that the perceived location of finger belonging to congenitally blind subjects was drifted in the opposite direction to the fingers of sighted and noncongenitally blind subjects. Hand representation measured in localization tasks might therefore already be impregnated by vision and not only result from declining proprioceptive information. The relative contribution of perceptual and cognitive factors involved in hand position sense remains largely unexplored. This is topic for further research.

This is especially true for hand width overestimation which was similar across hands and objects. Moreover, no significant differences were found in absolute landmark location error along the x axis of hand and objects. Hence biases in width estimates remain better explained by general mechanisms than body specific properties.

2

29

Fig.7. Item-centered axes (y and x axes) on the hand. Similar item-centered axes were used for objects.

Fig.8. Actual and perceived landmark location for the rake, right hand and rubber hand in upright and rotated orientation (data from chapter 2). The items are displayed along the y and x viewercentered axes. There was a significant drift in absolute location for the right hand in upright and rotated orientations along the item-centered y axis (see this axis on Fig.7). This drift was not significant for the rake and the rubber hand. There were no differences in absolute landmark location across items along the item-centered x axis.

30

Fig.9. Drift of the hand position in time. The graphs represents the mean normalized error in hand landmark position along the y axis in function of time during the first block and second block of trials for the left hand in upright posture (for more details, see the method sections of the localization tasks reported in each chapter). Note subjects saw their hand between block 1 and 2. A normalized time interval of 1 corresponds to the normalized duration of a block.

II.2 A multimodal version of the body model mediating position sense Altogether, the observations from the thesis and previous position sense studies (drift in hand position with time, see above) suggest that the body model underlying position sense might be more closely related to the notion of a dynamic and multimodal representation of the body: the body schema. The difference between the postural schema and the body model is theoretically unclear. The body schema is often defined as a combination of central processes (e.g. motor commands) and peripheral multisensory information (vision, audition, proprioception etc.) which are constantly registered and updated during movement to keep track of the position of our body parts in external space (de Vignemont, 2010; Haggard & Wolpert, 2005; Holmes & Spence, 2004). In that sense, the body schema appears as a highly plastic and continuously changing body representation. In contrast, the body model is, in its original form, described as a more rigid, unimodal metric representation of the body which needs to be combined to the postural schema to locate one’s hand in space: “an innate organization of mental body representations” (Longo et 31

al., 2012) which “appears to retain distortions characteristic of the somatosensory homunculus” and “is distinct from the body schema” (Longo & Haggard, 2010). Despite their apparent differences in properties those two representations might nevertheless be more related than expected and describe different stages of the same body representation. In the context of the localization task, the hand is static and hidden from view. The body schema is therefore not updated by direct dynamic visual and motor information usually available when performing actions (e.g. motor commands and all proprioceptive signals e.g. skin, joints, and muscles receptors involved when moving; Wolpert, Goodbody, & Husain, 1998). This distorted body model could therefore simply correspond to a non-fully processed or non-updated version of the body schema, similar to the distorted perceived limb position observed in the case of patients with phantom limbs or anaesthetized body parts (Gross & Melzack, 1978; Inui, Walsh, Taylor, & Gandevia, 2011; Melzack & Bromage, 1973). While the lack of updated visual and motor signals might lead to a distorted percept of hand location in static conditions, in the case of motion, the combination and integration of different motor and sensory sources of information about bodily parameters might generate an accurate estimate of body part location (Ernst & Bülthoff, 2004; Frith, Blakemore, & Wolpert, 2000; Wolpert et al., 1998). In that sense, body model and body schema might correspond to different stages of the same representation involved in spatial location of bodily information.

II.3 How many body representations do we need? Theoretically, most of the literature suggests the existence of two types of position sense (Proske, 2015; Proske & Gandevia, 2012; Proske et al., 2014; A. J. Tsay et al., 2016; A. Tsay, Savage, Allen, & Proske, 2014). One position sense involved in positioning one limb in relation to another and the second position sense being related to locating the body or body parts in external space. Research indicates that matching behavior between two limbs e.g. aligning the arms with one another, can be explained by a comparison of proprioceptive signals (muscle spindles and cutaneous receptors) between the two body parts, without necessarily involving reference to a central map of the body (Proske, 2015; A. Tsay et al., 2014). In contrast, localizing one’s body part in external space assumes referral to a multisensory body schema (Proske, 2015; Proske & Gandevia, 2012). Surprisingly, this literature never mentions the notion of a stored body model neither the necessity for sub distinctions within the notion of body schema. There is therefore a lack of unifying positive definition for the notion of body model/body schema. This generally points to the broader question of how to dissociate different body representations. Until recently, localization tasks measuring hand position in external space were interpreted as measuring the influence of a central body schema on position sense and not a stored body model (Gross et al., 1974). Hence, the same effect has been interpreted in different ways (body model/postural schema) which highlights the difficulty to ascribe an experimental effect to a specific cause. 32

The body schema is a highly complex body representation, one might therefore be tempted to distinguish or dissociate the various processes involved in the body schema into different types of body representations. But how many and on the base of which criterion can we distinguish them? One representation integrating different stages of spatial information processing used for action performance? Two representations based on functional criteria distinguishing a body model used for perception and a body schema used for action? While drawing distinction between body representations might sometimes be necessary, the taxonomy between the stored body model and the dynamic body schema needs further experimental and theoretical justifications (on a similar topic, see review from de Vignemont, 2010). This is an exciting research endeavor for the future.

III.

Conclusion Overall this thesis explored a whole range of arguments suggesting that hand distortions measured in localization tasks could reflect more general processes than previously expected. First, we found striking similarities between hands and objects measured in localization tasks. Second, we showed that typical characteristics associated with hand distortions in localization tasks could be detected in tasks supposedly void from proprioceptive/somatosensory influences. Finally, we have demonstrated that the magnitude of hand distortions was sensitive to non-somatosensory factors, especially visual e.g. viewer-centered biases and conceptual factors. When merging those factors together, hand distortions appear as non-specific to the body and to localization tasks. It is therefore necessary to interpret the relative distortions measured between estimated landmarks on the hand with caution. One interesting alternative explanation to hand distortions measured in localization tasks is the idea of a proprioceptive drift, which could relate the notion of a stored body model to a more dynamic multisensory schema of the body. We suggest investigating further the exact relation between these two types of body representations, especially how the distorted body model is involved in the integration of action information with the body schema. This is important for any research aiming to relate body size perception to action performance (Gagnon, Geuss, Stefanucci, Baucom, & Creem-Regehr, 2015; Gandevia et al., 2002; Marino, Stucchi, Nava, Haggard, & Maravita, 2010).

33

III.1 Supplementary material Principal Component Analysis on finger length underestimation of the 5 digits measured in the localization task for 66 subjects (see, p.25).

Fig.S1.This scree plot shows the eigenvalue for each finger component.

Finger Index little middle ring Thumb

PC1 0.94 0.82 0.91 0.89 0.76

ss loadings Proportion var cumulative var Proportion Explained Cumulative proportion

PC2 -0.14 0.23 -0.28 -0.27 0.58

PC3 -0.11 0.52 -0.18 0.06 -0.29

PC4 -0.2 -0.08 -0.13 0.35 0.08

communalities 1.2 1.9 1.3 1.5 2.2

PC1 3.75 0.75 0.75

PC2 0.57 0.11 0.86

PC3 0.39 0.08 0.94

PC4 0.19 0.04 0.98

0.76

0.12

0.08

0.04

0.76

0.88

0.96

1

34

IV.

Declaration of the contribution of the candidate This thesis is presented in the form of a collection of manuscripts that are, at the time of the thesis submission, either published or prepared for publication. Details about these manuscripts are set out below together with a description of the contribution of the authors. This work was also presented at international conferences by the candidate.

Chapter 1 Saulton A, Dodds TJ, Bülthoff HH and de la Rosa S (May-2015) Objects exhibit body model like shape distortions. Experimental Brain Research. 233(5) 1471-1479. The ideas presented in chapter 1 were proposed by the candidate. Design, experimental work and analysis of chapter 1 were performed by the candidate. Stephan de la Rosa supervised the work; Trevor Dodds and Heinrich Bülthoff helped revising the manuscript and provided constructive advices during the continuity of the studies.

Chapter 2 Saulton A, Longo MR, Wong HY, Bülthoff HH and de la Rosa S (February-2016). The role of visual similarity and memory in body model distortions. Acta Psychologica 164 103–111. The ideas presented in chapter 2 were proposed by the candidate. Design, experimental work and analysis of chapter 2 were performed by the candidate. Stephan de la Rosa supervised the work; Matthew Longo, Hong Yu Wong and Heinrich Bülthoff helped revising the manuscript and provided constructive advices during the continuity of the studies.

Chapter 3 Saulton A, Bülthoff HH and de la Rosa S (March-2017). Conceptual biases explain distortion differences between hand and object in localization tasks. Journal of Experimental Psychology: Human Perception and Performance. The ideas presented in chapter 3 were proposed by the candidate. Design, experimental work and analysis of chapter 3 were performed by the candidate. Stephan de la Rosa supervised the work; Heinrich Bülthoff helped revising the manuscripts and provided constructive advices during the continuity of the studies.

35

V.

Chapter 1: Objects exhibit body model like shape distortions

This chapter is published in Experimental Brain Research: Saulton A, Dodds TJ, Bülthoff HH and de la Rosa S (May-2015) Objects exhibit body model like shape distortions Experimental Brain Research. 233(5) 1471-1479 Abstract Accurate knowledge about size and shape of the body derived from somatosensation is important to locate one’s own body in space. The internal representation of these body metrics (body model) has been assessed by contrasting the distortions of participants’ body estimates across two types of tasks (localization task vs. template matching task). Here, we examined to which extent this contrast is linked to the human body. We compared participants’ shape estimates of their own hand and non-corporeal objects (rake, post-it pad, CD box) between a localization task and a template matching task. While most items were perceived accurately in the visual template matching task, they appeared to be distorted in the localization task. All items’ distortions were characterized by larger length underestimation compared to width. This pattern of distortion was maintained across orientation for the rake item only, suggesting that the biases measured on the rake were bound to an item-centric reference frame. This was previously assumed to be the case only for the hand. Although similar results can be found between noncorporeal items and the hand, the hand appears significantly more distorted than other items in the localization task. Therefore, we conclude that the magnitude of the distortions measured in the localization task is specific to the hand. Our results are in line with the idea that the localization task for the hand measures contributions of both an implicit body model that is not utilized in landmark localization with objects and other factors that are common to objects and the hand. Keywords: body representation, body schema, position sense, somatosensation

Introduction Stored representations of body size and shape are important components of perception and action (Marino, Stucchi, Nava, Haggard, & Maravita, 2010; van der Hoort, Guterstam, & Ehrsson, 2011). The body model refers to an implicit representation of the body size and shape mediating position sense of the human body (Longo & Haggard, 2010). For instance, information about the metric properties of the body contributes to determining the relative locations of body parts when using proprioception. 36

Recent investigations measured the implicit body model using a localization task in which participants indicate the spatial positions of the felt locations of ten landmarks (finger tips and knuckles) on their occluded hand (Longo & Haggard, 2010; Longo et al., 2012). The results of this localization task showed a highly distorted representation of hand shape consisting of a shortening of the fingers' length and a widening of the hand width. These effects generalized across both hands, and across different hand orientations. The latter finding implies that the observed distortions were not caused by a general foreshortening of the perspective or a motor bias (Longo & Haggard, 2010). The distorted characteristics of hand shape were interpreted as mirroring distortions of somatosensory representations (Longo et al., 2010; Longo & Haggard, 2010, 2011). Specifically, the distortion pattern matches the tactile acuity and geometry of the receptive fields of sensory neurons covering the dorsum of the hand (Brown, Koerber, & Millecchia, 2004; Longo & Haggard, 2011). The implicit body model has been dissociated from another important body representation, namely the conscious body image (Longo & Haggard, 2010, 2012). To assess the body image, participants pick out the image of their own hand among an array of hand images differing in size or shape (template matching task). Participants’ performance in this task is very accurate (Gandevia & Phegan, 1999; Longo & Haggard, 2010). Such an accurate recognition would not be expected if the body image and the body model were sharing the same distorted representation. Hence, larger distortions in the localization task than in the template matching have been interpreted in favor of a dissociation of the implicit body model from the body image (Longo & Haggard, 2010, 2012). Because the larger distortions in the localization task than in the template matching task are a defining property for the body model, it is important to ensure that this taskspecific distortion is specific to the body. To address this point, we compared participants’ performance in a localization and template matching task using noncorporeal objects and the participants' hands. We chose a broad range of objects, from a square CD box (least hand like control), to a rectangular post-it pad sharing a similar aspect ratio to the average hand, to a rake with similar structure to the hand. Here we define localization task specific distortions as the distortions in the localization task that go beyond the distortions observed in the template matching task, i.e. localization task distortions minus template matching task distortions. If localization task specific distortion effects are associated with the body only, they should only be found with the hand but not with non-corporeal items. Specifically, only the hand should present an overestimation of width relative to length in the localization task due to the anisotropies in tactile sensitivity and receptive field geometry on the hand dorsum (Longo & Haggard, 2010, 2011). Moreover, because internal body representations were found to be item centered, rotating items 90º should preserve the pattern of distortions only in the case of the hand (Longo & Haggard, 2010). In contrast, other non-corporeal items should be more sensitive to biases in retina or torso centered 37

coordinates and present a different pattern of distortion when presented in upright versus 90º rotated orientation (Künnapas, 1958).

Methods Participants. 16 right handed individuals (10 males) between 19 and 42 years of age (Mean=28.2) participated in the experiment. Participants gave written informed consent prior to the study. The research was approved by the ethics committee of the University of Tübingen. Stimulus & Apparatus. We used four items as stimuli (Fig. 1): the participant’s left hand, a rake, a rectangular post-it pad, and a square CD-box. In order to compare the stimuli with each other we quantified the item’s shape using its width to length ratio, referred to as its Shape Index (SI=100*width/length), and assumed to reflect the overall aspect ratio of the item (see method, Longo & Haggard, 2012). In Fig. 1A, the width of an item is marked with a yellow line, and the length is marked with a red line. We calculate SI from these item-centric width and length dimensions, so when items are rotated by 90º the resulting SI should remain the same if measured distortions were itemcentric, or change if distortions were viewer-centric. Following Longo and colleagues’ studies, hand length was defined as the distance between the knuckle and the tip of the middle finger, while hand width corresponded to the distance between the knuckles of the little to the index fingers (average hand SI≈64). For the rake, the length was defined as the distance between the bottom and top of the middle branch while the width was the distance between the bottom of the first and fifth branches (SI=40). In the case of the post-it pad (SI=60) and the CD-box (SI=100) we simply referred to the vertical and the horizontal dimensions of the items to define their length and width (Fig. 1A). Ten landmarks were used in the localization task for the hand and the rake: the finger tips and center of the knuckles at the bottom of each finger and the top and bottom of the five branches for the rake (Fig. 2). Four landmarks were used for the CD-box and the post-it: one at each corner. For the template matching task, we used silhouette images of each item. We used an image of a hand that was gender-matched to the participant to generate the hand silhouette. We used custom written scripts with Unity 3.5 (Unity Technologies, San Francisco, USA) for displaying the stimuli (silhouettes and landmark names) and for collecting participants' responses. All stimuli were displayed on a Dell U2412M monitor with a 16:10 widescreen aspect ratio at native resolution (1920x1200 pixels). Our experimental setup differed from the one employed in Longo and colleagues (2010) in the following ways. We used a cursor on a rectangular monitor instead of a long baton (35cm) to point on a square board; the screen was positioned 10 cm higher than the board in Longo et al. (2010) above the table. Indirect pointing via a mouse peripheral was chosen as a preferred method to help alleviate direct motor command influences which might supplement position sense (Fel’dman & Latash, 1982). Previous results have found that hand shape distortions underlying position sense were not viewer 38

centered (Longo & Haggard, 2010); as such we assumed that using a widescreen compared to a square board should not play a major role on the results. The height of the screen was constrained by the hand location just below: in order not to touch the hand, the screen had to be positioned 16 cm above the table top. .

Fig.1. (A) Images of the items used in the experiment. From left to right: hand, rake, post-it, box presented to the participant in upright (top row) and rotated (bottom row) orientation. The yellow and red lines on the items were not present during experimentation and have been drawn to illustrate the itemcentric width and length dimensions used to calculate the Shape Index (SI). The terms 'length' and 'width' always refer to the item-centric length and width, for example in the upright orientation the ‘length’ of the hand, rake, post-it and box is vertical (top row, red line), and in the rotated orientation the ‘length’ is horizontal from the viewer’s perspective (bottom row, red line). The lines on the green background were of a known size and used to calculate the actual hand size from the images. (B) Image of the experimental setup in the localization task, in the condition where the participant estimated the landmarks on the left hand.

Fig.2. Actual and estimated landmarks for the upright hand (left) and rake (right) averaged across 16 participants in the localization task. The filled circles indicate the mean location of actual landmarks (blue) and estimated landmarks (red). The error bars depict the standard deviation (shown for x and y directions separately). For sake of clarity the actual and estimated landmarks are connected with thin lines to highlight the structure of the item. Estimates and actual landmark positions were aligned on the knuckle of the little finger for the hand and on the bottom of the leftmost branch for the rake.

39

Procedure. The order of the localization and template matching tasks was counterbalanced across participants. Localization task. The center of the bottom knuckles and tips of each finger of the participant's left hand were marked with red non-permanent pen. An experimental block measured participants’ ability to localize predefined landmarks on a particular item (hand or object). First, the experimenter explained and familiarized participants with the landmark names and their corresponding locations on an item resting on a flat surface (for about two minutes). Tools have been shown to affect both spatial and bodily representations after performing and observing tool actions (for review, see Maravita & Iriki, 2004). To avoid this confound, none of the objects were touched or manipulated by the participants, nor held in front of participants by the experimenter. Participants sat at a table with their body midline aligned with a mark on the table which indicated the placing position (a cross) for the items. An item was placed centrally with its lower edge at the center of the cross. Participants viewed the item for 15 seconds while the position of the item was recorded using an overhead mounted camera (Canon, EOS 40D; Zoom lens, EF- 28-135mm). Using these images, we derived the exact size of the items (Fig. 1A). Afterwards a computer monitor was slid in parallel to the table top, over the item thereby occluding it (Fig. 1B). Participants were told to use the following strategies. For objects they should imagine the screen to be transparent so that they could 'see' the landmarks below it. For the hand, participants were asked to rely exclusively on the felt location of their finger tips and knuckles without using visual imagery. An experimental trial started by presenting the name of an item's landmark (e.g. tip of middle finger) in white font at the top center of the black computer screen. After a 2 s delay, the mouse cursor was presented at a random y-axis location on the right edge of the screen. Participants indicated as accurately as possible the perceived location of the queried landmark by positioning the mouse cursor over the corresponding position on the computer screen and left-clicking with the mouse. The hand directing the mouse pointer was hidden from view. The answer interval was not time restricted and provided no feedback. Then the next trial started. After testing each landmark in random order five times, the computer monitor was removed for 15 seconds making the item visible to the participant and the item’s location was photographed to ensure that it had not moved. Then each landmark was again tested five times. The ten measures for each landmark constituted one experimental block. Each experimental block probed all landmarks of an item (four items) in one specific orientation (upright or 90° clockwise rotation). There were a total of eight blocks. The testing order of experimental blocks was randomized across participants. At the beginning of the localization task participants received one experimental block as training with a different object (pen). The training data were not included in the analysis.

40

Template matching task. The template matching task is based on the item's SI. Participants estimated the SI using a 1-up 1-down adaptive staircase method (50% threshold; Levitt, 1971). The actual item was first presented on a table to the participant in an upright position and participants had one minute to remember the item’s SI. The item was then hidden from view. Afterwards, participants indicated whether an item’s silhouette displayed on a computer monitor was wider (right arrow key) or narrower (left arrow key) than the actual item. The area of the silhouette matched that of the item and only the SI was modified. The SI of the displayed silhouette was altered using two randomly interleaved staircase procedures. The first staircase had a start value of 125% of the actual SI; the second started at 75%. The procedure ended after each staircase had a total of thirteen reversals, with an upper limit of 80 trials. Step sizes were reduced after each reversal and the step sizes were 16 (initial step size), 8, 4, and 3 SI. The threshold (SI at convergence) was calculated from the mean of last 5 reversals (across the two procedures). The testing order of items was randomized across participants. Statistics. We used Mauchly’s sphericity test to validate the analyses of variance used on our data. When violations of sphericity were observed, we reported the results with Greenhouse-Geisser sphericity corrections.

Results Dissociation between localization and template matching tasks

Assessing distortions of each item separately. Previous studies have shown that participants have an accurate representation of hand shape in the visual template matching task and a distorted representation of hand shape in the localization task characterized by a value that is largely superior to the actual shape index of the hand (Longo & Haggard, 2010, 2012). If the localization task distortions are mainly associated with bodily items then it should be primarily present in the hand and not in non-corporeal objects. We normalized each item’s shape index (SI) by dividing the estimated SI by the actual item’s SI, to create a baseline of 1 and allow between item comparisons. We compared the normalized SI of all items in upright posture to the baseline (=item’s actual shape index) in both tasks. In the localization task, baseline comparisons showed that all normalized SIs were significantly larger than 1 (all p≤.013, all effect sizes r≥.60; p values were Holm-corrected, see Table 1) suggesting that all items showed larger estimations of width relative to length (mean width estimate > mean length; see supplementary material, S1). In contrast, responses on the visual template matching tasks were close to accurate for the majority of items (all p>.05 except for the box t(15) = 5.22, p < .001, r = .80 , p values were Holm-corrected ; see Table 1). The shape estimations are depicted in Fig. 3.

41

Item

Hand

Rake

Post-it

Box

Localization vs. baseline

t(15) = 4.80 p=.0009 r=.77

t(15) = 3.14 p=.013 r=.63

t(15) = 3.35 p=.013 r=.65

t(15) =2.88 p=.013 r=.60

Template matching vs. baseline

t(15) = 4.80 p=.93 r=.19

t(15) = 3.14 p=.93 r=.052

t(15) = 3.35 p=.09 r=.52

t(15) =5.21 p=.0004 r=.80

Localization vs. Template matching

t(15) = 4.58 p=.0014 r=.76

t(15) = 3.32 p=.014 r=.65

t(15) = 2.16 p=.094 r=.49

t(15) = 1.39 p=.19 r=.34

Task

Table 1. Holm corrected t-tests comparing the SI to the veridical performance of 1 for each upright presented item in the localization and template matching tasks (first and second row). Holm corrected comparisons of SI between localization and template matching tasks (third row). Each cell provides the t statistic, p value, and the effect size measure r. Significant effects are highlighted with a grey cell background.

Assessing distortions between items. To assess differences in distortions between items we conducted a within-subjects analysis of variance (ANOVA) on the normalized shape index with items (hand, rake, post-it, and box) and task (localization vs. template matching task) as within-subject factors. There was a significant effect of item [F(2.11, 31.58)=8.60, p