Race and Intelligence bn999920q

Citation preview

ANTHROPOLOGICAL STUDIES

Number 8 a |

Kace and Intelligence

edited by C. LORING BRACE, GEORGE R. GAMBLE, and JAMES T. BOND

Published by the AMERICAN ANTHROPOLOGICAL ASSOCIATION 1703 New Hampshire Avenue, N.W. Washington, D.C. 20009

Library of Congress Catalog Card Number: 75-181732 Copyright © 1971 by the American Anthropological Association. All rights reserved. Contents may not be reproduced in any form or by any means without prior written permission from the publisher. Printed in USA.

CONTENTS

Preface 2... ee ne ee eee eee ee eee tence eee ee eeececeee 1 Introduction to Jensenism .......... 0.0.0. eee ee ee eee ee ee eee es GC, Loring Brace 4 Can We and Should We Study Race Differences? ................. Arthur R. Jensen 10

Intelligence in Black and White ........................... Alexander Alland, Jr. 32

Whose Is the Failure? 2... ee ee ee eee eee cee eeees Vera dohn 37 The Influence of Conceptual Rule-Sets on Measures of Learning Ability .............

See ee ee ee ee ee eee ene eee eee e teen eee eceeeeee HOSAlie Cohen 41

Genetic and Environmental Components of Differential Intelligence ................ See ee ee eee eee ence eee eeeeseececeeee Lhomas G. Gregg and Peggy R. Sanday 58

On Creeping Jensenism ..................€. Loring Brace and Frank B, Livingstone 64

Preface IN THE WINTER OF 1969, the Harvard Educational Review published an article of monographic proportions by the California (Berkeley) educational psychologist Arthur R. Jensen.

The article, the longest ever published by the HER, expanded upon a theme which Jensen had ~ presented in a “most unfortunate speech” (Hirsch 1967:434) before the American Educational Research Association meetings in New York. His conclusion, which, as was pointed out, did not follow from the data presented (Kagan 1969, Deutsch 1969, Bodmer and Cavalli-Sforza 1970), was that the lower average IQ test score achieved by Negro Americans when compared with that achieved by white Americans is due to innate intellectual inferiority. With all the statistics and pro forma academic phraseology stripped away, he believes that blacks are stupider than whites and

that the difference is genetically determined. |

Coming as it does when the entrenched racism in American society is being widely and seriously

questioned for the first time (Silberman 1964, Rose 1968, Schwartz and Disch 1970), such a conclusion offered in the name of “‘science’”’ could not fail to elicit a vigorous reaction to say the very least. It was regarded as newsworthy by the US News and World Report, Time, Life, and Newsweek. It was given a version of “‘balanced” treatment by the New York Times Magazine, - guarded suspicion by Psychology Today, and warm approval by White Power, the newspaper of the National Socialist White Peoples Party. It was immediately put to use by southern lawyers and politicians still seeking to oppose the even-handed administration of the law to all citizens, and it was the subject of varied and thunderous rebuttal by professional scholars in the spring and summer issues (1969) of the Harvard Educational Review. Most of the initial scholarly reaction was from other educational psychologists since they had

watched the development of Jensen’s views at their professional meetings and in their own specialized publications. This in effect had given them something of a head start. If scientists in

other related areas were slower to react, they were in fact no less interested. Sociologists, psychologists, geneticists, and anthropologists have all joined the parade, and, to their credit, they have been unanimous in recognizing that Jensen’s conclusions simply do not follow from the available evidence. On the other side, one distinguished scientist has risen in support of Jensen’s views. This is William Shockley, co-winner of the Nobel Prize for his role in the invention of the transistor. While he is not fully familiar with the nature and reliability of what constitute data in the social sciences, he is convinced that “‘there is a difference in the wiring patterns” of white and

black minds (Neary 1970:581), and he has repeatedly urged that the National Academy of

Sciences sponsor an investigation of this assumed difference. : |

Any such investigation, of course, is laden with social and political dynamite, and emotions

have been running strong. To approve of such an approach would represent a concession to the . racists who back the assumptions of the Shockley-Jensen position. Refusal on the other hand lays

the Academy open to the charge of unscientific behavior; that is, of accepting the contrary viewpoint without having undertaken formal study. Adding to the anguish of the Academy’s dilemma is the hostility being directed at science from both the political left and the political right, and the reduction of money available for its pursuit. So far they have managed to keep a reasonable semblance of balance (Walsh 1971). Somewhat ironically, the branch of science which has been the slowest to produce a coherent

response to the Jensen challenge has been anthropology. Some of this may be the result of a - certain amount of vacillation on the part of those who have their hands on the levers of power in

the profession, and some may come from the preoccupation with the dispute over ethics,

2 RACE AND INTELLIGENCE | [AS8 government support, and research in southeast Asia. But one also gets the feeling that anthropology has been caught napping to a certain extent because it may have felt that the basic _ issues were resolved a generation and more ago. The insights of Tylor and Boas from before the turn of the century, and their applications to combat the racism of the Immigration Restriction League, the eugenics movement, and the grisly rise of Hitler’s “‘Aryan” theories have now retreated so far into the past that they have become practically a dead issue. It is not that anthropologists do not believe in them any more. In fact, they are enshrined as fundamentals at the beginning of every textbook in elementary general anthropology, but they are

so very basic that they tend to be taken somewhat for granted. To reapply the old cliche, anthropologists find themselves in a position analogous to that of the discussant of a problem in differential calculus who discovers that his opponent has not yet learned to multiply. The difference is that the basic anthropological discoveries concerning the nature of culture are very nearly impossible to quantify. How many “‘facts’’ must an aborigine command in order to survive in the Australian outback as compared to those necessary in suburbia USA? Are the perceived minimal bits of information in a literate culture comparable to those of a non-literate culture? What kind of measurement can demonstrate whether the basics of one culture are easier

or more difficult to learn than those of another? Despite the difficulties in quantifying the investigation of such questions, the unanimous conclusion of all those who have done field work over the last hundred years is that there are no differences in basic human learning potential when any group of people is compared with any other group. If anything, the enormous respect for human capacity which is so often derived from an anthropologist’s field experience might lend fuel to the suspicion that the average level of human competence is greater in less complex societies where the penalty for incompetence can be more readily equated with failure to survive. This,

however, is sufficiently unprovable so that it is scarcely more warranted than the common ethnocentric assumption that technological simplicity proves mental inferiority. | This has become so patently self-evident to anthropologists that one feels a little foolish at - being put in the position of having to repeat it before an intelligent audience. However, since it has been conspicuously absent from the claims and counterclaims that Jensen and his critics have produced, it would appear that the time has come to assert the truth of what should be an obvious

truism. It was this feeling that lay behind the organization of the symposium on “Differential Intelligence in Populations?” at the 68th Annual Meeting of the American Anthropological Association in New Orleans in November of 1969, although if it had not been for the organizing efforts of my two young co-editors, it appears that American anthropology would have registered

no reaction at all. To be sure, there has been some sentiment to the effect that any such anthropological attention would constitute undeserved publicity for the Jensen viewpoint. On the other hand, however, it was evidently not going to disappear just because the anthropologists refused to recognize it. Ample publicity from other sources has kept it very much alive and we feel that it would be a serious lapse in responsibility if the anthropological viewpoint continued to be represented by silence.

The session was chaired by Frederick Gearing and featured presentations by myself (co-authored by Frank B. Livingstone), Alexander Alland, Jr., Vera John, Rosalie Cohen, and Sherwood L. Washburn. At the conclusion, there was some agreement that the papers shouid be published after suitable rewriting and editing, although it has taken well over a year to agree just who would publish the papers, which authors would in fact agree to have their papers published, and who should serve as the editor. At one point, those willing to contribute had accepted the invitation of Sol Tax to have their papers published in Current Anthropology along with a paper that Jensen had prepared summarizing his viewpoint and a volunteered contribution by Gregg and Sanday, covering an important point left out by Livingstone and myself—the whole to have been submitted to critique with replies to the critiques. This alas proved too time-consuming and unmanageable even for the formidable talents of Sol Tax, so the copy-edited manuscripts were returned. The Executive Board of the AAA, in response to a request by George R. Gamble and | James T. Bond, who never let up in their efforts, agreed to consider publishing the collection asa AAA publication provided that they could get a proper figurehead to serve as editor. I was considered suitably ‘‘proper’’ and, upon being approached, agreed to the task.

, PREFACE 3 After all the wrangling and the delays in the publication of work that is now two years old, we hope that some good shall come of our efforts. While we had conceived of this collection as a specifically anthropological response to the issues raised by Jensen, one of our contributors is a sociologist, one is a psychologist, and one a geneticist. We justify this as a demonstration of the interdisciplinary character of the scope of anthropology and hope that the contributions of our

various colleagues do not contradict what we feel is the anthropological flavor of the whole. Finally, we should stress that this should in no way be regarded as an official reaction by the AAA to the Jensen controversy. The papers included herein represent the individual approaches of | the authors. Nor is the choice of authors as much the result of a carefully calculated sampling of

the areas they represent as it is the operation of chance in the shadowy area of academic communication networks and availability. This is not to say by any means that the choice was purely random, but much depended on who could and would be on the spot at a given time, and,

finally, who could and would prepare a manuscript for this collection. Despite the very “‘iffy”

, : C. Loring Brace

nature of this process, I think we have more than a little reason to be pleased with the results.

| REFERENCES CITED : Bodmer, W. F., and L. L. Cavalli-Sforza |

: 1970 Intelligence and Race. Scientific American 223(4):19-29. | Deutsch, M. .

1969 Happenings on the Way Back to the Forum: Social Science, IQ, and Race Differences Revisited. Harvard Educational Review 39(3):523-557. Hirsch, J.

1967 Behavior-Genetic Analysis. In Behavior-Genetic Analysis, pp. 416-435. J. Hirsch, ed. New York: McGraw-Hill.

Kagan, J. S. , Neary, J. .

Jensen, A.R.

1969 How Much Can We Boost IQ and Scholarly Achievement? Harvard Educational Review

39(1):1-123.

1969 Inadequate Evidence and _ TIllogical Conclusions. Harvard Educational Review

Rose, P. I. ; , | 39(2):274-277. -

1970 A Scientist’s Variations on a Disturbing Racial Theme. Life 68(2):58B-65.

1968 The Subject is Race. New York: Oxford University Press.

Silberman, C. | Schwartz, B. N., and R. Disch a

| 1970 White Racism. New York: Dell.

1964 Crisis in Black and White. New York: Vintage Books.

Walsh, J.

1971 National Academy of Sciences: Awkward Moments at the Meeting. Science 172:539-542.

Introduction to Jensenism C. LORING BRACE DEPARTMENT OF ANTHROPOLOGY UNIVERSITY OF MICHIGAN

THE URGE to capture the essence of a people—whether a geographical enclave, a nation, or just another town—in a single pat stereotype is as old as the perception that people differ from one area to another in their appearance and their behavior. This human tendency to categorize appears

in the earliest transcriptions of remembered history and tradition, as for instance in the unflattering characterization of the Biblical Philistines, or the Pharisees of the New Testament. Systematic stereotyping was canonized in the philosophy of Plato as the basis for a general approach to dealing with the apparent graded variety of the world, and this in turn was deliberately applied by the naturalists of the post-Renaissance world in their efforts to classify all the phenomena of “‘creation.”’ To this tradition we owe the enshrinement of “type specimens” in museum collections and the never-ending wrangle over the specific identity of each new fossil as it is discovered.

Even in earliest times, the fact that typological thinking could not encompass the full nature of reality was occasionally recognized, as for instance in the tale of the atypical—i.e., good—Samaritan. This, however, did little to shake the assuined reality of the stereotype. Quite the reverse. In good

Platonic fashion, the fact that individuals so often failed to embody the stereotype which should presumably encompass them was taken as an indication of the imperfection of this the secular world. By the eighteenth century, at least, it became practically a sacred duty to discover the ideal

essence in even the most sinfully deviant members of an assumed category. Recognition of variation while maintaining unshaken faith in the reality of the stereotype is classically expressed in

the phrase, “Some of my best friends are ”” (fill in with whatever categorical label you choose).

One can suggest that this tendency to reduce the graded variation of the world to finite symbolic form is as close to being instinctive as anything in the human behavioral repertoire. Language by its very definition is the symbolic representation of the more complex real world. Linguistic acquisition in the maturing human infant marks the sharpest distinction in the course of the behavioral development of man as compared with that of his closest simian relatives. Few would seriously question the claim that the exponential growth in ecological dominance by Homo

sapiens over the last 100,000 years or more is due at bottom to his ability to reduce the complexity of nature to a verbal analogue for ease of manipulation. The secret of human power, in a word, is language. This, too, is an ancient insight. As the Gospel put it, “In the beginning was the Word ... and the Word was God”’ (John 1:1). Not only does this illustrate the long-time realization that language is the key to human power, but it projects the importance of linguistic categorization to a superhuman level. Whether it is the “open sesame’”’ of ancient mythology or the ability to summon the Devil by calling his name in Medieval Christianity, the elevation of ‘‘the Word’’ to the level of sympathetic magic illustrates an

aspect of human behavior that has enjoyed favor for a span of time that doubtless equals the antiquity of language itself. As was recognized by the late V. Gordon Childe, “It is an accepted principle of magic among modern barbarians as among the literate peoples of antiquity that the name of a thing is mystically equivalent to the thing itself, in Sumerian mythology the gods 4

Brace | INTRODUCTION TO JENSENISM 5 ‘create’ a thing when they pronounce its name. Hence to the magician to know a thing’s name is to 7 have power over it, is—in other words—‘to know its nature’ ”’ (Childe 1946:128).

At the effective level, this continues with undiminished application at present and for the forseeable future. For instance, “‘bourgeois”’ is the personification of evil to both the Old and the New Left; ‘“ccommunist” is the personification of evil to the Middle American. And ‘“‘democratic’”’

is the epitome of the good to all three groups, although the details of what would actually be regarded as “democratic” would differ to an enormous extent. The discussion of the nature of | _ human biological variation is clouded by the same tendency to depict the world with a series of labels which give the user a sense of control over the nature of what is labelled. The very use of the

term race creates a reality which retains a mystical hold on the mind of the user. Even so sophisticated a biologist as Ernst Mayr, who has long warned about the Platonic background to continuing typological thinking, tumbles headlong into the trap against which he cautions with his extraordinary declaration of faith: ‘Races there are; how to delimit them, how to draw the line between them is not only difficult, it is impossible’ (Mayr 1968:103). The reality of races as biological entities, then, is to be found in the human conviction that they exist. Analogous to the Platonic Universe in which the dimensions of existence are determined by the ideas in the mind of God, the dimensions of the. human world are determined by the ideas in the minds of men. Even those of us who deny the validity of the race concept on biological grounds and who maintain that a genuine understanding of the nature of human variation can only be achieved after it is abandoned (Livingstone 1962, 1964; Brace 1964)—even we are forced to , admit that races are very real. They are real because people believe they are, and social reality—the human world—is determined by human belief.

Normally an individual acquires the dimensions of his belief by the process of enculturation, and only rarely does he seriously question the bases for their establishment. As a result, the dimensions of social reality accepted by most people are derived from the verbalization of experiences which first occurred some time in the past. Where racial categories are concerned, despite the vast amounts of information now available, those groups which are accepted as being

valid races still reflect the stereotyped impressions of seventeenth and eighteenth century European merchants. Both to the scientist and to the average man, what are considered the - significant dimensions of human biological variation owe more to the location of the ports of call of the old European mercantile system than to the average biological differences between major seographic and population centers.

Along with the labels of a long-since altered world, we have accepted some aspects of the meaning attributed to human differences by the thinkers of that age. Organic reality was arranged in a scale—the Great Chain of Being—ranging from invertebrates up to man, “a little lower than the

angels” in the words of the Psalm, who themselves ranked just below God (Lovejoy 1936). Relative position on the scale was equated with relative worth in the eyes of God. Explicit, too, was the idea that everything lower on the scale was created for the specific purpose of exploitation

by man at the top, and we are only now beginning to wake up to the ecologically ruinous consequences. Naturally the perceived varieties or races of men were installed on the chain also. Since Europeans did the installing, they tended to rank the other races below them, the distance below being determined by the extent to which they differed physically and culturally from the accepted European ideal. From the end of the eighteenth century, up to the union of evolutionary theory and genetics in the 1940s, the study of human physical variation has been a history of

almost unrelieved efforts to measure and quantify the extent of the inferiority of non-Europeans. , The racism which pervaded the general concern dating from the eighteenth century for the significance of human variation has been well documented (Harris 1968). If the ethos of this first century of anthropological effort “‘was brought to an end, insofar as it has been, by the discovery | of culture” (Bohannan 1971:2), there is still the possibility that the “insofar as it has been” did not extend beyond the bounds of cultural anthropology (in the American sense) until very much more recently (Barzun 1965). “Anthropology” in Europe has continued to mean what is called

“physical anthropology” (in the narrow sense) in America, that is, a primary concern for the _ tangible evidence for human variation, past and present, generally to the exclusion of any interest

in the cultural dimension. One could make a good case for the claim that the late survival of

6 , RACE AND INTELLIGENCE [AS8 evident racism in physical anthropology, even in England and America, is one of the consequences

of this unfortunate professional compartmentalization. :

If racism was terminated in general anthropology by ‘“‘the discovery of culture,”’ perhaps one would suspect that its survival elsewhere is in direct proportion to ignorance concerning the nature — and dynamics of the cultural realm. It was the feeling that a reduction in this ignorance would be

the best means of combatting racism and ethnocentrism which led many of us to push for the expansion of the field of anthropology after World War II. From a small field with the image of dilettantism and a concern for the esoteric, it has grown to occupy a respected niche in the general curriculum of college-level education. It would be too much to claim that this alone is responsible for the increasingly enlightened treatment of matters dealing with race expressed by the news

media as well as the elected representatives of the public at large, but at the very least

anthropology can be credited with having made major contributions to this trend. | Despite the gratifying nature of the change, it has still been agonizingly slow in the eyes of those engaged in promoting it as well as from the viewpoint of those suffering from the survival of the old views where individual worth was judged by group stereotype and groups themselves were

: assigned different ranks in an assumed hierarchy. Part of the reason for this leisurely pace can be

attributed to the fact that many of the practitioners of the very social sciences that have spearheaded the change had not themselves fully discovered the nature and power of the cultural realm. If some parts of anthropology—notably the biological—were slow in coming to this realization, it is hardly surprising that the nature of culture has been of less than central concern to the other social sciences. This has been particularly true in psychology and especially that part of psychology that has attempted to discover the portion of human behavior solely under genetic

control. )

Binet and Simon developed their intelligence tests during the first decade of the twentieth century just after the rediscovery of Gregor Mendel, and many felt that their application would determine the amount of human intellectual difference under the control of heredity. At the time, they warned that the results of their tests were useful only when the children tested came from similar environments—a warning that has been often repeated and still more often ignored. Even with the criterion of environmental similarity satisfied, the question of what could be learned from intelligence tests continued to be bothersome and elicited the comment early in the 1920s that _ ‘“‘measurable intelligence is simply what the tests of intelligence test,” even then considered “‘an old saw” (Boring 1961:210). Now more than a generation later, the issues are still being disputed. In spite of the application _ of multivariate statistics, matrix algebra, and a bewildering array of other computer-assisted analytical techniques, questions concerning the utility of intelligence tests and what it is they measure are still unresolved. Just recently the distinguished molecular biologist Erwin Chargaff came to the conclusion that “‘the all too frequent performance of intelligence tests is more likely to make the testers more stupid than the tested more intelligent” (1971:638). Yet the pre-evolutionary eighteenth century view continues on in the unshaken conviction that individuals can be understood in terms of a stereotyped image of the group of which they are considered a part, and that the groups in turn can be arranged in a hierarchical and linear array

| with Europeans at the top. The prejudice is neatly expressed by the long-time chairman of the Department of Psychology at Columbia University and author of a widely used textbook in statistics, Henry E. Garrett, ““No matter how low (in a socioeconomic sense) an American white

may be, his ancestors built the civilizations of Europe; and no matter how high (again in a _ socioeconomic sense) a Negro may be, his ancestors were (and his kinsmen still are) savages in an African jungle” (1962:984). This represents no change from the eugenics viewpoint of the 1920s that ‘‘the negro lacks in his germ plasm excellence of some qualities which the white race possesses and which are essential for success’’ in western civilization (Popenoe and Johnson 1926:285). Nor is there any basic change from the opinion of the racist opponent of Darwin and Huxley, James

Hunt. In reply to Huxley’s Man’s Place in Nature (1863), Hunt, as President of the Anthropological Society of London, read a paper entitled ““On the Negro’s Place in Nature” in

which he concluded, among other things, that “the Negro is inferior, intellectually, to the

Brace | RACE AND INTELLIGENCE 7

Negro” (1864:xvi). ,

European,” and that “European civilization is not suited to the requirements and character of the

One can understand how such an attitude of complacent and parochial ethnocentrism could exist among the educated in 1863. Evolution was still very new. Genetics had yet to be expounded and there was no faint suspicion of the complexity of environmental inputs shaping genetically

controlled substrates which act during the growth process to produce the adult phenotype of all complex structures. Freud was still a child, and Tylor had not yet published his definition of _ culture: “that complex whole which includes knowledge, belief, art, morals, law, custom, and any other capabilities and habits acquired by man as a member of society” (1871 [1958]:1).

But more than a century has passed since that time. We at least know something of the complexities involved in translating genotype to phenotype. And we know, too, that the process of enculturation, and the sequence of individual experiences that occurs while this takes place, plays such an overwhelming role in shaping human. behavior that it has been a major challenge to psychologists to discover whether any aspects of human behavior are inherited. If it has been

difficult to find the very existence of such dimensions, it has proven even more difficult to compare the extent of their development from one individual to another, and next to impossible when the comparison pits one population against another. Thus when one reads a statement claiming that “‘the Stanford-Binet IQ test measures the ability

to adapt to Western civilization” (Jensen 1969:14) it gives one the feeling that the author’s education in the social and biological sciences—to say nothing of history—must have been woefully inadequate. The author, however, Arthur R. Jensen, is a professor of educational psychology at

one of the most prestigious universities in the United States, the University of California at

then, can hardly be to blame. 7 .

Berkeley, and holds a PhD from another prestigious school, Columbia. Inadequacy of background,

Others have been equally puzzled by the consistent slant of Jensen’s views, given the apparent lack of a racist bias in his background. To be sure, Jensen’s approach bears an uncomfortable similarity to that of the racist Henry E. Garrett quoted above, and the latter retired as chairman of the Psychology Department at Columbia University the very year that Jensen earned his PhD, but Jensen’s degree was granted by Columbia Teachers College, an administratively and academically

separate unit. 7 ]

) Whatever the specific influences in shaping his career, the fact remains that ‘“‘the conclusions he

draws are ... unwarranted by the existing data, and reflect a constant bias towards a racist hypothesis” (Deutsch 1969:525). The study of intelligence in relation to race is bound to generate a highly emotional response. Even if clear-cut scientific conclusions were possible, and they are not, the presentation of a report in ‘‘sloppy and premature style,” to use the words of the Nobel Laureate Joshus Lederberg (Neary 1970:64), would certainly justify the claim that Jensen was guilty of gross insensitivity if not of the more pointed terms that have been used to describe him. The root of the issue, I suspect, is that Jensen is caught in the same Platonic trap that has plagued the study of man ever since the eighteenth century. Although he is aware of the hazards of approaching race as a “‘Platonic essence,’ he nonetheless dismisses the alternative as “utter nonsense” without the benefit of analysis, citations, or even cursory discussion. At the same time he accepts a definition of race—“breeding groups which differ in the frequencies of one or more - genes’’—-which is so vague that it provides no basis for differentiating between a clan and a village, a tribe and a whole continent, or even between two nuclear families. Adding the further comment that “major races’’ are characterized by a “relatively high degree of inbreeding’’ provides no help .

either. —

The problem is the same one which faced Ernst Mayr. Jensen, like Mayr, accepts the reality of races although he can neither define them nor delimit them. In his zeal for performing invidious comparisons, however, he has taken a further step, one which Barzun warned about when he said,

“For all purposes, the chief value of race-worship is that it stimulates group conceit after paralyzing the critical faculties” (1965:183). Finally, we come to the question of the responsibility of the pursuit of scientific endeavor. One . can applaud Jensen’s feeling that problems should not be avoided just because they are complex or difficult, but this still does not justify tackling, to the exclusion of others, a particular one just

8 INTRODUCTION TO JENSENISM [AS8 because, like the proverbial mountain, “it is there.” To ignore this is to elevate to near certainty the probability that the research practiced will be one-sided or “‘underdimensioned,” and to give | further fuel to those who have sought to portray science as irresponsible or even evil (Blackburn 1971). If, as Jensen claims, the contribution of heredity to intelligence is an “important question,”

all facets of the problem should be studied with equal thoroughness. One cannot avoid the problem of the legacy of slavery just because it is difficult to infer things accurately from historical accounts. Nor can one avoid the criticism which the council of the Society for the Psychological Study of Social Issues directed toward Jensen (namely, that he failed to set social conditions as equal before pronouncing on the contribution of heredity), by complaining that ‘‘no operationally testable meaning is given to ‘equal’ social conditions.’’ Again, he would appear to have avoided an absolutely essential part of the research on his “important question”’ just because it is complex and

difficult. |

While it may come as a surpise to Jensen, a great many scientists do take the SPSSI statement 7

seriously. It is in fact the basis for the reason why so many regard Jensen’s conclusions as “‘sloppy and premature,’ and many are quite convinced that the kind of research pursued by Jensen, or at least the drawing of conclusions from it, should indeed be postponed for several generations or perhaps indefinitely. It has already been carried on for more than two generations and nothing yet has come of it. Meanwhile, enormous social changes have occurred which have been accompanied

by equally enormous changes in performance on intelligence tests by whole categories of the American public. These performance changes could hardly be genetic. More than anything else, they point to the absolute necessity of controlling for social conditions before pronouncing upon the role of the hereditary components in shaping the behavioral distinctions between populations of human beings.

In conclusion, one could say that if in fact Jensen were really interested in an unbiased testing of the heritable component of intellectual differences between human groups, he should have been devoting his efforts to setting up a scientifically acceptable test situation. The very first step would involve engaging in an attempt to produce an operational definition of equal social conditions and the systematic effort to see that these be extended to all those whom he might wish to test. Better yet, to everyone. Then, and only then, could the question of inherited differences in ability be posed. In fact, whether or not the question is indeed “important”? could only be decided under - . such circumstances. Viewed from a humanitarian perspective, it would substantially improve the

lot of mankind if the energy currently being devoted to the dubious demonstration of innate human unworth were rechannelled to the task of removing the non-innate but very real social inequities that cripple the lives of the very people for whom Jensen professes such concern.

| ' REFERENCES CITED Barzun, J. 1965 Race: A Study in Superstition. Revised. New York: Harper and Row. Blackburn, T. R. 1971 Sensuous-Intellectual Complementarity in Science. Science 172:1003-1007.

Bohannan, P. _ 1971 Ethics. Newsletter of the American Anthropological Association 12(2):2.

Brace, C. L. ; Boring, E. G.

Chargaff, E. 1961 Intelligence as the Tests Test It. In Individual Differences, pp. 210-214. J. J. Jenkins and D. G. Patterson, eds. New York: Appleton-Century-Crofts.

1964 A Non-Racial Approach Toward the Understanding of Human Diversity. In The Concept of Race, pp. 103-152. M. F. A. Montagu, ed. New York: Free Press.

1971 Preface to a Grammar of Biology. Science 172:637-642.

Childe, V. G.

1946 What Happened in History. New York: Penguin Books. Deutsch, M.

1969 Happenings on the Way Back to the Forum: Social Science, IQ, and Race Differences | Revisited. Harvard Educational Review 39(3):523-557.

Brace | INTRODUCTION TO JENSENISM 9 Garrett, H. E. 1962 Racial Differences and Witch Hunting. Science 135:982-984. Harris, M.

1968 The Rise of Anthropological Theory: A History of Theories of Culture. New York: Crowell.

Hunt, J. 1864 On the Negro’s Place in Nature. Anthropological Review 2:80-98. Jensen, A. R. 1969 Edited Reply to a Letter by D. N. Robinson. New York Times Magazine, Sept. 21, p. 14. Livingstone, F. B. 1962 On the Non-Existence of Human Races. Current Anthropology 3:279-281.

1964 On the Nonexistence of Human Races. In The Concept of Race, pp. 46-60. M. F. A.

Montagu, ed. New York: Free Press. Lovejoy, A. O.

1936 The Great Chain of Being: A Study of the History of an Jdea. Cambridge: Harvard

University Press. Mayr, E.

1968 Discussion of Part Two: Biological Aspects of Race. In Science and the Concept of Race, pp. 103-105. M. Mead, T. Dobzhansky, E. Tobach, and R. Light, eds. New York: Columbia University Press. Neary, J. 1970 A Scientist’s Variations on a Disturbing Racial Theme. Life 68(22):58B-65. Popenoe, P., and R. Johnson 1926 Applies Eugenics. New York: Macmillan. Society for the Psychological Study of Social Issues 1969 Council Statement on Race and Intelligence. Journal of Social Issues 25:1-3. Tylor, E. B. 1871 (reprinted in 1958) The Origins of Culture. New York: Harper.

Can We and Should We Study Race Differences? ARTHUR R. JENSEN INSTITUTE OF HUMAN LEARNING UNIVERSITY OF CALIFORNIA, BERKELEY

All Rights Reserved

Copyright © 1970 by Arthur R. Jensen

MOST PERSONS experience some difficulty in discussing the topic of race differences in intelligence—a difficulty over and above that which is ordinarily inherent in the scientific study of

any complex phenomenon. There is an understandable reluctance to come to grips with the problem, to come to grips with it, that is to say, in the same straightforward way that we would try to approach the investigation of any other problems in the behavioral sciences. This reluctance is manifested in a variety of “symptoms” found in most writings and discussions of the psychology

of race differences, particularly differences in mental ability: a tendency to remain on the remotest fringes of the subject, to sidestep central questions; to blur the issues and tolerate a degree of vagueness in definitions, concepts, and inferences that would be unseemly in any other realm of scientific discourse; to express an unwarranted degree of skepticism about reasonably well established quantitative methods and measurements; to deny or belittle already generally accepted facts—accepted, that is, when brought to bear on inferences outside the realm of race differences;

to demand practically impossible criteria of certainty before even seriously proposing or investigating genetic hypotheses, as contrasted with extremely uncritical attitudes toward purely environmental hypotheses, a failure to distinguish clearly between scientifically answerable aspects of the question and the moral, political, and social policy issues; a tendency to beat dead horses and to set up straw men on what is represented as the genetic side of the argument; appeals to the

notion that the topic is either really too unimportant to be worthy of scientific curiosity or too complex, or too difficult, or even forever impossible for any kind of research to be feasible, or that answers to key questions are fundamentally “unknowable” in any scientifically acceptable sense; and, finally, the complete denial of intelligence and race as realities, or as quantifiable attributes, or as variables capable of being related to one another, thereby dismissing the subject altogether. These tendencies will be increasingly overcome the more widely and openly the subject is discussed among scientists and scholars. As some of the taboos against the public discussion of the topic fall away, the issues will become clarified on a rational basis, we will come to know better just what we do and do not yet know about the subject, and we will be in a better position to deal with it objectively and constructively. IS INTELLIGENCE AN ATTRIBUTE?

Intelligence is an attribute of persons. Probably for as long as man has been on earth it has been a common observation that persons differ in brightness, in speed of learning, in ability to solve problems, and so on. Parents, teachers, and employers are able roughly to rank children and adults in terms of a subjective impression of brightness or capability, and there is a fairly high agreement among different observers in the rank order they assign in the same groups of children. It is helpful to think of the subjective perception of intelligence as analogous to the subjective perception of temperatures, which is also an attribute. Before the invention of the thermometer, temperature 10

Jensen | CAN WE AND SHOULD WE STUDY RACE DIFFERENCES? 11 was a matter of subjective judgment. The invention of the thermometer made it possible to objectify the attribute of temperature, to quantify it, and to measure it with a high degree of reliability. With some important qualifications, the situation is similar in the case of intelligence tests. The most essential difference is that intelligence, unlike temperature, is multidimensional rather than unidimensional. That is to say, there are different varieties of intelligence, so that persons do not maintain the same rank order of ability in every situation or test that we may

regard as indicative of intelligence. It so happens that from among the total spectrum of human behaviors that can be regarded as indicative of some kind of “‘mental ability’ in the broadest sense, we have focused on one part of this spectrum in our psychological concept of intelligence. We have emphasized the abilities characterized as conceptual learning, abstract or symbolic reasoning, and abstract or verbal problem solving. These abilities were most emphasized in the composition of intelligence tests because these were the abilities most relevant to the traditional school curriculum and the first practical intelligence tests were devised to predict scholastic performance. When tests were devised to predict occupational performance, they naturally had a good deal in common with the tests devised for scholastic prediction, since the educational system is intimately related to the occupational demands of a given society. Much the same abilities and skills that are important in

schooling, therefore, are important also occupationally. Thus, we find that in industrialized countries practically all intelligence tests, scholastic aptitude tests, military classification tests, vocational aptitude tests, and the like, are quite similar in composition and that the scores obtained on them are all quite substantially intercorrelated. In short, there is a large general factor, or g, which the tests share in common and which principally accounts for the variance among individuals. When tests are devised to measure this g factor as purely as possible (i.e., in a factor analysis including a host of other tests it will have nearly all of its variance loaded on the general factor common to all the other tests and have little or no variance loaded on factors found only in

certain tests [special factors] or factors found only in small groups of tests [group factors]), examination of their item content leads to the characterization of it as requiring an ability for abstract reasoning and problem solving. Raven’s Progressive Matrices Test is an example of such a test. Tests having quite diverse forms can have equally high loadings on the g factor—for example, the verbal similarities and block design tests of the Wechsler Intelligence Scales are both highly loaded on g. Tests of g can be relatively high or relatively low in degree of “‘culture fairness.”’ (The

question “In what way are a wheel and a penny alike?” is probably more culture fair than the question “In what way are an oboe and a bassoon alike?’’) In short, it is possible to assess essentially the same intelligence by a great variety of means. Standard IQ tests measure the kinds of behavior in abstract and verbal problem situations that we call abstract reasoning ability. These tests measure more of g—the factor common to various forms of intelligence tests—than of any of the other more special ability factors, such as verbal fluency, spatial-perceptual ability, sensory abilities, or mechanical, musical, or artistic abilities, or what might be called social judgment or sensitivity. But a test that measured everything at once would not be very useful. IQ tests do reliably measure one important, though limited, aspect of

human performance. The IQ qualifies as an appropriate datum for scientific study. If we are to study intelligence, we are ahead if we can measure it. Our measure is the IQ, obtained on tests which meet certain standards, one of which is a high g loading when factor analyzed among other

tests. To object to this procedure by arguing that the IQ cannot be regarded as being interchangeable with intelligence, or that intelligence cannot really be measured, or that IQ is not

the same as intelligence, is to get bogged down in a semantic morass. It is equivalent to arguing that a column of mercury in a glass tube cannot be regarded as synonymous with temperature, or that temperature cannot really be measured with a thermometer. If the measurements are reliable and reproducible, and the operations by which they are obtained can be objectively agreed upon,

this is all that need be required for them to qualify as proper scientific data. We know that individually administered IQ tests have quite high reliability; the reliability coefficients are around .95, which means that only about 5% of the total individual differences variance is attributable to measurement error. And standard group administered tests have reliabilities close to .90. The standard error of measurement (which is about +5 for the Stanford-Binet and similar tests) ’ must always be taken into consideration when considering any individual’s score on a test. But it is

12 ) RACE ANDINTELLIGENCE [AS8 actually quite unimportant in comparison of the means of large groups of subjects, since errors of measurement are more or less normally distributed about zero and they cancel out when N is large. The reliability (i.e., consistency or freedom from errors of measurement) per se of the IQ is really | not seriously at issue in making comparisons between racial groups. If the samples are large, the © mean difference between groups will not include the test’s errors of measurement. The validity or importance of the measures derives entirely from their relationship to other variables and the importance we attach to them. The IQ correlates with many external criteria, and at the most general level it may be regarded

as a measure of the ability to compete in our society in ways that have economic and social consequences for the individual. In the first place, the IQ accords with parents’ and teachers’ subjective assessments of children’s brightness, as well as with the evaluations of children’s own peers. In terms of assessments of scholastic performance, whether measured in terms of school grades, teachers’ ratings, or objective tests of scholastic achievement, the IQ accounts for more of | the individual differences variance than any other single measurable attribute of the child. IQ accounts for about 50% of the variance in scholastic achievement at any single grade level, and over the course of several years or more of schooling it accounts for over 70% of the variance in overall scholastic performance. Since considerably less than 100 percent of the variance is accounted for, it means the IQ is not

an infallible ,predictor of the performance of any one individual. When used for individual diagnosis it must be evaluated in terms of many other factors in the child’s makeup and background and condition at the time of testing, and even then not too much stock should be placed in the IQ in predicting for the individual case, since the predictive validity of the IQ is not sufficiently high to override the effects of possibly unassessed traits or unpredictable unusual future circumstances which may radically alter the course of the individual’s development or performance in a statistically small proportion of cases. Thus, I am emphasizing the importance of evaluating the IQ somewhat differently when used for individual diagnosis and prediction than when used in making statistical predictions on large groups of individuals. It is somewhat analogous to actuarial predictions of insurance risks. Predictions for large groups classified by various criteria can be made with high degrees of certainty, while predictions for individual cases

are highly uncertain. ae Recently I received a letter from a high school senior who described himself as coming from a disadvantaged background. He had a strong desire to go on to college in hopes of becoming a lawyer, and he was wondering about his IQ and how much stock he should put in it in deciding his

further course. I doubt if there is much more sense in worrying about one’s own IQ than in worrying about the age at which one will die, as predicted by the insurance company’s actuarial tables. Among other things, | wrote the following to my student inquirer: “My own attitude toward tests, when I was a student, was not to give much thought to them but simply to set my sights on what seemed to me a realistic goal and then do my best to achieve it. You find out from those who have already made it what you have to know, what you have to be able to do, what skills you need to develop, and you set about doing these things just as you’d go about doing any kind of job that you know has to be done. If you set your goals too low, it’s too easy and you

won’t develop your potential. If you set goals that are unrealistically high, you become . discouraged. | reeommend one step at a time, each step being something you really think you can achieve if you really work for it. When you have made the first step successfully, then you will have a better idea of how to take the next step. That way, if you have whatever it takes, you’ll make it; if you haven’t got whatever it takes, you’ll find this out. But you’ll never really know without trying your best. I wouldn’t let any kind of test score determine what I try for. ‘The reality of your own performance in meeting the competition in striving toward your goals is the only real test. I believe this approach gives one the best chances of finally doing what he is best suited for, and this is one of the conditions for a satisfying life.”’ In statistical terms, however, the correlation is quite substantial between IQ and occupations, when the latter are merely ranked in the order of persons’ average judgment of the occupation’s prestige. Various studies have shown correlations in the range of .50 to .70. This is sufficiently high that the mean differences between groups of persons in occupations arranged according to a

Jensen | CAN WE AND SHOULD WE STUDY RACE DIFFERENCES? 13 prestige hierarchy (which is highly related to income) show highly significant differences in IQ or other mental test scores. In general, any two groups which differ in possessing what are perceived as “‘the good things in life’? according to the criteria and values of our society, will be found on the average to differ significantly in 1Q. Upward social mobility is related to IQ: the brighter children in a family tend to move up in socioeconomic level and the least bright tend to move down. There are exceptions to the general rule. Those who are born to wealth tend to be less able than those

who made it themselves—a quite predictable finding in terms of “regression to the mean.” Usually the regression of ability is much greater than the regression of cumulated wealth. The most ‘conspicuous exceptions, however, involve various disadvantaged minorities, whose social and economic positions are different from what one would predict in terms of IQ. For example, Negroes earn less income than whites of comparable IQ, education, family background, and work experience (Duncan et al. 1968). And American Indians, though considerably more impoverished than Negroes in the United States, score higher than Negroes on tests of intelligence and scholastic achievement (Kuttner 1968). Oriental children, who generally score at least as high on IQ as white children, also score considerably higher than would be predicted from their socioeconomic status (SES). This appears especially true of low SES Oriental children, who perform on a par with

middle-class white children on nonverbal tests (Lesser, Fifer, and Clark 1965). In predicting scholastic performance in school and in college, however, the evidence indicates that IQ tests and

scholastic aptitude tests work with about equal accuracy for all persons from whatever background. In this one respect, at least, the educational system seems to be one of the least | discriminatory institutions in our society. For example, there is no evidence that IQ tests predict. scholastic performance of Negro children less well than for white children, or that college entrance exams predict college grades less well for Negro than for white students (Jensen 1968b; Stanley and Porter 1967). The predictive validity of such tests could be lowered or changed, of course, by

altering the curriculum such that the predictors would no longer be as relevant and other predictors might then become more valid.

, When groups.are selected from the lower or upper extremes of the IQ distribution, the contrasts - are enormous. A classic example is Terman’s study of gifted children, selected in elementary school for IQs over 140, which constitutes the upper one percent of the population. These 1,528 children have been systematically followed. up to middle age (Terman and Oden 1959). The group -as a whole greatly exceeds a random sample of the population on practically every criterion of a successful life, and not just intellectual criteria. On the average the Terman group have markedly greater educational attainments, have higher incomes, engage in more desirable and prestigious - occupations, have many more entries in Who’s Who, have brighter spouses, enjoy better physical and mental health, have a lower suicide rate, a lower mortality rate, a lower divorce rate, and have - brighter children (their average IQ is 133). These results should leave no doubt that IQ is related to

socially valued criteria. '

IS INTELLIGENCE INHERITED?

The evidence on this point is very clear. There is no doubt of a large genetic component in individual differences in IQ. The methodology of determining the heritability of intelligence (or other traits) and the results of the applications of these methods to the study of intelligence have been reviewed in detail elsewhere (Burt 1958; Jensen 1967, 1969a,b). Heritability (H) refers to the

proportion of individual differences variance in a measurable trait, like intelligence, that can be attributed to genetic factors. 1-H, therefore, is the proportion of variance attributable to

non-genetic factors. These non-genetic factors are both biological and psychological. Some substantial proportion of the non-genetic IQ variance is unidentifiable, that is, is due to random environmental effects and to random stochastic biological processes in embryonic development. The heritability of IQ as estimated from the average of all published studies of the subject is .80, which means that on the average the studies show that 80% of the population variance in IQ is attributable to genetic variation, and 20% to nongenetic factors. The value of .80 is merely an average of many studies which yield H values that range from about .60 to about .90. There is no

' single true value of the heritability of a trait. Heritability is not a constant, but a population

14 ) RACE AND INTELLIGENCE [AS8 statistic, and it can vary according to the test used and the particular population sample tested. H will be affected by the range of genetic and environmental variation that exists in the population.

It should be noted that all the studies of the heritability of intelligence have been based on European and North American Caucasian populations. The results cannot strictly be generalized to other populations such as American Negroes. We would need to conduct heritability studies within

the Negro population if we are to have any certainty that our IQ tests are measuring a genetic component to the same degree in the Negro as in the white population. (Determining H in both populations still would not answer the question whether the group mean IQ difference between

Negroes and whites has a genetic component.) © oo

Non-genetic or environmental sources of variance can be analyzed into two major components: variance attributable to differences between families in the population and variance attributable to differences within families. The sum of the between families and the within families variances — constitutes the total of the nongenetic or environmental variance. Expressed as a proportion, it is 1 — A, = EH, and, as already pointed out the average value of A, reported in the literature is .80, making the average value of # = .20. The conceptually simplest method for estimating FE is to obtain the correlation between identical (monozygotic) twins reared apart (r.) 5 a) in uncorrelated environments (families). # = 1 —r.,>7,. The correlation between identical twins reared together (‘y7z7) in the same family is used to estimate the within families environmental variance E\,, = 1—

'mzt: the between families variance is then E, = 2 — Ey, or Myyz7+ 7 'wmza- When these formulations are applied to all the relevant twin data reported in the literature, the average values

they yield are & = .20, #, = .12 and £,, = .08 (Jensen 1967). Little, if any, of the b, is controllable. Some of it is due to prenatal effects related to mother’s age, health, accidental perinatal factors, ordinal position among other siblings, etc. In terms of our present knowledge, no

prescription could be written for reducing E,,. Some of it, in fact, is almost certainly due to random, stochastic developmental processes in the first weeks after conception, which means that even if we had perfect control over all the identifiable factors usually classified as environmental,

genetically identical individuals would still show some differences. The between families

component, E,, is probably much more attributable to what we commonly think of as

health care, and the like. re

environmental differences in terms of cultural-educational advantages, quality of nutrition, general SPECIFIC COMMENTS ON GENETICS AND IQ

To say that IQ tests like the Stanford-Binet “measure present ability, not inborn capacity”’ is misleading. Surely it measures present ability or performance. But the fact that the heritability (H,,) of the Stanford-Binet IQ is about .80 in English and North American Caucasian populations also means that the test measures “innate capacity,’ if by this term we mean the individual’s genotype for intellectual development. Since H is the proportion of variance in IQs (which are phenotypes) attributable to variance in genotypes, the square root of H represents the correlation between phenotype and genotype, and this correlation is about .90, a very high correlation indeed. (This is the correlation that exists after correction for attenuation, that is, test unreliability.) What the evidence on heritability tells us is that we can, in fact, estimate a person’s genetic

. standing on intelligence from his score on an IQ test. If the correlation between phenotype and genotype were perfect (i.e., 1.00), a person’s test score would, of course, be an exact index of his genetic potential. But since the correlation is only about .90, such statements can only be made on a probabilistic basis. If education and culturally-derived motivation strongly affect intelligence test performance,

then these factors should show up as part of the & variance, mostly E,, i.e., between families environmental variance. Heritability studies, as pointed out, show the & variance to be only about

20% of the total and F, only about 12% of the total. If group differences in IQ are to be explained in terms of educational and motivational factors, and if the heritability of IQ were the same in both groups, it would have to be assumed that all the members of one group differed from the mean of the other group by a constant amount in these motivational or other environmental variables. More will be said on this point in the later section on proposed genetic research.

Jensen | CAN WE AND SHOULD WE STUDY RACE DIFFERENCES? 15 The twin method may actually underestimate, rather than overestimate, the heritability of IQ. The reason is that there is considerable evidence that twins are more subject to prenatal stresses and nutritional disadvantages than singletons. This is reflected in the much lower birthweights of twins, the higher infant mortality of twins, and the fact that twins average 6 to 7 points lower in IQ than single-born children. One member of the twin pair is usually prenatally favored over the other, and this is especially true for monozygotic twins, as reflected in their differing birthweights. Birthweight of twins is positively correlated with later 1Q (Willerman and Churchill 1967). These prenatal differences, reflected in later IQ differences between the members of twin pairs, are very

probably greater for twins than for singletons and therefore suggest a larger component of (prenatal) environmental variance in twins than in singletons. Thus the argument that the twin method of estimating heritability leads to an overestimate and thereby underestimates the environmental component is very weak. A stronger case can be made for just the opposite conclusion. The fact that the estimates of H from the twin methods are in close agreement with estimates based on other kinships indicates that the twin estimates are not very deviant in either direction. Indeed, it is the consistency of H estimates arrived. at by different methods that makes them so impressive and reinforces their validity and scientific credibility (see Crow 1969). Cultural and educational differences are probably the most important non-biological sources of individual differences in intelligence, but they are not necessarily the most important non-genetic source of differences. It is likely that prenatal and nutritional factors are at least as important sources of variance as social-psychological factors. The sociological emphasis on the non-biological aspects of the environment has resulted in a relative neglect of probably important nutritional factors and maternal factors (age, health, diet, number of births, spacing of births, etc.) which can affect the prenatal and early childhood development of the individual. In reply to suggestions that our national IQ may be declining due to the possibility that the least able segment of the population is reproducing at a faster rate than the most able segment, some writers draw the familiar analogy between intelligence and physical stature. Both IQ and height are polygenetic traits and the same quantitative genetic model can be applied to both and can predict the various kinship correlations for IQ and height about equally well. It is also known that height, like intelligence, shows a positive correlation with socioeconomic status. Thus, if poor people have larger families than the well-to-do, we should expect the average height of the population to decrease over a number of generations. Exactly the same line of reasoning applies also in the case of intelligence. To counter this pessimistic prediction, it has been noted that despite what we should predict from simple genetic principles, the mean height of the population

not only has not decreased in the past 200 years or so, but has in fact increased by a very significant amount. The increase, it is assumed, is due to environmental factors such as improved

- nutrition. And the implication is, of course, that intelligence, too, will increase over generations because of improvements in the environmental factors relevant to intellectual . development. I believe this line of argument is weak and can lead to an unwarranted complacency about a possibly serious social trend. First of all, Carter (1962) and Tanner (1965, 1968) have pointed out that much if not all of the

increase in adult height in the past 200 years can be attributed to genetic factors, namely, the outbreeding effect. Increase in height is closely associated with the increase in the population’s mobility. The offspring of parents from different Swiss villages, for example, are taller than the offspring of parents born in the same village. This outbreeding effect, or hybrid vigor, tends to saturate or level off in the population in a few generations, as has already occurred with respect to : height in the United States. Nutritional factors have their greatest effect on rate of growth rather than on final adult height. In World War I men reached their full adult height at age 26; today they attain their full height at 18 or 19. Although it is true that height is positively correlated with socioeconomic status (SES) and that low SES families are larger than high SES families, these facts alone are not sufficient to warrant the prediction that the mean height of the population should decline. It would have to be shown that the same numbers of low SES persons as high SES persons have offspring. When this point . was investigated for intelligence, it was found that persons of below average IQ have larger families than persons of above average IQ, but that fewer of the below average ever marry or have any

16 , RACE ANDINTELLIGENCE [AS8 children at all (Higgins and Reed, 1962; Bajema 1963, 1966). The net result is a balance between the low and high IQ groups in the number of offspring they produce. This finding holds only for the white population of the US of a generation ago. No studies of this type have been conducted _

in the US Negro population. Since the bases for marriage and mate selection may be quite - different in various subcultures, the results of investigation of this problem in one group cannot be

generalized to other population groups with any confidence. The analogy with height is not convincing, since we have established only a negative correlation between height and family size, but have not taken into account the relative proportion of short and tall persons who never marry or produce offspring. Since we know there is selective mating for height in our population (that is, taller persons are viewed as more desirable) it is likely that fewer short persons marry or reproduce and that therefore a similar equilibrium between reproductive rates of short and tall persons exists as in the case of low and high intelligence. As I have noted elsewhere (Jensen 1968a, 1969a), _ certain statistics raise the question of whether Negro intelligence is declining relative to white — intelligence as a result of more extreme differential birthrates in lower and upper social classes among Negroes than among whites. Negro middle- and upper-class families have fewer children than their white counterparts, while Negro lower-class families have more. In 1960, Negro women married to professional or technical workers had only 1.9 children as compared with 2.4 for white women in the same circumstances. Negro women of ages 35 to 44 who were married to unskilled — - workers had 4.7 children compared with 3.8 for non-Negro women in the same situation, and Negro women with incomes below $2000 per year averaged 5.3 children (Moynihan 1966). This could mean that the least able segment of the Negro population is reproducing most rapidly, a condition that could alone produce and increase a genetic difference between the Negro and white populations in a few generations. The possible genetic and social implications of these trends have

not yet come under investigation and there are no data at present which would warrant

complacency about this important question. ; : ,

- Can genetic changes in a population take place only very slowly, so that selective pressures acting over several generations would be of negligible consequence? The answer, of course, depends largely on the ‘degree of selective pressure. We already know enough to permit fairly

accurate estimates of genetic trends given certain criteria of selection. If selection were extremely _ rigorous, an enormous shift in the population mean would be possible, as can be inferred from the _ average IQ of the offspring of the Terman gifted group. The Terman subjects were selected for _ Stanford-Binet IQs of 140 and above; they had a mean of 152. There was no selection of their spouses, except by the normal assortative mating that occurs for intelligence in our society (i.e., a correlation of .5 to .6 between spouses’ IQs). The offspring of the Terman gifted had an average IQ of 133 (Terman and Oden 1959). This is more than two standard deviations above the mean IQ of children born to a random sample of the population. There is a regression from the selected parent generation toward the general population mean, but the regression happens only once, and the ~ offspring of the selected parents will in turn have offspring without further regression, provided, of course, they do not mate outside the group of offspring from the selected parents. Rats have been

bred for maze learning ability and it has generally required from six to nine generations of selection to produce two strains of rats whose distributions of maze learning scores are completely non overlapping. IS RACE A VARIABLE?

One of the easiest ways of avoiding the issue of race differences in intelligence is to make the claim that there is no such thing as race and therefore it is not a variable that can be related to any other variables. Thus, proponents of this view would claim that the concept of race is merely a

myth, not a phenomenon that can be subjected to scientific study. This is, of course, utter nonsense. But it will pay to clarify the concept of race as it figures in comparative studies of

intelligence. :

There are two general definitions of race: the social and the biological (or genetic). Both are arbitrary, but this need not mean they are unreliable or lacking in precision. Although most of the studies of racial differences in intelligence are based on social definitions of race, it should be

Jensen | CAN WE AND SHOULD WE STUDY RACE DIFFERENCES? 17 noted that there is usually a high correlation between the social and the biological definitions, and , it is most unlikely that the results of the research would be very different if the investigators had used biological rather than social criteria of race in selecting groups for comparisons.

The social criteria of race are simple: they are the ethnic labels people use to describe themselves and the more obvious physical characteristics such as skin color, hair texture, facial

features, and so on, by which persons roughly determine one another’s “race.” Admitted, the =. social definition is crude. It does not take account of “‘borderline’”’ or ambiguous cases that are hard to categorize and which make for some unreliability in classification, and it does not take account

of the fact that there are no pure racial types—and especially in the case of American Negroes there is considerable racial admixture. Almost no American Negroes are of pure African descent; most have from 5% to 90% Caucasian genes, the average degree of admixture now being between 20% and 30%. Thus there is great genetic diversity within socially defined racial groups.

Does this make the social definition of race useless as a variable? No. In the first place, there is undoubtedly a high correlation between social and biological classification. That is to say, if one

were to sort school children, for example, into three socially defined racial groups, Negro, Oriental, and Caucasian, one would find a very high concordance of classification if he used strict biological criteria based on the frequencies of blood groups, anthropometric measures, and other genetic polymorphisms. What one would not have obtained from the crude social classification is degrees of racial admixture. In other words, the major racial categories would be much the same

whether constituted by social judgments or strict biological criteria. But if we wanted to go beyond this crude system of classification to make more refined differentiations, we would have to resort to biological criteria. Social judgments of degrees of racial admixture are quite unreliable. The broad categories, however, are reliable. They also qualify as variables in the sense that they show significant correlations with other variables such as IQ and scholastic performance. This is not to say that such correlations by themselves tell us anything about a biological or genetic basis for the correlation, which might be due to other environmental, social-class and cultural variables related to the socially defined racial classification. If the crucial variables in IQ differences are not racial classification per se, but other correlated environmental factors, then, at least in theory, one should be able to reduce the racial correlation with IQ to zero by partialling out the truly causal factors that are only incidentally correlated with both race and IQ. So far no one has succeeded in doing this as regards Negro-white comparisons. Every combination of environmental variables that anyone has partialled out has always left behind some significant correlation between race (socially defined) and IQ (Shuey 1965). One can always claim that all the relevant environmental variables were not taken into account. This is a real weakness of such studies and they can be legitimately criticized on this score. It is largely for this reason that our understanding of racial differences will

not be greatly advanced until more refined criteria of race based on biological criteria are employed. Specific proposals are made in a later section. It is strange that those who claim that there are no genetic racial differences in ability are often

the most critical of studies that have employed the social criterion of race rather than more rigorous genetic criteria. If the observed IQ differences are due only to social factors, then the social definition of race should be quite adequate, and, in fact, should be the only appropriate definition. If it is then argued that the two socially defined racial groups being compared are not “pure” and that each group contains some genetic admixture of the other, it can only mean that the biological racial aspects of the observed IQ differences has been underestimated by comparing ; socially defined racial groups. The biological definition of race is based on gene frequencies. Races are breeding groups which differ in the frequencies of one or more genes. A breeding group is one in which there is a higher proportion of matings among members of the group than of matings in which one member of the pair is from outside the group. Breeding groups result from relative degrees of geographical, racial,

and cultural isolation of different population groups. The definition of race by these criteria is arbitrary only in the sense that differences in gene frequencies is a continuous variable, and where one wishes to draw the lines as criteria for classification purposes is not dictated by nature but by - the taxonomic considerations of the investigator. Rather than thinking in terms of races, we should think in terms of groups with different gene frequencies. The question we would ask is

18 RACE AND INTELLIGENCE [AS8 whether various groups differing in gene frequencies also differ in 1Q, other things being, in effect,

equal. The major races are simply breeding populations that have a relatively high degree of _ inbreeding and differ from one another in the relative frequencies of many genes. They differ in so many known gene frequencies, in fact, that it seems highly improbable that they would not also differ in the frequencies of genes related to behavioral traits such as intelligence.

A major block to clear thinking about race is to think of it as a kind of Platonic essence, independent of any particular population group. General statements about the mental abilities of the “white race,” the “black race,” “the yellow race,” and so on, make no sense in terms of any studies that have yet been done or that seem at all feasible for the future. Strictly speaking, to ask if there are race differences in any characteristic is scientifically meaningless if what we mean by

race is not clearly specified. All we can do is study samples selected from certain specified populations. These samples cannot be regarded as representative of some Platonic racial groups. They are merely representative (if properly selected) of the clearly specified population group from which they are selected.

We could ask, for example, whether a population subgroup that differs from the general population in its average response to the educational and occupational requirements of our society differs in its gene pool from other population subgroups which are more successful, and if so, are some of the genetic differences related to ability factors with high heritability? Population subgroups which have immigrated are not necessarily representative of their native parent populations. Studies of racial or national groups in the United States, therefore, cannot be generalized abroad, and the reverse is also true. This does not mean, however, that meaningful comparative studies of various population subgroups within the United States are not feasible. The notion that there are no genetic mental ability differences among population subgroups that differ in many other gene frequencies is, in principle, hard to defend. Populations that have been widely separated geographically or socially for many centuries and which have been exposed to climatic and cultural conditions that exert different selective pressures are almost certain to differ genetically in many ways. And, in fact, they do. Nearly every anatomical and physiological

system studied has shown race differences. It is not at all necessary to invoke the factor of differential selective pressures to validate or explain some of these genetic differences, many of which confer no discernible advantage or disadvantage to survival or adaptation in any particular environment. A chemical substance, phenylthiocarbamide (PTC), is one illustration. To some

persons PTC is completely tasteless; to others it has a very unpleasant bitter taste. Whether a person is a taster is determined by a single gene. This gene has markedly different frequencies in | different racial groups. No one knows why this should be. Similarly, blood types have markedly different distributions in various racial groups, although it is not at all clear that one blood type is more advantageous than another in any given environment. In short, genetic diversity is the rule; genetic uniformity is the rare exception. By definition the gene pools of racial groups differ, and it is not at all an unreasonable hypothesis that genetic factors that condition behavioral development | also differ. Biological evolution generally is a slow process, but genetic changes with respect to particular

traits can occur relatively fast in response to selective pressures in the environment. In any case, biological evolution, whatever its rate, has resulted in marked genetic differentiation of human populations. Concerning the one standard deviation average IQ difference between Negro and white American populations, one writer stated, “A review of present knowledge on interracial divergence in man makes it unlikely that a difference as large as the observed one is genetic.” This hardly seems tenable in view of the fact that other traits show even greater racial differences than are found for intelligence. Height, like intelligence, is a polygenically inherited characteristic and is

probably less subject to selective pressures than intelligence, and yet we find racial (and even national or regional) differences of more than one standard deviation. In fact, two racial subgroups on the African continent, the Pygmies and the Watusi, differ in height by five to six standard deviations. Obviously biological evolution has, in fact, been sufficient to create marked differences in genetic characteristics. It is hard to imagine that there have not been different selection pressures for different abilities in various cultures and that these pressures would be as great for intelligence as for many physical

Jensen | CAN WE AND SHOULD WE STUDY RACE DIFFERENCES? 19 - characteristics which are known to differ genetically among racial groups. Individual differences in the abilities most relevant to a particular culture are highly visible characteristics and if they have consequences for the individual’s status in the social hierarchy or the culture’s system of rewards they will be traits subject to the genetic effects of sexual selection and assortative mating. If a trait is not very relevant to the demands of a particular culture it will not become highly visible, it will

not be a basis for selective mating, and its genetic basis will not be systematically affected by pressures in the social environment.

_ Selective mating refers to the fact that certain characteristics are viewed as desirable in mate selection by virtually all members of the breeding population. The usual consequence is that those standing higher on the desired trait will have greater opportunities for mating and reproduction while those at the lowest end of the distribution on the trait in question will be least likely to find

a mate and to leave progeny. The net effect is to boost the mean value of the trait in the population. Assortative mating refers to the fact that like tends to marry (or mate with) like. Itis sometimes an inevitable consequence of selective mating with respect to generally desirable traits,

but also holds for traits which are merely subject to various individual preferences. It is noteworthy that of all measurable human characteristics the one with the highest coefficient of assortative mating (i.e., the correlation between mates) is intelligence. The correlation between spouses’ IQs, for example, are around .5 to .6 in various studies, as contrasted with a correlation of .3 for height and of zero for fingerprints. The high degree of assortative mating for intelligence means that it is highly subject to genetic change through social influences. For example, the variance of the IQ distribution in the population would be reduced by approximately 20% if there were no assortative mating for just one generation. Assortative mating increases the variance of the

characteristic in the population, and if there is selective mating (as well as assortative) for the characteristic, the individuals at the lower (least desirable) end of the distribution will be least likely to reproduce. The net effect is to raise the average of the population on the trait in question. Such trends have probably taken place with respect to different traits in different societies for many centuries. While sexual selection may be capricious and non-adaptive with respect to many physical characteristics (e.g., various societies have different criteria of beauty), selection is not likely to be capricious with respect to those abilities which are salient in the competition in a given

society. There has probably been quite strong and consistent selection for different patterns of

ability in different cultures. A high degree of genetic adaptation to the demands of one environment might not constitute optimal adaptive capabilities to the demands of another, quite different, environment. As stated by Spuhler and Lindzey (1967:413) in their chapter on the

behavior-genetics of race difference: ,

it seems to us surprising that one would accept present findings in regard to the existence of .. genetic, anatomical, physiological, and epidemiological differences between races and still expect to find no meaningful differences in behavior between races.

They continue to point out that there are enormous discrepancies between races in the efficiency with which culture is transmitted (for example, the difference between literate and nonliterate societies). Some of these differences are closely associated with race differences, have existed for many thousands of years, and presumably have been accompanied by very different selection pressures in regard to characters potentially relevant to culture transmission, such as ‘intelligence.’

Thus, it seems highly improbable that there have been no markedly differing selective pressures ; on different subpopulations even within the United States. The selective pressures on Negroes must have been very different from those in European immigrant populations. The history of slavery suggests quite extreme selective factors, involving even the deliberate breeding of slaves for | certain characteristics which were irrelevant or perhaps even negatively correlated with intellectual prowess. It would be surprising indeed if more than 300 years of slavery did not have some genetic

consequences. But since the possible nature of these consequences is highly speculative and cannot be accurately inferred from historical accounts, this retrospective approach to the study of racial differences is too unreliable to be of much real scientific value. Direct genetical studies of . present population groups can provide the only really satisfactory basis for the scientific study of

genetic differences in abilities. :

20 , RACE ANDINTELLIGENCE [AS8 ARE THERE RACIAL DIFFERENCES IN IQ? In the United States persons classed as Negro by the common social criteria obtain scores on | the average about one standard deviation (i.e., 15 IQ points on most standard intelligence tests) below the average for the white population. One standard deviation is an average difference, and it is known that the magnitude of Negro-white differences varies according to the ages of the groups compared, their socioeconomic status, and especially their geographical location in the United States. Various tests differ, on the average, relatively little. In general, Negroes do slightly better on verbal tests than on non-verbal tests. They do most poorly on tests of spatial ability, abstract reasoning and problem solving (Shuey 1966; Tyler 1965). Tests of scholastic achievement also show about one standard deviation difference, and this difference appears to be fairly constant from first grade through twelfth grade, judging from the massive data of the Coleman study © (1966). The IQ difference of 1 SD, also, is fairly stable over the age range from about 5 yearsto adulthood, although some studies have shown a tendency for a slight increase in the difference between 5 and 18 years of age. Another point that has been suggested, but which requires much more systematic investigation before any firm conclusions can be reached, is that there is a larger sex difference in IQs for Negroes than for whites (Bronfenbrenner 1967). The presumed difference favors the females. The point is especially worthy of research because, if true, it would have. considerable social and. educational consequences, which would be especially evident in the upper tail of the IQ distribution. For example, if girls are a few IQ points higher than boys, on the

average, one should expect a greatly disproportionate number of Negro girls to qualify, as compared with boys, in any selection based on cut-off scores well above the mean, such as |

of 2 tol.

selection for college. Assuming a general mean of 85, an SD of 15, and a normal distribution, a 5

| point IQ difference between Negro boys and girls and a college selection cut-off score of 115, for example, we would expect the number of qualified girls to boys to be approximately in the ratio A statistic which has been much less studied thar the mean difference is the standard deviation

(SD), that is, the measure of dispersion of scores within the distribution. | Most studies agree in finding a smaller SD in Negro than in white IQs. The single largest

normative study of Stanford-Binet IQs in a Negro population, for example, found an SD of 12.4 as compared with 16.4 in the white normative sample (Kennedy, Van de Riet, and White 1963). This study is based on a large sample of school children in five Southeastern states and therefore may not be representative of the Negro population in other regions of the US. In general, however, most studies of Negro intelligence have found a smaller standard deviation than the SD of 15 or 16 generally found in white samples. The point is of some consequence in considering the relative merits of the opposing hypothesis relating to the causes of the observed average IQ difference between Negroes and whites, namely, the hypothesis of genetic equality versus the hypothesis of — genetic differences. If the distribution of [Qs in the Negro population does, in fact, have a smaller SD than in the white population, and if we hypothesize no genetic differences between the two populations, we must conclude that there is less variance due to environmental differences within the Negro group than within the white group. Since the genetic variance is hypothesized to be exactly the same in both groups, the difference in the variances (i.e., the square of the SD) of the sroups must be all environmental variance. Thus, if the total variance of Negro IQs is less than of white IQs, the genetic equality hypothesis is forced to predict a higher heritability of IQ in the Negro population than in the white; that is to say, more of the variance in Negro IQs would have to be due to genetic factors. If a study of the heritability of IQ in the Negro population yielded a heritability coefficient equal to or less than that found in the white population, this finding would contradict the genetic equality hypothesis, at least as regards the equality of genetic variance in the

two populations. |

Let us take another look at the Kennedy et al. (1963) data in this connection, to see how the hypothesis of genetic equality of variances comes out for this one set of data comparing the

distribution of Negro IQs with the distribution of the white population sample on which are based the norms for the Stanford-Binet Intelligence Test. It will be recalled that the SDs for Negroes and whites were 12.4 and 16.4, respectively. The variances are thus (12.4)? = 153.76 and (16.4)* =

Jensen | CAN WE AND SHOULD WE STUDY RACE DIFFERENCES? 21 268.96. Now, the best estimate of the heritability and Stanford-Binet [Qs in white population samples similar to that on which the Stanford-Binet was standardized is .80 (Jensen 1969a). This means that 80% of the variance of the white IQ distribution is genetic variance: thus, .80 x 268.96 = 215.17 is the white genetic IQ variance. But this is still greater than the total Negro IQ variance. The heritability of IQ in the white group would have to be assumed to be .57 for the white genetic variance to equal the total IQ variance of the Negro group, and surely some of this total variance is

non-genetic. Furthermore, no reported study of the heritability of Stanford-Binet IQs is as low as . .of, Thus, a hypothesis of genetic equality with respect to variances leads to highly untenable conclusions when applied to the data of Kennedy et al. (1963). By any canon of statistical and logical reasoning one is forced to reject the hypothesis that the distributions of genotypes for intelligence are equivalent in these two samples. By assuming genetic equivalence, one simply cannot make any sense out of the available data. This is not to say that one cannot question the data with respect toevery parameter that is involved in this line of reasoning. But if one accepts the validity of the heritability estimates in the white population and the SDs given by Kennedy et

al., it logically follows that a genetic equivalence hypothesis is untenable. It is, of course, statistically unwarranted to generalize this conclusion beyond the populations sampled in the study by Kennedy et al. The cause of the lesser variance of IQ in the Negro group is not known. One can only speculate and suggest hypotheses. From the evidence on the white population, for example, we know that some 15% to 20% of the total variance is attributable to assortative mating for intelligence; if the correlation between mates’ IQs was markedly reduced, the white IQ variance would be substantially reduced. (Variance due to assortative mating is all genetic variance.) Also, the covariance of heredity and environment (i.e., there is some correlation between children’s

genotypes for intellectual development and the quality of the environment in which they are

reared) constitutes some 5% to 10% of the total IQ variance in the white population. If environments were more similar, there would be less covariance and this source of variance would

be diminished in the total. We could find out if these factors or others, or some combination of factors, are responsible for the lesser variance in the Negro population only by carrying out complex heritability studies in the Negro population. A point that should be stressed is the fact that neither the white nor the Negro population, by

common social classification, is genetically homogeneous. It has already been noted that the American Negro is not of pure African ancestry but has, on the average, an admixture of 20% to 30% Caucasian genes, varying from less than 5% in some regions of the country to 40% or 50% in

others (Reed 1969). The white population contains many different subgroups which most probably differ genetically in potential for intellectual development. To point to one particular subgroup of one socially defined racial population as being higher or lower in IQ than some subgroup in another racial population proves nothing other than the fact that there exists an overlap between the racial groups. The fact that relatively large mean IQ differences are found between certain subgroups within the same race does not mean that these differences must be entirely of environmental origin and that therefore racial differences of similar magnitude must

also be entirely attributable to environment. :

Finally, it should be noted that IQ tests are taken by individuals. There is no such thing as

measuring the IQ of a group as a group. Individuals’ IQs are obtained as individuals. The basis on

which individuals may be grouped is a separate issue, depending upon the purpose of the investigator. When test scores are grouped according to some criteria of racial classification, we

find mean differences between the groups. If we group test scores by some criteria of socioeconomic status, we find mean differences between the groups. Conversely, if we group persons by levels of IQ, we find the groups differ in their proportions of persons of different races

and social classes. |

ARE RACE DIFFERENCES IMPORTANT?

There is, of course, nothing inherently important about anything. Race differences in _ intelligence are important only if people think these differences, or their consequences, are important. It so happens that in our society great importance is given to these differences and their

22 | RACE AND INTELLIGENCE [AS8 importance is acknowledged in many official public policies. Racial inequality in educational and occupational performance, and in the social and economic rewards correlated therewith, is today

clearly one of the uppermost concerns of our nation. |

Most persons are not concerned with those racial characteristics that are patently irrelevant to performance. The real concern results from the observed correlation between racial classification and educational and occupational performance. Persons who feel concerned about these observed differences demand an explanation for the differences. It is apparently a strongly ingrained human characteristic to need to understand what one perceives as a problem, and to ask for answers. People inevitably demand explanations about things that concern them. There is no getting around that. We have no choice in the matter. Explanations there will be.

But we do have’a choice of essentially two paths in seeking explanations of intelligence differences among racial groups. On the one hand, we can simply decree an explanation based on

prejudice, or popular beliefs, or moral convictions, or one or another social or political ideology, or on what we might think it is best for society to believe. This is the path of propaganda. Or, on the other hand, we can follow the path of science and investigate the problem in the same way that any other phenomena would be subjected to scientific study. There is nothing to compel us to one path or the other. This is a matter of personal preference and values. And since persons differ markedly in their preferences and values, we will inevitably see both of these paths being followed

for quite some time. My own preference is for a scientific approach to the study of these phenomena. It is certainly the more interesting and challenging intellectually. And our experience tells us that the scientific approach, by and large, leads to more reliable knowledge of natural phenomena than any other method that man has yet devised. If solutions to educational problems depend upon recognizing certain psychological realities in the same sense that, say, building a workable spaceship depends upon recognizing certain physical realities, then surely we will stand a

better chance of improving education for all children by choosing the path of scientific investigation. In facing the issue of race differences in abilities we should heed the statement of John Stuart Mill:

science. . If there are some subjects on which the results obtained have finally received the unanimous assent of all who have attended to the proof, and others on which mankind have not yet been equally successful; on which the most sagacious minds have occupied themselves from the earliest date, and have never succeeded in establishing any considerable body of truths, so as to be beyond denial or doubt; it is by generalizing the methods successfully followed in the former enquiries, and adapting them to the latter, that we may hope to remove fhis blot on the face of

Once we subscribe to a scientific approach, we are obligated to act accordingly. This means, for one thing, that we entertain alternative hypotheses. To entertain an hypothesis means not just to

pay lip service to it or to acknowledge its possible merit and let it go at that. It means to put it into a testable form, to perform the test, and report the results with information as to the degree of statistical confidence with which the hypothesis in question can be accepted or rejected. If we can practice what is called “strong inference,’’ so much the better. Strong inference consists of formulating opposing hypotheses and pitting them against one another by actually testing the contradictory predictions that follow from them. This is the way of science. How much of our educational research, we may ask, has taken this form? How much of the research that we see

| catalogued in the already gargantuan ERIC bibliography on the causes of the educational handicaps of children called culturally disadvantaged has followed this path? The only sensible conclusion one can draw from a perusal of this evidence is that the key question in everyone’s mind about racial differences in ability—are they genetic?—has, in effect, been ruled out as a serious alternative hypothesis in the search for the causal factors involved in inequalities of educational performance. Sundry environmental hypotheses are considered, but rarely, if ever, are alternative genetic hypotheses suggested. If a genetic hypothesis is mentioned, it is usually for the sake of dismissing it out of hand or to point out why it would be impossible to test the hypothesis in any case. Often, more intellectual ingenuity is expended in trying to find reasons why a particular genetic hypothesis could not be tested than in trying to discover a way of formulating the hypothesis so that it could be put to a test. The emotional need to believe that genetic factors

Jensen | CAN WE AND SHOULD WE STUDY RACE DIFFERENCES? 23 are unimportant in individual or group differences in ability can be seen in many statements by dedicated workers in those fields of psychology and education most allied to the problems of

children called disadvantaged. For example, Dr. Bettye Caldwell, a prominent worker in compensatory and early childhood education has noted: _ Most of us in enrichment ... efforts—no matter how much lip service we pay to the genetic

potential of the child—are passionate believers in the plasticity of the human organism. We need _ desperately to believe that we are all born equalizable. With any failure to demonstrate the

__ effectiveness of compensatory experiences offered to children of any give age, one is entitled to

conclude parsimoniously that perhaps the enrichment was not offered at the proper time [Quoted by Baratz and Baratz 1969]. | But genetic factors in rate of development are never considered as a possible part of the explanation. It is important not to evaluate persons in terms of group membership if we are to insure equality of opportunity and social justice. All persons should be treated as individuals in terms of their own merits if our aim is to maximize opportunities for every person to develop his abilities to their fullest capacity in accord with his own interests and drives. But the result of individual selection (for higher education, better jobs, etc.) makes it inevitable that there will be unequal representation of the parent populations in any subgroup that might be selected whenever

there are average differences between parent populations. . Many questions about the means of guaranteeing equality of educational opportunity are still moral and political issues at present. When there is no compelling body of scientific evidence on which policy decisions can be based, such decisions must be avowedly made in terms of one’s personal social philosophy and concepts of morality. Many goals of public policy must be decided in terms of values. The results of research are of greatest use to the technology of achieving the

value-directed goals of society. The decision to put a man on the moon was not a scientific decision, but once the decision was made the application of scientific knowledge was necessary to achieve this goal. A similar analogy holds for the attainment of educational goals.

. CAN RACE DIFFERENCES BE RESEARCHED? It is sometimes argued that even though it is not unreasonable to hypothesize genetic racial

differences in mental ability, we cannot know the direction or magnitude of such genetic differences and the problem is much too difficult and complex to yield to scientific investigation.

Therefore, the argument often continues, we should go on pretending as though there is no question of genetic differences, as was officially stated by the US Office of Education in 1966: “It is a demonstrable fact that the talent pool in any one ethnic group is substantially the same as that in any other ethnic group.”’ First, we will never know to what extent research can yield answers on a subject unless we at least try our best to do the research. It is doubtful that any major scientific advances could have been made in any field if it were decided beforehand that the problems could not be researched. I cannot agree that a scientific approach should be restrictied to only the easy problems. If all the

necessary methodology for studying the genetics of race differences in psychological characteristics is not yet sufficiently developed, this should not be surprising, since so little effort

has been made thus far. The methodology of a field of inquiry does not grow in a vacuum. Scientists do not first develop a complete methodology for the investigation of a complex area and

then apply it all at once to get the final answers. An appropriate methodology evolves as a result of : prappling with difficult problems in the spirit of scientific research. Darwin’s theory of evolution

did not begin with a fully developed methodology adequate to prove the theory, nor did the theory of the inheritance of acquired characteristics—a theory which was later disproved after the development of an adequate methodology, a methodology which would not have developed in the

absence of attempts to research this theory. No one would have been inclined to invent the necessary research methods in the absence of the problems these methods were needed to solve. One critic states “The scientific problem [of genetic race differences in ability] itself seems of -. dubious validity, if one considers how great are the difficulties . . ., at least on the basis of present techniques.’ The same statement could have been made about research on the theory of evolution,

24 RACE ANDINTELLIGENCE [AS8 the atomic theory, the gene theory, and so on. We do not expect any single study or experiment to reduce all the uncertainty about a complex subject to absolute zero in one bold stroke! But asin _ dealing scientifically with most other complex phenomena, we should not regard ourselves as so intellectually impotent as to be unable to gradually chip away at the heredity-environment uncertainty with whatever tools that scientists can muster or devise with their present knowledge and ingenuity.

What are some of the thinking blocks in this area? One is the frequent failure to distinguish between raw facts, on the one hand, and inference from the facts in terms of some hypothesis, on the other. The Society for the Psychological Study of Social Issues (SPSSI), for example, in a press release (May 2, 1969) criticizing my article in the Harvard Educational Review (Jensen 1969a), stated, ““There is no direct [italics mine] evidence that supports the view that there is an innate | difference between members of different racial groups.’ Of course there is not direct evidence, nor - can there be direct evidence if by “‘direct’’ we mean evidence that is immediately palpable to our physical senses. The gradual disappearance of ships over the horizon is not direct evidence of anything, but it can be interpreted in terms of the hypothesis that the earth is round. It would be

harder to explain if we hypothesized that the earth is flat. So even as relatively simple an hypothesis as that the world is round cannot be proved by direct evidence, but depends upon logical inference from diverse lines of evidence. If all that was needed was direct evidence, even a monkey would know that the world is round, in the same sense that it knows that a lemon is sour.

The substantiation of an hypothesis in science depends upon objective evidence but does not necessarily depend upon direct evidence alone.

Another inhibition to thought on this topic is the notion that before research can yield any answers, the environment must be absolutely equal for all groups involved in comparisons. The SPSSI statement went so far as to say that “. .. a more accurate understanding of the contribution of heredity to intelligence will be possible only when social conditions for all races are equal and when this situation has existed for several generations.’ Since no operationally testable meaning is given to “‘equal”’ social conditions, such a statement, if taken seriously, would completely preclude

the possibility of researching this important question, not just for several generations, but indefinitely. Actually, large environmental differences between racial groups can be revealing when the environmental ratings are positively correlated with IQ or scholastic performance within _ the groups but show a negative correlation between the groups. If group A on the average has a poor environment in terms of variables claimed to be important to intellectual development and sroup B has a good environment, and if group A performs better than group B on intelligence tests which are appropriate to the experience of both groups, this is evidence that some factors other than the measured environmental variables are involved in the relatively higher intellectual performance of group A as compared with group B. If environmental factors cannot be found that _ will account for the difference, it is presumptive evidence in favor of the genetic hypothesis. Genetical tests of the hypothesis are preferable, of course. (These are discussed in a later section). But what one also looks for are consistencies among various lines of evidence, especially lines of evidence that lead to opposite predictions from different hypotheses. Many investigators now would question the view that the lack of early stimulation in the preschool years can be counted among the chief causes of the poorer IQ performance of Negro

, children, since when children are grouped in several categories according to their parents’ socioeconomic status, the Negro children in the highest SES category still score two to three IQ points below white children in the lowest SES level (Shuey 1966). Thus, what we generally think of as a reasonably good environment is apparently not sufficient to equalize the performance of

Negro and white groups. .

Such findings lead to hypothesizing increasingly subtle and hard to measure environmental effects. But it should be recognized that at present most of the environmentally ‘“‘damaging’’ effects that are assumed to be accountable for performance differences are hypothetical and not factual. Poor self-concept and alienation are among the currently prevailing explanations, but what

has not yet been satisfactorily explained is why such general motivational dispositions should affect some cognitive abilities so much more than others. Performance is not uniformly low on all _ tasks, by any means. There are distinct high and low points in the profile of various abilities in __

Jensen | CAN WE AND SHOULD WE STUDY RACE DIFFERENCES? 25 different ethnic groups (Stodolsky and Lesser 1967), and no one has yet attempted to explain how such profile differences, which are invariant across social classes, could come about as a result of

differences in generalized attitudes and motivation in the test situation. | |

Finally, unnecessary difficulties arise when we allow the scientific question to become mixed up with its possible educational, social, and political implications. The scientific question and its solution should not be allowed to get mixed up with the social-political aspects of the problem, for _

when it does we are less able to think clearly about either set of questions. The question of whether there are or are not genetic racial differences in intelligence is independent of any questions of its implications, whatever they may be. But I would say that the scientific question should have priority and the answer should be sought through scientific means. For although the answer might have educational and social implications, and there are indeed grave educational and social problems that need to be solved, we must first understand the causes of problems if we are to do anything effectively toward solving them. Gaining this knowledge is a scientific task. As it is

accomplished, we are then in a better position to consider alternative courses of action and evaluate their feasibility and desirability in terms of society’s values and goals. This moves the problem into the realm of public policy, where all the answers cannot be scientifically derived. But ,

policy cannot be wisely or effectively formulated unless-it is informed by the facts. No matter how . well-intentioned it may seem to be, it can only be less effective and-less beneficial if it is based on false premises or in contradiction of reality.

| . GENETIC RESEARCH

TO REDUCE THE HEREDITY-EN VIRONMENT UNCERTAINTY

Today there is virtually no uncertainty among those who have attended to the evidence that individual variation in intelligence is predominantly conditioned by genetic factors and that environmental factors account for a lesser proportion of the phenotypic variance. One can point to variations among studies that have estimated the heritability of intelligence. Such variations in estimates of the proportion of variance attributable to genetic factors are to be expected in view of the great variety of populations sampled and the differences among the variety of tests of mental ability that have been used. Despite these expected variations in heritability estimates, it is

important to note that no major study contradicts the conclusion that heredity contributes something more than twice as much to the variance in IQ as environment in white European and American populations. (We do not have good heritability data on other populations. ) The term “heredity-environment uncertainty’’ refers mainly to the question of race differences in intelligence. The answer to this question is still in the realm of uncertainty in terms of the

normal scientific meaning of this word. Absolute certainty is never attained in an empirical science. Absolute certainty can be had only in pure mathematics, the certainty of which rests upon

the fact that pure mathematics is, as Bertrand Russell pointed out, just one vast tautology. Empirical science deals in probability statements, and “certainty” refers to a high degree of probability that a proposition is “‘true,’’ meaning that certain objective consequences can be predicted from the proposition with a stated probability. A decisive increase in this probability with respect to any given scientific proposition rarely results from a single experiment or discovery. I take exception to the impression that might be given by some writers that unless a scientific study can be perfect and 100% certain, we cannot know anything. This is not how scientific knowledge advances. We do not devise perfect methods or obtain complete answers on : the first try. Certainty, in the sense of probability, is generally increased very incrementally in

science. Research aims to add reliable increments to statements of probability. | This we must continue to do with respect to the question of genetic race differences in

intelligence. It is still an open question by all reasonable scientific standards. The existing evidence is in all cases sufficiently ambiguous, due largely to the confounding of racial and environmental

factors, as not to permit statements with a sufficiently high probability such that all reasonable and qualified persons viewing the evidence will agree that it is conclusive. The issue of genetic race differences may be likened to theories of the moon’s craters—whether they were caused by volcanic eruptions or by the impact of meteors. All the evidence obtainable by astronomers could

26 | RACE AND INTELLIGENCE [AS8 support either interpretation, and different scientists could argue for one theory or the other. A substantial increment could be subtracted from this uncertainty only by obtaining new evidence not obtainable through telescopic study, namely, directly obtaining and analyzing material from —

the surface of the moon. :

I believe that, similarly, the heredity-environment uncertainty about race differences in IQ will

be substantially reduced only by obtaining new evidence—new kinds of evidence. Exclusive reliance on anthropological, sociological, and psychological evidence would probably not substantially advance our knowledge. I believe that application of the methods of biometrical genetics (also called population genetics or quantitative genetics) to the question of race differences will substantially reduce our uncertainty. Someone suggested that the only way one could prove race differences in intelligence would be

to dye one member of a pair of white identical twins black and adopt it out to a Negro family while the co-twin is reared by a white family. How much difference would it make in their IQs? Better yet is the suggestion of Professor Arthur Stinchcombe (1969): find pairs of identical twins in which one member of each pair is Negro and one is white, separate them at birth and rear them in Negro and white families and see how their IQ differences compare with those found for twins where both are of the same race! These suggestions sound ridiculous; one is unfeasible and the

other is impossible. Yet as conceptual experiments they are good, because they suggest the necessary ingredients of the information we must obtain to reduce the heredity-environment uncertainty. Both examples rightly recognize skin color (and, by implication, other visible racial features) as a part of the individual’s environment. They are based on comparing genetically equivalent persons reared in different environments. Another possibility consists of rearing genetically and racially different persons in essentially similar environments—including the factor of skin color, etc. Is such a study possible? Yes. Geneticists already know the frequencies of a large number of genetically independent blood groups in European and African populations. On the basis of such data, it is entirely possible to determine the proportion of Caucasian genes in a population sample of Negroes, socially defined. Furthermore, it should be possible by the same means to classify individuals on a probabilistic basis in terms of their relative proportions of African and Caucasian genes. Since the average admixture of Caucasian genes for American Negroes is between 20% and 30%, there should be |

enough variance to make it possible to assign large numbers of individuals to at least several categories according to their amount of admixture, and the probable error in classification could be quite definitely specified. A sufficient number of blood groups or other genetic polymorphisms

with known frequency distributions in African and Caucasian populations would have to be employed to ensure a high degree of statistical certainty that the categories represented different degrees of genetic racial admixture. A wide range of admixtures probably exists among Negroes living in highly similar environments, so that it should be quite possible in such a study to obtain samples which do not differ across the admixture categories in a number of socioeconomic or other environmental indices. What about skin color? It is polygenetic and is very imperfectly correlated with the amount of Caucasian admixture. Individuals, for example, whose genes are derived in equal (50-50) proportions from African and Caucasian ancestors evince the full range of

skin colors from white to black, including all the shades between. This makes it possible . statistically to control the effect of skin color; that is, one can compare a number of persons all of whom have the same skin color but different degrees of African/Caucasian admixture, or conversely, the same degree of admixture but different skin colors. (Skin color can be quantified precisely and objectively by means of a photoelectric device which measures reflectance.) The question, then, would be: do the mean IQs (or any other mental ability tests) of the several categories of racial admixture differ significantly and systematically? The genetic equality

hypothesis would predict no difference; the genetic inequality hypothesis would predict a difference between the groups. A further refinement, in order to ensure greater equality of environmental conditions across the

admixture categories, including prenatal environment, would be to include in the study a large number of half-siblings all related through the mother and reared together. Some half-siblings will inevitably fall into different admixture categories. Do they differ significantly on mental tests

Jensen | CAN WE AND SHOULD WE STUDY RACE DIFFERENCES? 27 when skin color is controlled? Birth order, maternal age, and other factors would have to be noted,

but in large samples these factors would probably tend to be random with respect to racial admixture. One would also want a white control group with no African admixture in order to rule out the remote possibility that the blood groups themselves are causally related to IQ, since they are intended in this study only as genetic markers or indices of racial admixture. Such a study

would go further toward answering the question of Negro-white genetic differences in intelligences . than the sum total of all the other studies that we now have.

The possibility has been suggested of using genetic linkages for studying the inheritance of intelligence and race differences, but evaluation of its potential merits will have to be decided by

geneticists. If the genes for some clearly identifiable physical trait are located on the same chromosome as the genes for some measurable mental ability, we should expect to find a marked correlation in the population between the appearance of the physical characteristic and the mental attribute whose genes share the same chromosome. The physical characteristic would thus serve as

an objective genetic marker for the mental trait. :

The major difficulty with this approach may be that what we call intelligence is so polygenetic that the relevant genes are carried on most or all of the chromosomes, so that specific linkages could never be established. If intelligence consists of a large number of subabilities, each of which is conditioned independently by a very limited number of genes which are carried on a single chromosome, then it may be possible to study linkages, provided we can reliably measure the subabilities. I have described elsewhere how psychologists might make their measurements of abilities of greater interest and value to researchers in genetics (Jensen 1968b). Briefly, it would consist of the fractionation of mental abilities to the most extreme limits that reliability of measurement will permit, and then seeing if these subabilities show any signs of relatively simple genetic inheritance (such as showing Mendelian ratios) or genetic linkages.

Are there any known linkages between physical and mental characteristics in the normal distribution of intelligence? I do not know of any established examples. We should begin looking for such possible mental linkages with blood groups, biochemical variations, and other physical traits. One set of interesting findings concerns the association between uric acid level in the blood and intellectual achievement. Whether this is an instance of genetic linkage or whether there is a causal connection between uric acid and brain functions is not yet established. Stetten and Hearon

(1958) reported a correlation between serum uric acid concentration and scores on the Army intelligence test of 817 inductees. A study of serum urate levels of 51 University of Michigan professors found a positive correlation with drive, achievement, and leadership (Brooks and Mueller 1966), and high school students have been found to show a similar relationship (Kasl, Brooks, and Cobb 1966). It would be interesting to know if these correlations are found within other racial groups and also if there are differences between groups in serum uric acid levels. Every bit of such various kinds of information, if it points consistently in the same direction, reduces to some extent the heredity-environment uncertainty.

There are other promising approaches to this problem through biometrical genetics, but explication of the technical aspects of these methods is clearly beyond the possible scope of the present discussion.

| IMPLICATIONS FOR EDUCATION Since educators have at least officially assumed that race and social class differences in , scholastic performance are not associated with any genetic differences in growth rates or patterns

of mental abilities but are due entirely to discrimination, prejudice, inequality of educational opportunity, and factors in the child’s home environment and peer culture, we have collectively given little if any serious thought to whether we would do anything differently if we knew in fact that all educational differences were not due solely to these environmental factors. There have been. and still are obvious environmental inequities and injustices which have disfavored certain minorities, particularly Negroes, Mexican-Americans, and American Indians. Progress has been made and is continuing to be made to improve these conditions. But there is no doubt still a long way to go, and the drive toward further progress in this direction should be given

28 RACE AND INTELLIGENCE [AS8 top priority in our national effort. Education is one of the chief instruments for approaching this goal. Every child should receive the best education that our current knowledge and technology can provide. This should not imply that we advocate the same methods or the same expectations for all children. There are large individual differences in rates of mental development, in patterns of ability, in drives and interests. These differences exist even among children of the same family. The © good parent does his best to make the most of each child’s strong points and to help him on his weak points but not make these the crux of success or failure. The school must regard each child, and the differences among children, in much the same way as a good parent should do. ——I believe we need to find out the extent to which individual differences, social class differences, and race difference in rates of cognitive development and differential patterns of relative strength and weakness in various types of ability are attributable to genetically conditioned biological growth factors. The answer to this question might imply differences in our approach to improving the education of all children, particularly those we call the disadvantaged, for many of. whom

school is now a frustrating and unrewarding experience.

Individuals should be treated in terms of their individual characteristics and not in terms of their group membership. This is the way of a democratic society, and educationally it is the only procedure that makes any sense. Individual variations within any large socially defined group are always much greater than the average differences between groups. There is overlap between groups in the distributions of all psychological characteristics that we know anything about. But dealing

with children as individuals is not the greatest problem. It is in our concern about the fact that when we do so, we have a differentiated educational program, and children of different socially identifiable groups may not be proportionately represented in different programs. This is the “hang-up” of many persons today and this is where our conceptions of equal opportunity are most

likely to go awry and become misconceptions. .

Group racial and social class differences are first of all individual differences, but the causes of the group differences may not be the same as of the individual differences. This is what we must find out, because the prescription of remedies for our educational ills could depend on the answer. Let me give one quite hypothetical example. We know that among middle-class white children,

learning to read by ordinary classroom instruction is related to certain psychological developmental characteristics. Educators call it “readiness.” These characteristics of readiness appear at different ages for different kinds of learning, and at any given age there are considerable individual differences among children, even among siblings reared within the same family. These

developmental differences, in middle-class white children, are largely conditioned by genetic. factors. If we try to begin a child too early in reading instruction, he will experience much greater | difficulty than if we waited until we saw more signs of “‘readiness.”’ Lacking readiness, he may

even become so frustrated as to “turn off” on reading, so that he will then have an emotional block toward reading later on when he should have the optimal readiness. The readiness can then not be fully tapped. The child would have been better off had we postponed reading instruction for six months or a year and occupied him during this time with other interesting activities for which he was ready. Chances are he would be a better reader at, say, 10 or 11 years of age for having started a year later, when he could catch on to reading with relative ease and avoid the unnecessary frustration. It is very doubtful in this case that some added “‘enrichment”’ to his preschool environment would have made him learn to read much more easily a year earlier. If this is largely a matter of biological maturation, then the time at which a child is taught in terms of his own schedule of development becomes important. If, on the other hand, it is largely a matter of

preschool environmental enrichment, then the thing to do is to go to work on the preschool : environment so as to make all children equally ready for reading in the first grade. If a child’s difficulty is the result of both factors, then a combination of both enrichment and optimal

developmental sequencing should be recommended. :

There is a danger that some educators’ fear of being accused of racial discrimination could . become so misguided as to work to the disadvantage of many minority children. Should we deny differential educational treatments to children when such treatment will maximize the benefits they receive from schooling, just because differential treatment might result in disproportionate representation of different racial groups in various programs? I have seen instances where Negro

Jensen | CAN WE AND SHOULD WE STUDY RACE DIFFERENCES? 29 children were denied special educational facilities commonly given to white children with learning

difficulties, simply because school authorities were reluctant to single out any Negro children, despite their obvious individual needs, to be treated any differently from the majority of youngsters

in the school. There was no hesitation about singling out white children who needed special attention. Many Negro children of normal and superior scholastic potential are consigned to classes

in which one-fourth to one-third of their classmates have IQs below 75, which is the usual

borderline of educational mental retardation. The majority of these educationally retarded , children benefit little or not at all from instruction in the normal classroom, but require special attention in smaller classes that permit a high degree of individualized and small group instruction. Their presence in regular classes creates unusual difficulties for the conscientious teacher and detracts from the optimal educational environment for children of normal ability. Yet there is reluctance to provide special classes for these educationally retarded children if they are Negro or Mexican-American. The classrooms of predominantly minority schools often have 20% to 30% of such children, which handicaps the teacher’s efforts on behalf of her other pupils in the normal range of IQ. The more able minority children are thereby disadvantaged in the classroom in ways that are rarely imposed on white children for whom there are more diverse facilities. Differences in rates of mental development and in potentials for various types of learning will not disappear by

being ignored. It is up to biologists and psychologists to discover their causes, and it is up to educators to create a diversity of instructional arrangements best suited to the full range of educational differences that we find in our population. Many environmentally caused differences can be minimized or eliminated, given the resources and the will of society. The differences that remain are a challenge for publit education. The challenge will be met by making available more ways and means for children to benefit from schooling. This, Iam convinced, can come about only through a greater recognition and understanding of the nature of human differences.

, NOTE

1 Por example, one need not accept the IQ scale as the most appropriate. If it could be argued and demonstrated that some transformation of the IQ scale produced more orderly and lawful data in studies of heritability, in the degree of normality of the distribution of scores, and in more closely approximating a genetic model, then such a transformation would be justified: It could very well affect the variances of the distributions in different population subgroups. Berkeley geneticist Dr. Jack King, for example, has suggested that if we assume that the factors (genetic and environmental) that affect intelligence do not behave additively but interact multiplicatively (i.e., a factor adds or subtracts a given percentage to the total measure rather than a fixed amount) a logarithmic transformation of the IQ scale is theoretically justified. In the multiplicative model, the logarithm of the observed measure is normally distributed. The logarithmic transformation in fact makes the IQ distribution more normal (Gaussian) in a number of studies, and it tends to

equalize the variances of the Negro and white distributions, although it also has the effect of pulling their means slightly further apart. The proper transformation is 100 (1 + 1n IQ/100),

which leaves the general population mean at IQ 100. (1n is the natural logarithm.) Past studies of the heritability of intelligence should be re-analyzed using this logarithmic transformation of the TQ scale to see if it gives a closer and more parsimonious fit to a polygenic model.

Bajema, C. J. | REFERENCES CITED

1963 Estimation of the Direction and Intensity of Natural Selection in Relation to Human Intelligence by Means of the Intrinsic Rate of Natural Increase. Eugen. Quart. 10:175-187.

1966 Relation of Fertility to Educational Attainment in a Kalamazoo Public School Population: A Follow-up Study. Eugen. Quart. 13:306-315.

Baratz, S. S., and Joan C. Baratz |

1969 Early Childhood Intervention: The Social Science Base of Institutional Racism. Paper

presented to the Society for Research in Child Development, Santa Monica, Calif., March. Bronfenbrenner, U. : . 1967 The Psychological Costs of Quality and Equality in Education. Child Developm. 38 :909-925.

30 RACE AND INTELLIGENCE [AS8

195:415-418. | Coleman, J. S., et al. a Crow, J. F. FF Higgins, J.,S. Reed and E. Reed : Jensen, A. R. ) Brooks, G. W., and E. Muller

1966 Serum Urate Concentrations Among University Professors. J. Amer. Med. Assn.

Burt, C.

1958 The Inheritance of Mental Ability. Amer. Psychol. 13:1-15.

Carter, C. O.

1962 Human Heredity. Baltimore: Penguin Books.

1966 Equality of Educational Opportunity. US Dept. of Health, Education, and Welfare.

1969 Genetic Theories and Influences: Comments on the Value of Diversity. Harvard

Educational Review 39:301-309. -

Duncan, O. D., D. L. Featherman, and Beverly Duncan | 1968 Socioeconomic Background and Occupational Achievement: Extensions of a Basic |

Model. Final Report, Project No. 5-0074 (EHO-191) US Dept. of Health, Education, and ; Welfare, Office of Education, Bureau of Research, May.

1962 Intelligence and Family Size: A Paradox Resolved. Eugen. Quart. 9:84-90.

1967 Estimation of the Limits of Heritability of Traits by Comparison of Monozygotic and Dizygotic Twins. Proc. Nat. Acad. Sci. 58:149-157.

1968a Social Class, Race, and Genetics: Implications for Education. Amer. Educ. Res. J. 5:1-42.

1968b Another Look at Culture-Fair Tests. In Western Regional Conference on Testing Problems, Proceedings for 1968. ‘‘Measurement for Educational Planning,” pp. 50-104. Berkeley: Educational Testing Service, Western Office.

1969a How Much Can We Boost IQ and Scholastic Achievement? Harvard Educ. Rev. 39:1-128.

1969b Reducing the Heredity-Environment Uncertainty. In Environment, Heredity, and Intelligence. Harvard Educ. Rev. Reprint Series No. 2, pp. 209-248. ~ Kas], S. V., G. W. Brooks, and 8S. Cobb

1966 Serum Urate Concentrations in Male High School Students. J. Amer. Med. Assn.

198:713-716. Kennedy, W. A., V. Van de Riet, and J. C. White

1963 A Normative Sample of Intelligence and Achievement of Negro Elementary School Children in the Southeastern United States. Monogr. Soc. Res. Child Developm. 28(6). on

Kuttner, R. E. 1968 Letters to and from the Editor. Perspect. Biol. Med. 11:707-709. Lesser, G. S., G. Fifer, and D. H. Clark 1965 Mental Abilities of Children from Different Social-Class and Cultural Groups. Monogr. _ Soc. for Res. in Child Developm. 30(4). Moynihan, D. P. 1966 Employment, Income, and the Ordeal of the Negro Family. In The Negro American, pp. 134-159. T. Parsona and K. B. Clark, eds. Cambridge: Houghton-Mifflin. Reed, T. E. 1969 Caucasian Genes in American Negroes. Unpublished manuscript, March.

Shuey, Audrey M. ,

1966 The Testing of Negro Intelligence, Second edition. New York: Social Science Press. | Spuhler, J. N., and G. Lindzey 1967 Racial Differences in Behavior. In Behavior-Genetic Analysis. J. Hirsch, ed. New York:

McGraw-Hill. ; Stanley, J. C., and A. C. Porter ; 1967 Correlation of Scholastic Aptitude Test Score with College Grades for Negroes Versus

Whites. J. Educ. Meas. 4:199-218. Stetten, D., Jr., and J. Z. Hearon

1969 Intellectual Level Measured by Army Classification Battery and Serum Acid

Concentration. Science 129:1737. |

Stinchcombe, A. L.

1969 A Critique of Arthur R. Jensen’s “How Much Can We Boost IQ and Scholastic

Achievement?’ Harvard Educ. Rev. 39(3). :

Stodolsky, 8. S., and G. Lesser

: 1967 Learning Patterns in the Disadvantaged. Harvard Educ. Rev. 37:546-593. Tanner, J. M.

1965 The Trend Towards Earlier Physical Maturation. In Biological Aspects of Social

Problems, pp. 40-66. J. E. Meade and A. S. Parkes, eds. New York: Plenum Press. 1968 Earlier Maturation in Man. Sci. Amer. 218:21-28.

Jensen | CAN WE AND SHOULD WE STUDY RACE DIFFERENCES? 31 Terman, J. M., and M. Oden 1959 The Gifted Group at Mid-Life. Stanford: Stanford University Press. Tyler, Leona E.

1965 The Psychology of Human Differences, Third edition. New York: Appleton-Century-

Crofts. Willerman, L., and J. A. Churchill 1967 Intelligence and Birth Weight in Identical Twins. Child Developm. 38 :623-629.

Intelligence in Black and White . ALEXANDER ALLAND, JR.

DEPARTMENT OF ANTHROPOLOGY COLUMBIA UNIVERSITY |

SINCE I AM NEITHER physical anthropologist nor statistician, my criticism of Jensen’s paper (1969a) draws upon the criticisms advanced by others more competent to judge his work. I should like to say in passing that, in my opinion, the most important criticism of the genetic hypothesis which attempts to separate whites and blacks in the United States into unequally endowed groups with regard to intelligence is that offered by Cohen (1969), which, incidentally, was written before this controversy began and represents independent research rather than a direct answer to what has come to be known as “‘jensenism.”’

The population geneticist Crow (1969:307-308) finds much of value in Jensen’s analysis, but offers an important caveat when it comes to “‘group differences, especially racial differences’’: Heritability studies have been confined almost exclusively to white populations and to highly normal environments. How relevant are they to other populations and environments? We are currently especially concerned about culturally disadvantaged groups and racial minorities.

Strictly, as Jensen mentions, there is no carry-over from within-population studies to

between-population conclusions.

I agree that it is foolish to deny the possibility of significant genetic differences between races. Since races are characterized by different gene frequencies, there is no reason to think ©

that genes for behavioral traits are different in this regard. But this is not to say that the magnitude and direction of genetic racial differences are predictable.

; It is clear, I think, that a high heritability of intelligence in the white population would not, even if there were similar evidence in the black population, tell us that the differences between

the groups are genetic. No matter how high the heritability (unless it is one) there is no

assurance that a sufficiently great environmental difference does not account for the differences ©

in the two means, especially when one considers that the environmental factors may differ qualitatively in the two groups. ... It can be argued that being white or being black in our society changes one or more aspects of the environment so importantly as to account for the difference. For example, the argument that the American Indians score higher than Negroes on IQ tests—despite being lower

on certain socio-economic scales—can and will be dismissed on the same gounds: some

environmental variable associated with being black is not included in the environmental ratio. This criticism totally vitiates Jensen’s argument on black-white differences, for if it rests on a faulty set of comparisons no further counter-argument is needed. Nonetheless, there are other weak points in Jensen’s analysis worth noting. For example, the statistical manipulation which is employed to demonstrate the heritability of IQ differences between whites and blacks can be turned around to indicate that all differences are due to environmental effects. This has been

demonstrated by Sitgreaves (1969). |

32 |

This does not invalidate Jensen’s argument, but merely demonstrates that statistical theory alone cannot be used to buttress a theory of heritability. According to Sitgreaves (1969:3-5):

Professor Jensen believes that the best estimate of A, , derived from available data on

correlations between IQ scores of individuals with varying degrees of family relationships, including monozygotic twins reared apart, is about .80. However, as he points out, this is a sample value, based on calculations made from relatively small samples, in which the sampled ~ environments may not be represented in the same proportions as in the norming population, so

1|

Alland | INTELLIGENCE IN BLACK AND WHITE . 33 that the true value of H, is still in doubt. For example, in the Burt study of 53 identical twins reared apart, 29 of the natural parents and 32 of the foster parents were in semi-skilled or unskilled occupations, so that the occupational distributions of the parents are skewed, and

concentrated on one end of the occupational scale. |

Suppose, however, for the time being, that we accept the value of .80 as the value of Hf, .In

such a case some simple calculations give us )

G, = 100, Vq_ = -80 + (200) = 160, .

BE. = 0, VE, = .20 + (200)= 40 ;

SDE = 6.3. l

[where 200 = the variance] for the norming population. That is, a white population with a normal distirbution of environments. The corresponding standard deviations are S.D.q# =12.6,

Now let us consider a second population, namely a Southern Negro population. A number of studies have shown ... that for this group the mean IQ score is about 85 while the standard deviation is about 12.6. That is, we have P, =G, + #, =85,and Vp2= Ya, + VE, = 160. The values of the component means and variances depend upon the joint probability distribution of G and E in the Negro population. The hypothesis proposed for study by Professor Jensen states that the observed differences in the means and variances of the IQ scores reflect differences in the distribution of :the genetic

component in the two groups. The alternative hypothesis proposed here considers that_this distribution is the same in both the white and Negro populations with the result that G, =G, = 100, and Ya, = YG, = 160. Now, if we assume that for Negroes in the South, the totality of the environmental effects, represented by the component £, is to depress each IQ score by a fixed amount, and we assume that this amount is 15 points, we have E, =- 15, VE, = 0, so that P, = G, + E, = 100-15 = 85 and Vp, = YG, + VE, = 160+ 0= 160. Thus we obtain from the model exactly the values that have been observed. : Jensen’s estimates of heritability are based upon studies of twins, particularly those studies concerned with twins who were reared apart. In addition to taking the highest heritability figure available, Jensen relies upon mean. values rather than the observed range of IQ differences. While the mean values are relatively low, the range may, in fact, be wide. Since the available sample is quite small, a wide range of variation might be significant. In addition, as Sitgreaves points out, the

environments of many of the twins who were reared apart are not significantly different. According to Gottesman (1968:27-28), . Another way to gain perspective about the meaning of a 10 or 20 IQ point difference is to look at the data on within-pair differences in intelligence for identical (MZ) and fraternal (DZ) twins ... Even though the gene pools do not differ and even though each of the two groups has been raised under more or less the same regime, the mean difference amounts to 6 IQ points for the sample of fifty pairs studied by Newman et al. (1937). The range of within-pair differences was 0 to 20 points. Thus, even when gene pools are known to be matched, appreciable differences

differences. . ) :

in mean IQ can be observed that could only have been associated with environmental

A better appreciation of the influence of the environment on IQ can be gained from looking at the two unique samples of thoroughly described identical twins who have been reared apart and thus in discriminably different environments ... the average intrapair difference on the

Binet was 8 IQ points. The range of differences was 1 to 24 points. A very similar picture is . given in a remarkably large sample of thirty-eight pairs of identical twins reared apart and studied by Shields (1962). When the tests used in this larger study are converted into IQ point

equivalents (Shields and Gottesman 1965) the average interpair difference for the identicals is 14 points on a verbal IQ test and 10 points on a nonverbal test ... At least 25 percent of the sample of identicals reared apart had within-pair IQ point differences exceeding 16 points on at

least one of the tests... : It is obvious from looking at the data on identical twins that individuals with exactly the

same genetic constitution can differ widely on the phenotypic trait we measure with IQ tests and label intelligence. The differences observed so far between whites and Negroes can hardly

. be accepted as sufficient evidence that with respect to intelligence, the Negro American is genetically less endowed.

34 RACE AND INTELLIGENCE [AS8 It is apparent from his article that Jensen accepts race as a valid biological division (at least in the statistical sense) lying between the species and the population; but there is growing evidence from statistical and taxonomic studies of within-species variation that race is an inappropriate concept in biology, particularly in regard to man. This suggestion, first made in anthropology by Livingstone (1962) and Brace (1964), has been borne out by Hiernaux (1968) in his most recent

book, La diversiteé humaine en Afrique subsaharienne. Hiernaux argues that far from being uniform, African populations reveal a diversity comparable to that of populations from other regions of the world. He presents us with a scale of interpopulational variability in Africa, isolating 33 characters, when he applies this scale to a world sample, 23 characters show a rapport of 50% or more. Thus he demonstrates that African populations are no more homogeneous than any other

group of human populations defined geographically. In addition, he shows that groupings based upon ethnic classification in Africa do not correspond with those few genetic groupings above the. | level of the individual population that he has been able to abstract from the data. In fact, of 101 7 ethnic groups open to cluster analysis, Hiernaux has been able to establish only 25 constellations, of which 23 comprised only two populations each; 49 populations escaped all grouping. This evidence of ethnic and genetic variation in Africa is particularly germane to the discussion in that while Jensen is talking about the American Negro, the genes he is talking about (or, more correctly, a good percentage of them) come from Africa. Gottesman (1968), in a book edited by Jensen and others, also stresses this diversity. He points out (1968:17), citing a study of William Pollitzer, that the regions of origin of the slaves imported to Charleston between 1733 and 1807—Senegambia 20%, Windward Coast 23%, Gold Coast 13%, Whydah-Benin-Calabar 4%, and Angola 23%—represent an area more than 1000 miles long and 600 miles wide. He goes on to argue (1968:20) that it is a tremendous oversimplification to speak of a single black or white population in the United States:

The variation[s] observed in the studies reviewed in this section are probably valid and reflect the genetic heterogeneity of Negro Americans living different geographical and social distances away from their white neighbors. Such heterogeneity prevents us from speaking validly of an ‘average Negro American” with x percentage of white genes.

To attribute a genetic basis to intelligence differences between groups is, to say the least, premature. If such group differences were, in fact, genetic, we might expect to find significant — differences between populations within each “‘racial’’ group rather than the consistency cited by Jensen. That consistency points to the effects of environment upon the performance of American

Negroes as a sociologically defined group. As Gottesman (1968:46) has put it, | At the present time Negro and white differences in general intelligence in the United States appear to be primarily associated with differences in environmental advantages.

In sum, genetic studies of black vs. white intelligence (whatever that is) which are based upon undifferentiated United States samples are naive in the extreme because they do not consider distributions of genetic variation in Africa or in the United States. Jensen finds an overall intelligence deficit of 15 percentage points among American Negroes. He

is willing to attribute about half of this difference to environment. The other half is assumed to reflect genetic factors. Yet he states (1969:100). In addition to these factors, something else operates to boost scores five to ten points from first to second test, provided the first test is really the first. When I worked in a psychological clinic, I had to give individual intelligence tests to a variety of children, a good many of whom came _ from an impoverished background. Usually I felt these children were really brighter than their IQ would indicate. They often appeared inhibited in their responsivenss in the testing situation on their first visit to my office, and when this was the case I usually had them come in on two to four different days for half-hour sessions with me in a “‘play therapy’’ room, in which we did nothing more than get better acquainted by playing ball, using finger paints, drawing on the

blackboard, making things out of clay, and so forth. As soon as the child seemed to be completely at home in this setting, I would retest him on a parallel form of the Stanford-Binet. A boost in IQ of 8 to 10 points or so was the rule; it rarely failed, but neither was the gain very

often much above this. So I am inclined to doubt that IQ gains up to this amount in young disadvantaged children have much of anything to do with changes in ability. They are largely a | result simply of getting a more accurate IQ by testing under more optimal conditions ... I

Alland ] INTELLIGENCE IN BLACK AND WHITE | 35 would put very little confidence in a single test score, especially if the child is from a poor

background and of a different race of the examiner. : Is Jensen not aware that these conditions are not met by the majority of studies he cites, particularly those drawn together by Shuey (1966)? If the deficit he notes is consistent in disadvantaged children, then all the IQ differences noted between blacks and whites in the United | States may be subsumed under a combination of testing errors and environmental effects.

For those of us who have done fieldwork in Africa, there can be little doubt that behavioral and cognitive patterns in the black community in the United States bear little resemblance to those in ‘Africa, and that, in fact, such patterns are highly variable in Africa and dependent more upon

cultural differences than shared gene pools. Hiernaux’s data on the lack of correspondence between ethnic and genetic identity is of direct relevance here. Anyone familiar with West Africa will know that the common folk myths about the behavior of

blacks are patently false. Let us, for example, take musicality. In the Abron village I studied, which had a population of 115 individuals, there were only three skilled drummers, andI was one _ of them—and this in spite of the fact that music, particularly percussion, is a major element in most West African cultures. On the other hand, I found a rather wide distribution of verbal skills which involved subtle plays on words and intense metaphorization. Such a process must involve a

high degree of what Jensen refers to as “conceptual learning.’ Great emphasis is placed on oratorical ability, which involves a high degree of creativity, and new combinations which make sense within the context of the local culture are constantly sought for and reinforced through the

_ wide appreciation which they receive. Skill with words is so important to the Abron that a hereditary chief who lacked this talent was subject to ridicule by this own villagers. In this society, at least, a chief has to earn respect, and such respect comes from the rendering of wise decisions

couched in elegant language. .

While I have done no statistical study of the distribution of this trait among the people of Abron (I was concerned with other problems at the time), two facts were obvious to me: (1) Oratorical ability was fairly widespread in the populations with whom I had contact, and was highly rewarded (much more so than musicality), particularly in older people. (2) The trait was apparently learned through practice, although there appeared to be variation which may have been due to individual genetic differences in some aspect of intellectual capacity. The main thrust of Jensen’s paper, which has been somewhat buried by popular accounts, is actually that there is a wide diversity of mental abilities in men and that there is a need to develop diversified educational programs. With this I concur heartily, and it is unfortunate that Jensen’s

paper has influenced some to attack this very point. To adopt the genetic hypothesis without recognizing that genetic potential is manifested only in relation to environmental experience is to ignore all we have learned about the interaction of heredity and environment. I should like to think that Jensen is himself committed to further experimentation in education. It is unfortuante

that he has published what can only be taken as a premature analysis of the genetic effect on intergroup differences. Jensen has taken a fairly safe hypothesis—that intelligence is heritable—and

forced it to carry the burden of a second argument for which there is still little evidence: that black and white performance on intelligence tests is determined primarily by genes. Recently, a reviewer of one of my manuscripts suggested that it was “baloney’’ to demand that the scientist take full responsibility for his published findings, and that when they are misused he should take it upon himself to vigorously counteract such misuse. Such an idea horrifies me. I am

not speaking for censorship, but only for good judgment. In the case of race and intelligence it . seems to me that the basic tenet of scientific research, the null hypothesis, must stand. We must begin with the assumption that there is no connection between race and intelligence, at least so far

as genetics is concerned. Until this assumption has been disproved, it remains our basis of operation.

In my opinion, none of the hundreds of studies which have attempted to disprove the null hypothesis concerning race has been successful. This is important for all of us, and there is no need for us to be defensive on this issue. We must constantly remember that it is not our task to prove that the races are equal. We are, unfortunately, faced over and over again with hasty attempts to * prove otherwise. These must be met with a rigorous analysis, for this is the true task of science,

36 RACE AND INTELLIGENCE [AS8 whether we are discussing some new theory of planetary motion or biological differences between men.

REFERENCES CITED Cohen, R.

1969 Conceptual Styles, Culture Conflict, and Nonverbal Tests of Intelligence. American

Anthropologist 71:828-856. Crow, J. F.

1969 Genetic Theories and Influences: Comments on the Value of Diversity. Harvard

Educational Review 39:301-309. Gottesman, I. I. 1968 Biogenetics of Race and Class. Jn Social Class, Race, and Psychological Development, pp: 7-51. M. Deutsch, Irwin Katz, and Arthur R. Jensen, eds. New York: Holt, Rinehart and

Winston. Hiernaux, J.

1968 La Diversite Humaine en Afrique Subsaharienne. Brussels: Editions de l’Institut de

Sociologie Universite Libre de Bruxelles. Jensen, A. R. 1969 How Much Can We Boost IQ and Scholastic Achievement? Harvard Educational Review 39:1-1238.

Shuey, A. M. 1966 The Testing of Negro Intelligence, Second edition. New York: Social Science Press. Sitgreaves, R.

1969 Comments on the Jensen Report. Paper read at the meetings of the National Academy of Education, University of California, Los Angeles, October 11.

Whose Is the Failure? VERA P. JOHN DEPARTMENT OF PSYCHOLOGY YESHIVA UNIVERSITY

OF THE MANY recent attempts to blame the victims of public education for their misfortune,

none is as sweeping, as detailed, as pessimistic, as Jensen’s (1969) genetic argument, which concludes: ‘Social class and racial variations in intelligence cannot be accounted for by differences in the environment.’’ Jensen’s approach, though persuasive, is based upon many fallacies.

The most important of these is the notion that intelligence is a measurable entity which is accurately reflected by scores on existing tests of intelligence. Intelligence tests were developed in |

Europe in the early years of this century, when there was as yet little research extant on the development of thought and problem-solving in children. The test items were chosen solely on the basis of empirical considerations, standardized on a culturally homogeneous group of children, and

validated against the judgment of teachers. Alfred Binet, the inventor of IQ tests, realized the limitations of his approach; he strongly opposed any concept of fixed intelligence (Schwebel 1968).

The format of intelligence tests has remained unchanged since their invention. As Schwebel (1968) so aptly argues, American psychology has achieved little to date in probing the processes of learning and teaching; rather, it has devoted itself to psychometrics, that is, the quantification of learning ability. This is a strange example of the codification of ignorance. In the past few decades, some work has been done in the exploration of human intelligence and thought, for example, by Piaget, Guilford, Bruner, and Vygotsky. But the collection of diverse items referred to as the IQ

test does not reflect this newer work. The notions of the structure of intelligence and the exploration of the processes of thought (both convergent and divergent, as illustrated in reasoning and creativity) are absent from these instruments. Psychologists have barely begun to free themselves from the classical, philosophical tradition, : on the one hand, and the anticognition behaviorist position, on the other. We are just starting to think about thinking. To represent the state of knowledge in the field as precise enough for the

quantification of intelligence, which is but one form of thought, is to propose to measure something we cannot even define. To use such a tool as an accurate indicator of race differences, as Jensen does, is absurd. A second fallacy is the idea of equal opportunities for the tested. There is a sizeable literature in psychology on test-readiness, motivation, and linguistic variables as these affect the performance of the tested. Examples illustrating the unequal conditions confronting children drawn from diverse ethnic groups abound. Zintz (1969), for instance, discusses the impact of cultural variables on test behavior among children in the Southwest. Labov (1969) has addressed himself to the

influence of being a speaker of Negro nonstandard English upon test performance and has examined some sociolinguistic variables involved in the testing situation. In a devastating critique

of Jensen’s work, Labov points out that an experienced black interviewer, himself a child of Harlem, gets only monosyllabic, defensive answers when questioning Harlem boys he Knows in an

individual setting, but that these same children, in discussion with their peers, reveal intricate verbal reasoning skills in talking about their experiences. The testing setting is perceived by most of these children as an accurate sample of a world that has traditionally excluded and often physically and/or mentally injured them. Rejected by, or rebelling against, that world, they show 37

38 | RACE AND INTELLIGENCE [AS8

among Cherokee. , : defensiveness and hostile passivity in such a setting and are unable or unwilling to demonstrate their knowledge or ways of thinking. Dumont and Wax (1969) have described a similar situation

The oft-mentioned cumulative deficit of minority children—a relative deterioration in

performance in terms of national norms as children get older—can best be explained in light of the kind of process these workers have described. Once minority children are old enough to build a society of their peers, they will rely upon that group and withdraw increasingly from the school,

though physically they may still remain within it.

Another fallacy is Jensen’s view of the nature of the biological environment and its effect upon intelligence. Jensen argues that since a higher birth rate exists among the poorest and least able segments of the Negro community than among the “educationally and occupationally most able segment of the Negro population” (which has a lower birth rate than its white counterpart) and among whites as a whole, intelligence differences between whites and blacks will increase over — time. This type of argument has been made many times in the past, and forecasts based upon such arguments have been shown to be incorrect. In the last quarter of a century, however, the environment of lower-middle-class Americans, who are predominantly white, has greatly improved, through preventive medicine, better prenatal care, improved nutrition, better housing, and other benefits. Such benefits have not extended to — the non-white residents of ghettos and reservations. The lack of adequate nutrition, during both the prenatal and the postnatal period, is an important cause of what has been called minor brain damage. Impairment in hearing, vision, and attention span constitute environmentally caused, but biologically mediated, limitations of the growing organism that are important to learning (Birch and Gussow, 1970). Further, the effect of frequent illness on learning and school achievement is

clear. |

The intervention strategies based upon these two views of the relationship between body and mind—Jensen’s, emphasizing gene pools, and Birch’s emphasizing environmental medicine—are very different. Jensen’s strategy excuses this society for its racist policies toward those it has exploited for up to four centuries and offers them a permanent spot on the lowest rungs of a caste system. Birch’s requires the reshuffling of the resources and benefits of this society; it insists on

economic justice as a prerequisite to educational improvements. co A fourth fallacy is the idea of “cultural deprivation’? rather than culturally patterned

differences in behavior. Anthropologists were aghast when the term “cultural deprivation”’ first became popular among the antipoverty warriors. The term was not accidental. It brought in its wake its own inevitable logic: a new form of the old concept of the “‘primitive.’’ According to Jensen, disadvantaged children have adequate abilities in associative learning but are deficient in abstract and verbal intelligence and should therefore be taught by associational methods. The most powerful refutation of this argument has come from a long-time collaborator and former student of Jensen’s, William Rohwer. Jensen and Rohwer interpret the meaning of the

studies of paired-associate learning on which they have at various times collaborated, in diametrically opposed fashion. In these studies, the child is presented with 20 word-pairs, such as “elephant and pillow,” or “pear and tree.” In subsequent trials, after hearing the first word in the pair, the child is supposed to recall the second. What the experimenters have found, says Rohwer

| (1969), is that if you teach a child to elaborate on a pair, to use the words to construct a meaningful sentence which includes some action, such as “‘the elephant is looking for his pillow,” his rate of performance increases sharply. Once this method of learning is acquired by children drawn from varied backgrounds, the investigators can find no class differences in performance, even though the children’s tested IQ differs. Jensen argues that this is due to the genetically more even distribution of associational learning. Rohwer, on the other hand, argues that this experiment stresses the importance of an active process of learning which includes not a rote principle, but a highly inventive process of imagining! Once children are taught effectively, as they were in this experiment, the contribution of previously acquired skills decreases. Our task, then, is to discover how to teach—an art with few practitioners and a science barely begun. (In a dissertation research project now being carried out by M. Rockewitz at Yeshiva University, Rohwer’s experiments are

John | WHOSE IS THE FAILURE? 39 being transformed into educational practice; inner-city children are being taught effective strategies of learning, and the results thus far are dramatic.) _ Other research concerning ethnic differences in learning styles has been conducted by Lesser, Fifer, and Clark (1965), who have shown that mental abilities are organized differently in children of varied groups: while Chinese children excel in spatial abilities, Jewish children perform best in

the verbal area. The high quality of performance of American Indian children on nonverbal tasks, such as the Goodenough Draw-A-Man Test, has repeatedly been shown by researchers over the past

30. years (see, for example, Dennis 1940). The relationship between culturally patterned experiences of socialization and the development of particular kinds of intellectual processes is of interest. However, current research methodologies often yield misleading results in this area. As Labov (1970) has argued, the assumption that black children-are nonverbal has been advanced and documented because of the total failure of white social scientists to familiarize themselves with

ghetto life and language. Consequently, testing methods militate against the gathering of an accurate sample of the verbal know-how of the ghetto child. If research in this area is to be continued, it is essential that it be initiated and implemented by social scientists from the ethnic group being studied. The comparative framework (pitting white

children against non-whites) fails to yield insight into the relationship between the cultural dynamics of a particular group and the development of cognition among the children of that sroup. Detailed and systematic observations are needed before such relationships can be effectively formulated. Anita Pfeiffer, a Navajo educator, has reported (personal communication 1969) that Navajo children mature early in skills required for reading. They are keen observers; they display high motor-perceptual coordination. What aspects of Navajo child-rearing may be causally linked to the rapid development of these skills? Through the work of Navajo social scientists, the answer

will eventually be forthcoming. In this connection, Jensen’s comparison of black and Indian children is particularly distressing. Quoting the Coleman report (1966) he argues that in spite of the greater poverty of the reservation child, he scores higher on IQ tests than his black peer. Jensen, being ignorant about Indian life, fails to note many relevant factors. For instance, many Indian children are not in school, so the

two populations are not equivalent in terms of representativeness. It is likely that fewer brain-damaged children survive in Indian than in black communities because of the lack of roads and hospital facilities on the reservations. Therefore, the contribution of low scores affects the two populations differently. The danger of such comparisons need not be documented further. Testing diverse groups with the same instrument without considering the specific conditions under which they live, learn,-think, and believe cannot lead to anything but fallacious generalizations.

To a psychologist, the most shocking aspect of Jensen’s approach (though one that is unfortunately to some extent inevitable) is his view of human intellect. | much prefer the alternative view offered by Sartre, who speaks of man as a creature who makes his being by launching himself toward the future. In such a view, man creates himself through his activity; and during periods of great social ferment, millions of men seize opportunities to change themselves by changing their relationship to others and to nature. The basically ahistorical view of many social scientists leads to a comparative, quantitative, and static understanding of the intellect of man. A simple example of this ahistorical fallacy was illustrated in a recent article on the American schools. Throughout its history, American schools have failed to educate, effectively, about 50% of their students, (Greer 1969). In the past, 50% of the children who were failing to gain from the public schools were predominantly white, from Italian and Irish immigrant families and even some Eastern European Jewish families. These children went on to become part of the very same gene pool which, according to Jensen, is thought to possess, on the average, 15 points more of abstract intelligence than blacks. Jensen points to the lack of substantial improvement in achievement by children participating in contemporary compensatory programs. Substantial numbers of children ' educated in these programs have profited from them, however. The problem is that these programs, like all other programs in American mass education, have failed to reach 100% of those exposed to them. - Why do we bother to try to refute Jensen? Because we social scientists, in our role as advisers to a country which cannot solve its social problems, have lost our ability to examine fundamental

40 | RACE AND INTELLIGENCE [AS8 premises. We support transitory policies and trends rather than base our thought upon a true and highly developed science—for it does not exist as yet. The varying moods of. the country give prominence to one or another half-baked theory buttressed with disparate facts. The year 1969 _ witnessed a wholesale and frightened attack on the Negro people, who had never been more vocal, never more competent in their demand for a total redistribution of power and resources. The best possible education, based on the little we do know, could be a powerful ally to black children in

their struggle. |

In this context, Jensen’s article furthers a racist doctrine. Some have argued about his

motivation. Jensen himself has defended himself against what he considers the emotional and

violent attacks upon him by some members of his own institution and by student activists elsewhere. Whatever forces have impelled him to present anew a genetic theory of racial — differences, the impact of his presentation has been to hurt the cause of non-white peoples. Can > we stand by? Silent? — |

REFERENCES CITED | ,

Birch, H. G., and J. D. Gussow | 1970 Disadvantaged Children: Health, Nutrition, and Social Failure. New York: Harcourt.

~ Dennis, Wayne 1940 The Hopi Child. New York: Wiley.

Dumont, R. V., and M. L. Wax

Greer, Colin .

1969 Cherokee School Society and the Intercultural Classroom. Human Organization 28 :217-226.

1969 Public Schools: Myth of the Melting Pot. Saturday Review, November 15, pp. 84-86+. Jensen, A. 1969 How Much Can We Boost IQ and Scholastic Achievement? Harvard Educational Review 39:1-124. Labov, W.

1969 The Logic of Nonstandard English. In Report of the Twentieth Annual Round Table Meeting on Linguistics and Language Studies, pp. 1-43. James E. Alatis, ed. Washington,

D.C.:G.S., Georgetown University Press. | , Lesser, G. Fifer, and D. H. Clark BS 1965 Mental Abilities of Children from Different Social-Class and Cultural Groups.

Rohwer, W. D., dr. Schwebel, M. | | Monographs of the Society for Research in Child Development 30(4).

1969 Language, Race, and School Success. Manuscript.

1968 Who Can Be Educated? New York: Grove Press. | :

Zintz, M. V.

1969 Education Across Cultures. New Mexico: Kendall, Hunt. |

The Influence of Conceptual Rule-Sets on Measures of Learning Ability ROSALIE COHEN DEPARTMENT OF SOCIOLOGY TEMPLE UNIVERSITY

IT HAS BEEN OBSERVED repeatedly that pupil performance on standardized measures of intelligence and achievement varies in certain systematic ways. Children from rural areas perform less well as a group than do urban children, those from poor families less well than those from

middle class families, girls less well in mathematics and physics than boys, and so forth. _ Explanations of these variations have ranged from minute physiological differences among children to broad-scale social influences on them. Most related investigations have focused on differential response to performance measures as they now exist; few have incorporated differential pupil

performance with analysis of school performance requirements within a common frame of reference. Many studies have questioned the representativeness of the content presented in test instruments to the experience backgrounds of children; few suggest that the tools of learning may be relatively unrelated to the content which is learned, and that they may be related separately from the content to the measurement of potential. Many studies have focused on selective behavior characteristics of poor achievers as sources of devaluation of them by teachers, making learning differentially effective; few have focused on systematic relationships between different behavior characteristics and different kinds of cognitive performance. Most studies have accepted the social organization of the school and the cognitive organization of learning materials as given; few have related them to each other, to variations in the organization of other settings in which learning takes place, nor to the occupational settings in which learning is used.

The body of research which is viewed here has explored all of the above minority stances. Within a single frame of reference, it has studied the requirements for performance in school in interaction with those brought to it by pupils; it has separated the transfer of information in the school from the transfer of cognitive tools to produce new information and from the transfer of

the norms of social organization; it has studied the relationships of school valued behavior characteristics as correlates of good and poor performance to different cognitive tools; and it has dealt with the relationships of varied styles of social and cognitive organization which are used by pupils to the social and cognitive requirements of the family, the school, and of different life tasks. This research views the modern-day school as a formally organized institution and its required rules of cognitive organization as Analytic. It suggests that, because of its pervasive formal focus, the transfer of information has been confounded with the transfer of cognitive skills and with

organizational norms, and that they are subsequently confounded in their translation into ) conventional measures of learning performance. As a result, the effectiveness of performance indicators as measures of learning ability among children, or as predictors of success in life tasks, has been only selectively appropriate. In order to obviate this problem, it has been necessary to separate for analytic purposes the functions of the school: (1) to transfer old knowledge (general information) from its functions, (2) to transfer cognitive skills to produce new knowledge, and (3) norms, or work rules for the organization of individuals around tasks. Learning “potential” per se cannot be measured with our present tools. Current measures of

.learning ability are performance measures; and inferences are drawn from placement on a performance range to placement on a similar range representing learning capacity. That is, learning

4]

42, | RACE AND INTELLIGENCE [AS8 performance measures are indicators of what has been learned as evidenced by what is used and/or produced in response to appropriate stimuli. Such measures are focused around what has been presented to be learned, either directly in the school—a setting which, despite variations, presents

relatively similar inputs in quality and quantity—or, indirectly, in supportive agencies and contexts. Such measures may thus be “‘normed”’ against what is commonly learned by other children of similar ages and school levels in a large general sample. It is the assumption of this approach that all individuals can learn any body of information which has been ordered logically, presented clearly, and reinforced regularly and appropriately in relation to their capacity to learn; and that they will use it or reproduce it in response to appropriate stimuli. This approach would not question the effect on performance of a multitude of social, cultural and interpersonal factors, nor the limited state of empirical evidence concerning the extent to which each may be involved. It tends rather to view these multitudinous factors as non-systematic (idiosyncratic to individual children). They are viewed, thus, as not within the purview of standardized assessment, and their manipulation as not within the legitimate functions of the school. Despite the commonality of this

point of view, however, the research which is reviewed here finds that, rather than being idiosyncratic, many learning related characteristics of individuals are systematic in. their interrelationships and they are systematically related to conventional measures of learning ability. They are, further, intricately interwoven with the legitimate functions of the school as with those

of other social arenas.

The progress of these studies has been systematic. Their early phases are reported as

background; the findings of later studies are reviewed together. This, then, is not a research paper in the sense that it reports the findings of a single study. It is rather an overview of many studies, both published and unpublished, placed within a single explanatory framework.! The purpose of this review is to begin to illuminate the different cognitive rule-sets which individuals use, and their relationships to conventional measures of learning performance. It also speculates on. how they may develop, and how they may be used. /

Four conceptual styles are described—two Polar or mutually incompatible rule-sets, the Relational and Analytic, and two conflict resolution rule-sets, the Flexible and Conflict-concrete styles. It is noted in advance that as they are described, these styles have the character of “‘pure styles.” They appear among many individuals in the form in which they are presented; many other individuals, however, demonstrate the use of mixed patterns or combinations of styles which are not commented upon. Most recent investigations have thus been directed toward the manner in which rules are linked and rule-sets transformed. In general, most of the research which is reviewed

here has focused on the identification of internal consistency of patterns, on the related characteristics of individuals which may follow from using each of the response styles, and to their relationships to existing measures of learning performance and to life tasks. With the exception of

early studies, it has not been concerned with the development of populations values. It is understood that, once alternative performance measures have been developed along with a sense of their reference to other abilities, the development of population values will follow. EARLY STUDIES

) Early research was restricted to a conventional approach to the identification of learning problems.’ That is, its efforts were to identify generic performance requirements in the school, to isolate specific performance failures among pupils, to attempt to identify sources of these failures, and to suggest remedies. A first task was to identify generic requirements for learning performance

which testing instruments make. Through a content analysis of the ten most commonly used standardized tests of intelligence and achievement along with a sampling of researchers who develop and revise these tests, three types of requirements emerged. They were the demonstration

of incremental development on successive age and grade levels of: (1) breadth and depth of informational content, (2) the ability to abstract analytically, and (3) the ability to extract salient

information from its embedding context (field articulation) (Cohen 1967). Compensatory education has been concerned primarily with the first of these requirements, i.e., helping children

Cohen ] CONCEPTUAL RULE-SETS AND LEARNING ABILITY | 48 with limited backgrounds to acquire more information about the world. Analytic abstraction and field articulation skills, the second and third requirements, have been assumed to be constants. Although these reflect only one mode of cognitive organization—Analytic—intelligence tests have traditionally been based on the assumption that the ability to use these skills in unfamiliar as well as familiar contexts is a measure of intelligence. Modes of cognitive organization other than those

required by these tests, however, are the subject of a substantial literature (see Cohen 1969a, Appendix A). This literature indicates that these skills are independent of intelligence and they are definable without reference to specific substantative content (Kagan, Moss, and Siegel 1963). Different conceptual styles have been observed to appear also among members of the same family, i.e., within a similar biological and social framework.

A second study analyzed and reconceptualized the relevant psychological and linguistic literature (Cohen 1969a). This procedure identified two polar response patterns following on two dominant modes of conceptual organization. The mode of selecting and organizing sense data required by the school is “‘analytic’’; that which has characterized a deviant approach to cognitive functioning is called here “Relational,” although in the literature it is more commonly called ‘self-centered.’ Sociobehavioral correlates of the two dominant modes of conceptual organization cover a wide range of classic test behaviors (Cohen 1969a). A new observation emerged from this study. That is that many social and behavioral expectations of the school environment which may

act as informal performance criteria are those which are associated with the Analytic style. Expectations for progressively increasing attention and concentration spans around which lesson plans are designed and curricula organized, increasing objectivity on higher grade levels, increasing responsibility and self-discipline, regularity of attendance, respect for the time schedules and social

arrangements which define .teacher-pupil relationships also appear as performance criteria. Assessment of pupil performance in these related areas are commonly stated explicitly on report cards in the elementary school. In the later grades, they are assembled under the broad assessment grade given to “citizenship.” A system of progressively stringent negative sanctions attached to violations of the norms is part of the definition of school organization around learning tasks, and it contributes to the assessment of relative pupil performance. Most educators would agree that “sood work habits’ enhance the learning process as they are frequently associated with good learning performance among pupils. Because certain social and psychological characteristics are

systematically associated with different styles of cognitive behavior, however, such related performance characteristics (also used as supportive assessment criteria) may also be viewed as

derivatives of the Analytic style, rather than as separate phenomena (Cohen 1969a). One additional observation of note emerged from this study: among children who enter school with Relational patterns of response, movement has been observed during the first few years in school from Relational responses to Analytic ones (Kagan, Moss, and Siegel 1963). When the school is viewed as an Analytically oriented environment, this observations suggests that a dominant pattern of cognitive organization can be modified. A third study attempted to identify sources of these two styles in influences on early learning patterns (Cohen 1967). A year was spent in participant observation and in progressively structured

study in order to do so. More closely associated with them than with any other source of systematic variation was the style of family and peer group organization in which children were socialized and in which they continued to participate. These studies found that Analytic children had both originated in and continued to function in formally organized primary groups—those in

which group functions had been conceptually extracted, formally abstracted and assigned to } status-roles within the group. Children whose preferred mode of cognitive organization was Relational had been socialized in and continued to function in “shared function’ primary groups—those in which critical functions of the group were widely shared or indiscriminantly performed, resulting in a fluid and constantly shifting distribution of functions. Although the frequency of both primary group styles and their associated conceptual styles differed in various social groups, relationships between style of primary group organization were reliable and they cut across all common sources of demographic variation. _ A series of studies followed relating styles of primary group organization, styles of cognitive

organization and differential school performance. Instrumentation was multidimensional,

4A | RACE AND INTELLIGENCE [AS8 TABLE 1. MODE OF ABSTRACTION AND FIELD ARTICULATION RULES OF ANALYTIC AND RELATIONAL RULE-SETS

Analytic Relational | Mode of Abstraction Rules Stimulus-centered Self-centered Parts-specific Global Seek formal properties Seek Descriptive properties

Field Articulation Rules Extract infinitely Embed infinitely TABLE 2. MODE OF ABSTRACTION AND FIELD ARTICULATION RULES oe OF ANALYTIC, FLEXIBLE, CONFLICT-CONCRETE AND RELATIONAL RULE-SETS

Mode of Abstraction Field Articulation |

Analytic Stimulus centered Extract infinitely Parts specific

Find common hidden attributes (generalizable) | : ):

' Properties ... with formal meaning .

Flexible Stimulus centered Embed selectively Parts specific Unique properties (with formal meanings)

Conflict-concrete Stimulus centered Extract (alienation) )

Global Extract, except forby those concepts Descriptive held together physical, Large Categories biological, or hereditary orgins, by definable spatial boundaries,

or by era or time |

Descriptive |

Relational Self-centered Embed infinitely Global

Unique properties

TABLE 8. PRIMARY GROUP STYLES, CONCEPTUAL STYLES

. AND MODALITIES OF SELF IDENTIFICATION

Primary Group Style Conceptual Style Cognitive Skills Modalities of Self

Identification and

Referent “‘Others’’*

Formal persistently a Analytic |Field Analytic Status Independent ‘‘Generalized others” Formal and shared-function Flexible Analytic Individualized Action Field Dependent “Significant others”’

Within-group conflict Conflict-concrete Descriptive Concrete Field Independent No ‘‘others”’ (exceptions) (exceptions) Shared-function persistently Relational Descriptive Transcendental

, Field Dependent ‘‘[deal others”’

*Kuhn and McPartland 1954; Garretson 1961

Cohen ] CONCEPTUAL RULE-SETS AND LEARNING ABILITY 45 combining the methods and techniques of psychological, linguistic and sociological inquiry. These studies found: (1) styles of primary group organization and styles of cognitive organization were so - closely linked that mixed and conflicting participation were associated with mixed and conflicting

methods of cognitive organization, identifying two additional styles—the Flexible and Conflict-concrete styles; (2) users of the several methods of cognitive organization were differentially distributed in relation to measures of tested achievement, with Relational and . Analytic pupils dominating the poor and good achievement categories respectively, and users of

the mixed styles dominating the middle ranges; and (3) preferred use of a style of cognitive organization was analytically separable from the breadth and depth of a pupil’s information repertoire, and methods of relating to the varied dimensions of the test battery were so persistent that it appeared that they were following subliminal, internally consistent and well-integrated sets of rules.

Conceptual styles emerged, thus, as rule-sets for the selection and organization of sense data, | and for its transfer to new contexts, separate from the nature of the context in which they are used. The construct, ‘conceptual styles,’’ includes two kinds of cognitive rules: (1) mode of abstraction rules, and (2) field articulation rules. As they were formulated in early research, the Analytic and Relational rule-sets appear in Table 1. -More definitive rules of Analytic and Relational conceptual styles as well as those of two combination styles, the Flexible and Conflict-concrete styles, are assembled in Table 2, and relationships among style of primary group organization, conceptual style and self concept appear in Table 3.

| Use of the four styles appears to have a pervasive influence on the behavior characteristics of their users. Because of the kind of rules which make up the rule-sets, they can be seen to

determine the objectivity-subjectivity of their users (one’s perceptual distance from his environment), and to control the possible and permissible relationships among individuals, between individuals and objects, and among objects. Links among behavior patterns of individuals using the four styles are drawn in the following

sections. They describe individuals’ methods of primary group organization, their methods of

responding to tests of conceptual style, their language characteristics, and their mode of identifying “self” and “referent others.” A less structured portion of the review shows evidence of links between the use of the four styles and occupational self-selection. THE POLAR STYLES—ANALYTIC AND RELATIONAL—

THEIR MUTUAL INCOMPATIBILITY | In early studies, the Relational pattern was of interest because it appeared frequently among many urban children who achieved poorly in school. It is compared here with the Analytic pattern (that required by the school) because of their mutual incompatibility. A discussion of the two

styles follows. , , The Analytic Style |

The Analytic conceptual style consists of the rules: be (1) stimulus centered, (2) parts-specific, (3) find non-obvious attributes, and (4) abstract common or generalizable attributes of a stimulus which have (5) formal (relatively stable and long-lasting) meanings, i.e., ignore the idiosyncratic. In

addition, extract them from their embedding contexts, name them and give them meaning in themselves. Individuals who follow these rules relate to such attributes of a stimulus as its parts, and their relative size, length, width, weight, assigned values, etc., separately; e.g., a large object need not be heavy or of high value. Frequently relationships must be those distributed on linear continua, so that objects may be formally compared in terms of relative possession of a common attribute. Relationships which are drawn among objects by such individuals are more commonly static and descriptive rather than functional or inferential, and they seldom involve process or motivation as a method of relating one to another. . Following Analytic rules persistently appears to have certain consequences in the development

of conditioned behavior and personality traits among their users. Stimulus centering and extraction promotes in individuals a perception of conceptual distance between the observer and

46 RACE ANDINTELLIGENCE [AS8 the observed; an objective attitude, a belief that everything takes place “‘out there’’ in the stimulus.’ Since the stimulus and its attributes are viewed as formal, i.e., long lasting and relatively

constant, the opportunity to study it, to ask questions, to find reasons and processes, to take directions and compare results is stimulated and developed. Following Analytic rules persistently also appears to result in the development among its users of progressively longer attention and concentration spans, greater perceptual vigilance, a reflective attitude and a relatively sedentary nature. The language style of the Analytic conceptual style users is Standard English of controlled elaboration. Its relevant characteristics are its dependence upon relatively long lasting and stable meanings of words and formal and stable rules of organization. Communications are intended. to

be understood in themselves, i.e., without dependence upon non-verbal cues or idiosyncratic contexts. The language use presents many alternatives for-the expression of generalization and _ categorization along many different dimensions; and its organization is so formal that individuals — can discriminate parts of speech in nonsense sentences with the assistance of few grammatical or — syntactic cues. Because words have formal meanings, subjective intent is communicated through the selective use of modifiers and in dependent phrases and clauses. When spoken, the Analytic usage is characterized by “hesitation phenomena’” —pauses for verbal planning—by controlled vocal modulation,® and by revisions of sentence organization to convey specific meaning. Analytic conceptual style users were found to have originated in formally organized peer and social groups (Cohen 1968a, 1968b). Group organization is viewed by its members to have come about in order to perform tasks most effectively and efficiently. The group, its task, its sub-tasks,

its members and their task relevant characteristics have been abstracted analytically, extracted, named and given meaning in themselves. As a result, members can conceptually manipulate their

own characteristics so as to become variously appropriate to multiple status-roles, and to reformulate the characteristics of a task or a group through the conceptual manipulation of the attributes of the task, the group or of its members. Formal group members function effectively when activities are pre-planned and time-scheduled. In brief, in similar fashion to the manner in which Analytic rules are used by individuals to respond to standardized stimuli and to select and organize their language, they analytically abstract and extract to have meanings in themselves, the individual, his task relevant characteristics, his groups and their functions and sub-functions. The formal characteristics of individuals and those of groups are, thus, viewed as manipulable and | reformulable for multiple reference group relevance. The process of change with population srowth in formally organized groups is commonly accomplished through the differentiation and specialization of functions rather than through segmentation; and differentiated functions in such groups may be expressed by odd numbers rather than even ones. This phenomenon is explained by participants as a rational mechanism to resolve conflict in an orderly fashion so that group tasks may be expeditiously accomplished. Analytic style users responses to the ‘““‘Who Am I?” fall into the Status category. They identify themselves by their status-roles, e.g., “fathers,” “sons,” “engineers,” “Sunday School teachers,”

etc.; ie., they view themselves as multiple status-role carriers. When forced to elaborate, they identify with the functions which they perform in these status-roles or refer themselves to the interaction network by which each is defined. Their referent “‘others’”’ are like the Meadian

“generalized others’ who function interdependently with them in a complex network of . interlocking positions. Users of the status modality express commitment to this network, to its highly differentiated division of labor and to the complex time schedules which give it its patterning. Because of the conceptual integration of the characteristics of individuals with status-role functions, status responses lead to specific and verifiable predictions about the talents,

abilities and behavior of its users. : The Relational Style

In contrast to Analytic rules, the Relational conceptual style uses the rules: be (1) self-centered,

(2) global, (3) find descriptive characteristics, and (4) identify the unique, i.e., ignore - commonalities. In addition, embed for meaning. Each relevant concept must have either spatial or personal relevance to the observer and a unique meaning depending upon an immediate context or |

Cohen | CONCEPTUAL RULE-SETS AND LEARNING ABILITY Al situation. Generalizations and linear notions are generally unused and devalued, and parts of a stimulus and its non-obvious attributes are not given names and appear to have no meaning in themselves. Relationships which are drawn are frequently functional and inferential and more commonly static than dependent upon process or motivation. Since emphasis is placed on the unique and the specific, the global and the discrete, on notions of difference rather than on variations on common themes, the search for mechanisms to form abstract generalizations is not stimulated. __ Persistent following of the Relational rule-set appears to have certain consequences in the

, development of conditioned behavior and personality traits. Global responses require little time for scanning; thus, immediacy of response is stimulated. The descriptive characteristics of a stimulus tend to call forth affective responses; and the self-centering rule produces a narrowing of perceived conceptual distance between the observer and the observed. The observed is perceived to

be placed so close to the individual that it obscures what lies beyond it, and so that the observer cannot escape responding to it. The individual also appears to view ‘“‘the field’’ as, itself, responding to him; i.e., although it may be completely objective and inanimate to others, because

it demands response, it is accorded a kind of life of its own. This tendency is reflected in the language use of Relational style users in the personification of the inanimate. As a result of the intimate responsive perceptual field which is stimulated by following this rule-set, its users develop interactional tools to examine the field and to manipulate it—particularly social-interactional tools. Consequently, Relational style users appear to be active, distractible, emotional, over-involved in

all activities and easily angered by minor frustrations. Immediacy of response to a constantly changing field does not result in the conditioned development of long or deep attention or concentration spans or perceptual vigilance, but rather the reverse. Associated observations are that Relational rule-set users who follow the rule “be global’’ appear to be Gestalt learners, and | their most common method of progression of ideas is by free association.’ When the lexicon of Relational style users was subjected to a semantic feature analysis, it was found to be distinctly different language use rather than a sub-standard use of Standard English. It

was found also to have characteristics in common with some non-English languages (Cohen, Fraenkel, and Brewer 1968). Among its more salient characteristics are: it uses descriptive abstraction for word selection, words must be embedded in specific time-bound contexts for meaning; and it presents few mechanisms for the expression of generalizations or for the elaboration of meaning. Instead, such categorical combinations as the actor and the act, the cause and the effect, the place and what happened there are used to express in a single word or phrase what is usually elaborated in dependent clauses and phrases. In this language use there are few synonyms (equivalency ranges are wide)-and the selective use of available concepts for expression is dependent upon the unique context of the occurrence (contextual embedding) upon many interactional characteristics of the communicants, e.g., their ages,,sexes, and authority or other social relationships to each other (social standing), on the time and place in which they are spoken, and on inflection, muscular movements and other non-verbal cues (situational embedding). Formal

use of English is difficult for the Relational user to understand because it lacks necessary non-verbal cues; and reading presents an additional problem because the relationship of the

and inflection.

communicants (the writer and the reader) is not made clear. The spoken language of Relational style users is characterized by fluency,® by the use of strong and colorful expressions which are intended to express and to call forth emotion, and by a wide range of meaningful vocal intonation

Relational conceptual style users were found to have originated in “‘shared-function”’ families (those in which critical group functions are widely shared or indiscriminantly performed by all members without pre-patterning) and to participate repeatedly in shared-function peer and social | groups (for more details, see Cohen 1967). Shared function primary group organization is not centered around tasks, but around group mystique; it is responsive as a whole to chance changes in

its external and internal environment. Members function effectively when activities are immediately responsive to these changes, and their members appear to have developed a _conditioned sensitivity to hardly perceptible variations of mood and tone in other individuals and in their surroundings. They function poorly when time scheduled pre-planned activities tend to

48 , RACE AND INTELLIGENCE [AS8 interfere with immediacy of response to such stimuli. In brief, in similar fashion to the manner in which Relational rules are used by individuals to respond to standardized stimuli and to select and organize their language, they view the group as global, descriptive, group-centered and unique, and _

members consider themselves to have meaning only when they are embedded in it. With population growth, such groups segment into “like” groups rather than specializing in function and thus they remain primary groups. They are most commonly composed of even numbers of

members rather than odd. |

Relational style users’ responses to the ‘“‘Who Am I?” follow the Transcendental mode. They

identify themselves as reflections of global, descriptive “‘essences’; they refer themselves to transcendental notions, e.g., “I am an amoeba in the universe,” “I am rock and water, fire and ice,” or to universalistic perfect ideals, e.g., “‘“a seeker of the Truth,” “‘an advocate of Justice.”’ They appear not to notice structure, either in the physical or social sense. Their responses show no

evidence that they perceive themselves to have analytically measurable or identifiable attributes which may be related to functional status-roles. Their referent “others” as it appears in their responses are also transcendental, ideal or universal (the cosmos or mankind). Transcendental responses are considered to be beyond consensual validation or verifiable communication; and, in the sense that they are task and status-role free, they lead to few predictions about the talents,

abilities or behavior of the respondents. |

Discussion |

The Analytic and Relational conceptual styles are viewed as Polar styles because, in the combination of their rules of abstraction and field articulation, they require not only different skills and approaches to the selection and organization of relevant information from a stimulus or situation, but mutually incompatible ones. “‘Find the common, ignore the unique,” and ‘‘Find the unique, ignore the common” are examples of mutually exclusive rules. “Extract, don’t embed,” and “Embed, don’t extract” are likewise opposing rules. Use of each of the two sets of rules is associated among individuals with a wide variety of equally incompatible socio-behavioral correlates which appear to have been conditioned by following the respective rule-sets rather than

to have developed independently as separate phenomena. . oe Each set of rules was found to have been used repeatedly in rules for response to standardized

stimuli, in rules for language use, and in the organization of the social groups in which early socialization has taken place and in those in which participation was repeated over time. For this reason, they are viewed as subliminal rule-sets which individuals appear to follow without being aware that they are following rules. In addition, both the rules of cognitive organization and the rules of social organization appear to have taken on a value component to their users, i.e., they not

only view these rules as being used, but indicate that they should be used. It is therefore — hypothesized that related cognitive and social rules may be learned proverbially, as norms, in individuals’ earliest forms of social participation through the mechanism of positive and negative sanctions on random behavior. As norms, they appear to function generally and subliminally, below the level of consciousness, supported by value notions. The extent to which these rule-sets have become internalized is reflected in their carriers’ self concept, and in the nature of its group . origins in their perception of referent “‘others.’’ Since the rule-sets and their derivative behaviors are mutually incompatible, each set of rules devalues the other. This conflict of rules produces social conflict interactionally, associated with mutually incomprehensible language use and mutually restricting views of reality. When both rule-sets are used by a single individual, cognitive

conflict or role conflict might be expected, which may be resolved by using one of the conflict-resolution styles which are described in later sections. Under appropriate conditions in both dimensions, inter-individual and intra-individual behaviors characteristic of conflict resolution phenomena have in fact been observed, both experimentally and in natural situations.

Since the school is defined as an Analytic environment, requiring Analytic approaches to cognitive organization as well as analytically derived rules of social organization around tasks, and

since in the case of the school both Analytic rules and norms are legitimated by institutional definition, Relational style users could be anticipated to perform poorly according to both sets of -

Cohen | CONCEPTUAL RULE-SETS AND LEARNING ABILITY AQ criteria, with effective communication between Relational style users and the school constrained by the mutual incompatibility of the rule-sets. This appears to be the case. In sensitivity groups, however, in which Relational rules are legitimated, the reverse process has been observed to take place, and Analytic rules and norms are devalued and suppressed. Analytic and Relational styles are differentially related to common demographic variation and to school performance in our samples. Pupils who use Analytic and Relational conceptual styles , dominate good and poor achievement categories respectively. However, polar deviations occur -only when the use of the conceptual styles is combined with content absorption and feedback. Conventional performance measures combine the assessment of content learned by pupils with their ability to use Analytic tools; indeed, many items are constructed with this purpose in mind. — By selecting discriminating items and sub-routines from the batteries, however, content feedback and the use of cognitive tools can be analytically separated. The use of Analytic skills is an important determinant of the relative performance of pupils, although many pupils who use non-Analytic styles demonstrate large information repertoires, and in many cases, when the size of the information repertoires is held constant, they achieve reasonably high scores on performance measures. Item-by-item analyses of the test performance of individuals indicates that Relational _ pupils achieve their scores largely through curriculum related content feedback, however, and they lose scores most frequently on items which require high levels of Analytic abstraction and multiple extraction. Since all children “know” things which they have no opportunity to demonstrate in

response to these instruments, the curriculum related content which forms the basis for the development of test items is probably the only exposure common to all children. It may be properly viewed, then, as a measure of the relative ability to learn. When size of information repertoires among pupils is controlled for, many of the differences which appeared on combined content-skill became non-significant. Since, in many cases, the two measures of learning ability (content absorption and use of Analytic skills) have been sharply divergent, we were forced to consider the possibility that the mutual incompatibility of Analytic and Relational rules may be responsible (for more details, see Cohen 1967). It is apparent that both the size of pupils’ information repertoires and their use of Analytic skills are used as measures of ability to learn what is presented in school, and that the use of Analytic skills is an important determinant of the relative performance of pupils. What is not apparent, however, is that relative content absorption and feedback appear to create the greatest

performance differentials among pupils. In a sample of children of the same age showe performance assessments ranging from “barely educable” to “gifted,” the greatest differences at polar ends of the performance continuum were in the size of their information repertoires, despite a full range of conceptual styles on all levels, and despite a differential distribution of conceptual styles on each level in relation to achievement. Further, deviant school related personality and social characteristics among pupils were more closely associated with the use of non-Analytic styles than with content (Cohen 1968c). For a school population in which multiple styles appear,

then, the relative size of pupils’ information repertoires may be more directly related to their learning ability than is their preferred method of cognitive organization. In addition, when large and small information repertoires are combined with the effective use of the Analytic and Relational conceptual styles, especially when the behavioral correlates of the two styles are considered as differential performance criteria, these combinations are found to be effective predictors of relationships between tested ability and school achievement, as well as explanatory

criteria for good, poor, under- and over-achievement (Cohen 1968c, Table VII). / Differences between content absorption and cognitive skills do not explain variance in performance due to a multitude of other factors, of course. Such an analysis does suggest, however, that conventional combined measures of learning ability are differentially useful for , pupils with varied conceptual styles. They confound the transfer of information (a measure of - learning ability) with the transfer of skills to produce a specific kind of new information (also a measure of ability, but one which is selectively appropriate). As general measures of learning ability they are further hindered by the use of less structured observations of pupil behavior such

- as attention spans, perceptual vigilance, and other rule-set associated personal and social characteristics. This review also suggests that when using conventional performance measures for a

50 RACE AND INTELLIGENCE [AS8. multi-style population, the size of a pupil’s information repertoire may be a better measure of his

ability to learn than is the use of either cognitive or normative skills. } | CONFLICT RESOLUTION STYLES—FLEXIBLE AND CONFLICT-CONCRETE |

The two polar styles, Analytic and Relational, were discussed first because of their mutual incompatibility. Combinations of Analytic and Relational rules in mode of abstraction and field articulation categories produce two additional styles—the Flexible and the Conflict-concrete styles. In each case, when rule-set combinations are used by individuals, they exhibit conflict resolution

characteristics. For this reason, they are viewed as methods of dealing with the mutual incompatibility of Analytic and Relational rules in different ways. In the Flexible style, individuals _ conceptually isolate and insulate the social contexts in which each set of rules is appropriate and- - _ embed them selectively. Such individuals can use both sets of rules within a middle range level of

abstraction, and can conform to each set of social rules as they are appropriate. A conflict in reality organization is apparent, however, in many instances where users vacillate when decisions or general statements are required,’ and’ they are seen by professional observers to be highly

suggestible and anxious in new situations, and they tend to depend more upon their social | expertise than upon their skills or abilities. In the Conflict-concrete style, the incompatibility of Analytic and Relational rules appears to produce a conflict in reality organization which limits its |

users to concrete, physically observable characteristics of a stimulus or situation and to a denotative use of language. This conflict is also dealt with through the grouping of concepts into extremely large categories, by defining these categories as of pre-determined or unchangeable origins (e.g., biologically determined) or as having occurred by chance and therefore irrelevant, e.g., the grouping of unrelated objects on a table or the groupings of unrelated individuals in a

common town or neighborhood. -

The Flexible Style The Flexible style is characterized by the Analytic rules: be stimulus centered, be parts specific, and find formal attributes; combined with the Relational rules: identify the unique as meaningful ~~

and embed (selectively). Relationships formed by Flexible style users are more commonly functional and inferential than descriptive, and they lack the static character of polar Analytic or Relational style users’ responses. Flexible style users also commonly impute process or motivation to perceived relationships even when the content: is impersonal; and concepts are often held together only because a story has been created in which each item figures (embedding). General statements are possible only with qualifications, or with the identification of the circumstances | under which each may hold. The conceptual distance between observer and observed varies, and perceptual distance apparently can be manipulated, either at will, or as appropriate to the context. Flexible style users either cannot, or do not, move to higher levels of abstraction using either Analytic or Relational rule-sets, and, although they appear to be able to move freely among formally organized and shared-function social groups, they either cannot, or do not commit

themselves to either set of norms. ,

Language use characteristics of Flexible style users have not yet been subjected to linguistic analysis. Semi-structured observations, however, find (1) it isa highly elaborated form of Standard

English, i.e., it presents frequent use of contextual qualifiers (linguistic elaborations); (2) the quantity of communication is considerably greater than that of the users of any of the other styles; (8) its content contains many exemplary independent stories with process or motivation themes rather than general statements; (4) it presents frequent rhetorical appeals to other communicants for agreement and support; and (5) when spoken, vocal intonation and inflection is more varied than that of Analytic users and less controlled by the formal interpretation of the utterance, but it is less meaningful than that of the Relational.

Social participation studies of Flexible style users find that they have had effective participation in both shared-function and formally organized primary groups. It is believed that the cognitive and social rule-sets of both kinds of groups have been learned as composites in the

Cohen | CONCEPTUAL RULE-SETS AND LEARNING ABILITY 51 different styles of social organization, and that they are viewed by their carriers as selectively appropriate to them. Flexible style users appear to be able to participate effectively in both kinds of social groups without becoming committed to the norms of either. In moving from one kind of group to the other, Flexible style users have been observed in fact to change their style of behavior and their method of communication appropriately and to manipulate appropriately the perceptual distance between themselves and their fields. It is of interest that some Flexible style users remain

stably flexible, while others identify with either the Analytic or the Relational over time, i.e.,

1967). |

when later groups are of the shared-function type and earlier groups of the formally organized type, individuals’ response patterns are more like the Relational, and they reflect commitment to the Relational style of social and cognitive organization (a reference group phenomenon). When later groups are of the formal type and earlier ones shared-function, the reverse is true (Cohen Flexible style users’ responses to the “Who Am I?” fall into the Individualized Action mode. _ They relate to specific kinds of social contexts, e.g., they are characterized by qualifiers describing the circumstances under which each statement will hold, and they describe unique traits, feelings

and life styles of the individual (e.g., “I am a good mixer,” ‘‘a happy person,” and likes and dislikes). They exhibit a characteristic “faction” component rather than commitment to a stable system of relationships. Users describe themselves as having a kind of individualistic ‘“‘style’’ which

leaves them free to behave in various ways in various situations without abandoning their styles and without committing themselves to group norms. In reaction to norms, they make conditions

~ and exceptions and translate to suit their purposes. Their referent ‘‘others’ have similar characteristics to their own, they are more like Sullivan’s “significant others” than the Meadian

“generalized others.” | The Conflict-Concrete Style

The Conflict-concrete style of conceptual organization combines the Analytic rules of abstraction: be stimulus-centered and find formal attributes; with the Relational rules: be global and descriptive. Field articulation rules are: extract, except for those combinations which are immutably held together by Nature or those which are grouped together in space or in time or era by chance. Relationships which are drawn by Conflict-concrete style users are commonly static, and descriptive or functional, and they seldom depend upon process or inference. Their categories are so global as to ignore not only relevant parts of a stimulus, but its entire meaning as well; e.g., when three objects are presented and the subject is asked to select two which are alike or which go together in some way, the nature of each object may be ignored along with its parts to select two - which occupy a similar position in space. Because of the stimulus-centering characteristic of this style along with the formation of large general categories, it is sometimes viewed as a low level of Analytic abstraction. Perceptual distance between the observer and the observed is great. -

Following Conflict-concrete rules persistently appears to have certain consequences in conditioned behavior characteristics and personality traits. Response to large general categories _ does not stimulate either deep concentration nor immediacy of response; and the assumptions of immutability or chance origins of the categories does not stimulate curiosity or the desire to search for subtle attributes, reasons or processes. Focus on the concrete which is stimulated by this rule-set may, however, result in the practical and mechanical skills associated with it among its users, as well as their characteristic resourcefulness about practical matters. Its users tend to be * comparatively slow moving and deliberate in their actions, “‘set” in uncomplicated patterns of daily existence and unconcerned about “getting ahead.”’ The differential language use of Conflict-concrete conceptual style users has not yet been subjected to systematic linguistic analysis. Semi-structured observations, however, include the following: (1) the characteristic low level of Analytic abstraction is apparent in the language use, i.e., items are grouped into extremely large categories (e.g., a familiar bird may be identified as “‘an animal”). Common categories are the classic animal, vegetable and mineral; people (a separate _ category with distinctive biological origins) young and old, large and small; home and other buildings; things which grow and those which do not, and so forth. Language organization is

52 ) RACE AND INTELLIGENCE [AS8

appears infrequently. | standard but non-elaborated. Its users appear to observers to be taciturn; their speech is frugal, and it lacks common variations in vocal inflection and expressions of emotion. Personalistic content

On early measures of formal shared-function primary group style, Conflict-concrete style users’ responses were mixed and conflicting. On further analysis of their groups, they were found to have originated in and to continue to participate in a unique style of formally organized family and to

belong to no other organized peer groups. Collaborative activity outside of the family is cooperative—‘‘groups”’ are aggregates of individuals who are engaged in some common task in such a way that each individual retains his individual autonomy. Such groups are viewed as having: no structure or life of their own. They exhibit little differentiation of functions, and they therefore

| reflect no concept of status-role around differentiated functions and no related task—relevant attributes of role incumbents. Families are persistently and formally organized around physically _ or biologically determined characteristics of individuals, forming biologically determined roles; and >

functions are seen to arise in individuals from the same sources as roles. Status-roles are thus viewed as non-manipulable, pervasive and generic. Because of their deterministic origins, status-roles do not depend upon skills or abilities in their carriers for definition, and they do not lend themselves to reformulation. Such families are characterized by a low division of labor, overlapping functions, and by global, undifferentiated notions of role. In brief, in similar fashion to the manner in which individuals respond to standardized stimuli, the low level of abstraction based on global descriptive characteristics of immutable or chance origin is reflected in the social organization of their families. This low division of labor based on global descriptive characteristics

determined by chance or of unchangeable origin is reflected in the organization of their other ‘““sroups’’ as well.

Conflict-concrete style users’ responses to the ““Who Am I?” fall into the Concrete modality.

They identify themselves by their physical characteristics, e.g., height, weight, coloring, distinguishing marks; by location in physical space, e.g., “I live at 1234 Main Street,” or “I live in the country”; or by gross category, e.g., “I am a person.” Differentiation is also low (they make fewer responses). The concrete observable characteristics by which users of this modality identify

themselves show no evidence that they perceive themselves to have a place in any kind of interactive social configuration. No referent “others”? appear commonly in responses in this _ modality. In the sense that they are task-status-role free, such responses lead to no useful predictions about their carrier’s talents and abilities. Because Conflict-concrete responses appear

frequently among patients in mental hospitals, patients in institutions for the physically handicapped and among college students who drop out, this modality was originally called an

“alienation category.’ Studies of pupils in Austria, however, where independence is institutionalized, added a new dimension which was incorporated systematically into later research. In addition to the concrete, physical and locational responses which are listed above, — those involving blood ties and the listing of specific activities were added. Such additional physical

and locational responses identify the respondent’s direct family links: “I am the son of my father,’ “‘the brother of my sister,” “‘the grandson of my grandfather’; as well as “I do’s” of a very specific nature, e.g., “I work in an office,” “sometimes I go to a movie.” It is of note that the global categories of identification are those which are either physically or biologically determined and non-task oriented; they do not lend themselves to reformulation around tasks. Responses in

: this modality are free of interactive content and if any referent “others” are suggested, they are those who are grouped together by chance or by pre-determination into similar global categories. The tendency to reformulate them for better task performance or for multiple-context relevance is thus effectively impeded. Discussion

Flexible and Conflict-concrete conceptual style users are characteristically distributed on learning performance measures in the middle ranges, with the mean of the Flexible style users consistently higher than that of the Conflict-concrete. These middle range styles appear to be unrelated to race and social class. However, girls are most commonly Flexible style users, and the

Cohen | CONCEPTUAL RULE-SETS AND LEARNING ABILITY 53 largest numbers of Flexible style users are consistently girls. This observation suggests that female sex-role learning is either clearly Flexible, or a selective mixture of Analytic and Relational. It is possible that the Flexible style may be supported and legitimated by positive social sanctions on

the resultant suggestibility and variability associated with the style among women. Conflict-concrete style users dominate the response patterns found in rural populations; they appear to be found more frequently in Southern urban complexes than in Northern ones; and in varied urban complexes it is commonly found among individuals in independent occupations such as postmen, small businessmen and master craftsmen. Rules of the four rule-sets are contrasted in Table 2 and associations among conceptual styles, primary group origins and self-concepts are charted for convenience in Table 3. Associations among conceptual styles, conventional definitions of deviance and self-selection into occupational roles follow in the final section. CONCEPTUAL STYLES, DEMOGRAPHIC VARIATION, DEVIANCE AND OCCUPATIONS

Styles of conceptual organization are differentially related to race, sex and nationality origins in

our samples as independent variables, as well as to differing styles of religious and political participation, to attitudes toward planning, ideal health services, education and to value realms as dependent variables. However, when style of primary group organization is controlled for, these

differences become non-significant. Where relationships among method of primary group organization, conceptual styles and conventional performance measures, are reliable, they cut across all common sources of demographic variation. Inter-instrument and longitudinal reliability is highest at the means of the distributions of the four styles; it is lowest at the breaking points between styles. Despite this variation, composites exceed chance. Shared-function primary group organization is more common in low income environments and more common among black urban populations than in white urban populations, while formal organization is more common in the middle class. Both black and white rural populations appear predominantly Conflict-concrete. Relational and Analytic conceptual styles are more common respectively in low income and middle class populations, and poor and good achievement levels follow. Because Flexible and Conflict-concrete styles bind their users to middle range levels of abstraction, their users are found predominantly on middle performance levels on standardized measures. When performance in subject areas which require higher levels of Analytic abstraction are compared with those which require lower ones, those individuals who do not reach high levels of Analytic abstraction because they use one of the mixed styles do less well on them. Girls, thus, who demonstrate the Flexible style most frequently, do less well in mathematics and physics than boys and less well in these subject areas than in others. Similarly, rural pupils among whom the Concrete-conflict style is more common appear to be middle range performers as a group. Within the identified ranges, size of information repertoires of pupils as they are reflected in item analysis on the instruments are predictive of performance levels. We are led to hypothesize, therefore, that a full range of learning potential may exist among each of the four conceptual style users, despite differential limitations in their ability to deal with items requiring high levels of Analytic abstraction and multiple

extraction. |

Distributions of the four styles also vary from school to school. When schools are served by communities in which the dominant family and social group is formal, the distributions of styles

are more heavily weighted toward the Analytic and Flexible styles and performance in its student . bodies is generally high. In those communities whose school populations originate largely in shared-function family and peer groups, distributions of the styles are reversed and the performance ranges are low. In predominantly rural areas, where families are of the type associated with Conflict-concrete style, this conceptual style predominates and performance levels are low

middle and the range is narrow. : . Except for the shared-function style, schools also tend to reflect the dominant method of social

organization of their pupil populations. In rural areas, schools reflect the globality of the _Conflict-concrete style. Curriculum organization, pupil differentiation and teacher preparation is comparatively undifferentiated and unspecialized, and self-contained classrooms are more common

54. RACE AND INTELLIGENCE [AS8 than in urban centers. In large cities, however, where the Analytic and Flexible styles are more common, schools exhibit a high degree of differentiation and specialization in their organization, not only in curriculum, but also in pupil differentiation, teacher preparation, in counseling services _ and in the career paths to which they provide inter-system links. It is in these large school districts, however, where pupil populations are most varied. For segments of the population in urban

centers, then (i.e., for the Analytic and Flexible children) school organization may act as a

selective reinforcer of formal cognitive skills. .

School norms associated with the Analytic style are also linked to general definitions of normative behavior. The extent to which the socio-behavioral correlates of the Relational style is

devalued outside of the school is apparent both in recognized definitions of deviance and in occupational settings. Students in a sample of correctional institutions for juveniles were found to | be predominantly Relational, and the next most common style was Conflict-concrete (Cohen | 1969b). And similar findings emerged from a study of a sample of the hard-core unemployed ina community agency which is concerned with teaching job-relevant social skills. It is of note that agency administrators observed that the norms of formal organizations are more important than are job relevant skills. They find that among those with similar school records and similar job skills,

only those who can conform to the norms of formal organization around tasks as they are reflected in school related performance measures are employed and continue to hold their jobs. These comments above concern only those positions which are held in organizations which, like the school, are formally organized around tasks. Such norms, as well as the cognitive skills which accompany them, are differentially important in settings in which time schedules, pace of work and interdependence are differentially required. In studies of professions in which both practice and research components are present, practitioners are consistently found to be less Analytic than are the research professionals.’ ° Among them, median Miller Analogy scores and their ranges were found to be middle range; and in many such schools, such instruments were abandoned as being inappropriate to the selection of successful professionals. In a report on selection criteria for Child Care workers, it was concluded that with school level completed and sex, race and social class held constant, intelligence as reflected in conventional measures was inversely related to success (Chambers 1969). Conceptual styles are also found in mutual selection criteria into occupational roles. Although |

Analytic styles are commonly found among businessmen, executives, researchers, and in certain academic fields, the Flexible is mosi characteristic of the ‘helping professions’’—social work, elementary school teaching and nursing; Conflict-concrete styles are found commonly among engineers, independent and small businessmen, and craftsmen; and the Relational among salesmen, creative writers, artists and musicians, sensitivity group leaders and counselors.! Although samples for these studies are accrétionary and the populations not yet sufficiently varied for reporting, the suggestion of developing trends is that mutual choice into certain occupations and into the social contexts in which work is done may well take place on the basis of non-Analytic as well as Analytic norms and cognitive skills.

There is evidence that styles may change under certain circumstances,” e.g., college students

who enter and leave with Analytic response characteristics are found to have reflected the Flexible pattern while in the process of their preparation, and that among those who enter with

and continue to manifest the Conflict-concrete style, many drop out without completing (Garretson, in press). Changes have been induced in a sample of retarded children (K.C. Board of Public Education) who, when exposed to social activities which widened their social horizons, changed from a dominant Conflict-concrete pattern of response to the Flexible (Hartley 1971). Conceptual styles also changed among mothers who participated in a school-support program which was directed toward helping them help their children)! And a sample of high school children in grades ten to twelve who were at least four years below grade level on achievement tests made sizable leaps both in achievement level and in tested intelligence as measured by standardized instruments in a period of a year when taught the rules of Analytic selection and organization, divorced from the social norms with which they are commonly associated.’ * We are led to suggest, therefore, that the Analytic items and sub-routines on standardized instrumentation |

Cohen ]| CONCEPTUAL RULESETS AND LEARNING ABILITY 55 are only selectively appropriate for the testing of learning capacity among children, as well as

selectively appropriate to occupational success. :

This review does not suggest, of course, that the failures of all poor performers on conventional learning performance measures are hampered by their conceptual style, nor does it suggest that for many pupils the multitude of individual and social factors which were mentioned earlier are not .

separately or in combination also involved. It does suggest, however, that because learning .. . performance measures depend in part upon the use of Analytic skills, and because these skills are analytically undifferentiated from content accretion and from related but less obvious assessment criteria such as length and depth of attention spans, such instruments are limited in their ability to separate learning potential from either the kind of level of cognitive skills attainable, or from many

behavioral characteristics of individuals which are taught and learned as norms of social participation. It suggests also that the styles are manipulable, and that few definitive statements can be made about the learning capacity of sizable numbers of children on the basis of these measures until such factors as conceptual styles are effectively controlled for. An additional observation arises from the less structured portion of the review, i.e., that all work is not done in formally organized settings nor does even all professional work require high Analytic skills. Some work requires mixes of the styles and in some occupations, high Analytic skills may be a distinct hindrance. Until effective links are empirically demonstrated between the varied requirements of occupational settings, both cognitive and normative, the salience of performance measures in their present form, either as measures of learning ability or as predictors of success in occupations, will

continue to be only loosely linked.

NOTES : ‘The sizes and nature of some of the samples are listed below. Some were surveys and some ' depth studies. The battery which is described in the text was used for some samples, and partial units and additional units were used in others. The relationship of conceptual styles to behavior patterns which are not described in the text has also been studied in still other researches which are not listed here. In some cases, samples are accretionary and final results will not be published until

their total N’s have been fulfilled. It was thought, however, that a partial list of N’s and

populations might be useful to the reader in his interpretation of the text. Samples: 66 purposively selected ninth grade pupils in a large Eastern urban school district, 450 ninth grade pupils (total ninth grade samples) from selected schools in the same school district; 3 populations of Midwest college students and administrators totaling 900; 8 school teachers; 835 members of a service group; 450 students (total student body) of a Midwest college; 3000 adults in a random household survey in a Midwestern city; 100 Peruvian college students; 80 former patients of a mental hospital; 300 riot area residents; 3 sororities totaling 180 college students; 1800 students, staff, faculty, and administration of 4 Midwestern colleges; 1700 freshmen of a Midwestern University; 222 patients in two psychiatric hospitals; 125 freshman medical students in a testing program for orientation week; 18 participants in a graduate institute on nutrition and health; 30 students in a rescue program for drop-outs and disadvantaged college students; 300 students in a special education project for retarded children aged 8 to 13; 450 students in 3 juvenile detention centers; 210 participants in a program to teach job relevant skills in a community agency; 600 students entering graduate programs in Social Work, Education and Psychology for 3 successive years; 125 college students in career assessment programs; 125 Austrian high school students in Mittelschule and

Hauptschule; 13 Black mothers in a program to learn to teach their children in an Eastern

Research Center. The instruments were also adapted for use among small children in India by Dr. Eva Keockheis; for use among a Puerto-Rican population by Dr. Regina Holloman; and for use in

German by the author and Dr. Georg Wieser of the Institute for Advanced Studies and Scientific ° Research in Vienna, Austria. The author is indebted to Dr. Wynona Hartley of the University of Kansas Medical School for the development and use of the ‘‘Who Am IJ?” sub-routine; the Midwest

samples resulted from her work both in her present position and with associates at the Greater |

Kansas City Mental Health Foundation in Kansas City, Missouri. The author is also indebted to the late Dr. Gerd Fraenkel who developed the coding for the linguistic sub-routines and who worked tirelessly until his recent death on spoken tape analyses and in observation and analysis. 2 Early studies were completed at the Center for Psycho-social Studies, Learning Research and Development, University of Pittsburgh, pursuant to a contract with the Office of Education, HEW, under the provisions of the Cooperative Research Program. Some later studies were supported in part by Ford Foundation International Dimensions grants of the University of Pittsburgh Center for International Studies.

56 RACE AND INTELLIGENCE [AS8 3 Experiments in modification have all been successful to some degree. Degree of success

appears to have some relationship to the dominant conceptual style of individuals. In our

experience, the Conflict-concrete style is the most difficult to modify, and its carriers consistently score highest of the four groups on F scales. The polar styles, Analytic and Relational, are the next most resistant to modification with the Flexible style the most easily modified. Modification also

appears to be related to the point of intervention. That is, if one assumes that primary group norms, conceptual style, self-concept attitudes and values are linked by dominating rules, the changes in any of these behavior patterns would be expected to affect all of the others as well. Our experiences indicate that self-concept and referent ‘‘others’’ are most easily changed, and they change under normal conditions more frequently than other types of behaviors. However, no structured comparisons have been undertaken to date, and practical research problems have made doing so difficult. From the standpoint of theory, the planning of a controlled environment in which rules are taught directly and used consistently on multiple levels including those of social organization could be expected to be most effective. . * Perceptual distance between the observer and the observed appears to be sufficiently great so that he can study and manipulate his field, but not so close to himself as to be affected by it. It is also viewed as relatively stable and non-reacting even when personalistic. Tools for observation and

manipulation designed by Analytic style users are characteristically “‘distance”’ tools. Supported by research observations of Basil Bernstein and his associates at the Institute of

Education, University of London. -

6 Vocal modulation is used when subtle differences in meaning are intended for which syntax and grammar provide no mechanisms. 7A common conditioning mechanism is the word game, “‘topping’’ or “‘capping,’’ a game of

mutual insult played by both children and adults through free association of links. It may be

contrasted with a game like ‘‘Scrabble,’’ a word game played by Analytic style users by forming and reforming words from their parts. 8 This observation is supported by similar findings among the samples in England reported by

Basil Bernstein and his staff at the University of London. .

°For example, on attitude questionnaires, Flexible style users responses are characterized by the writing of many qualifications on the margins to identify the circumstances under which their statements may hold, and in frequent erasures on other sub-routines from one kind of answer to another. Stories and explanations exhibit similar vacillation. 10Samples in Social Work, Education and Psychology. Fielding areas were selected to reflect student bodies at Eastern and Midwestern universities. A Southern University sample was also- | planned; however, the untimely death of an associate has postponed the analysis of data. 11 earning Research and Development Center, University of Pittsburgh, Richard Goldman,

principal investigator. |

12May be found in the program reports of Project Succeed, Learning Research and

Development Center, University of Pittsburgh. Program of Mrs. Jane Taylor at Munhall High School, Munhall, Pennsylvania. REFERENCES CITED Chambers, Guinivere

1969 Report on Selection into Child Care Programs. In Research Conference on Urban Sub-cultural Differences. R. Cohen, org. University of Pittsburgh.

Cohen, Rosalie .

. 1967 Primary Group Structure, Conceptual Styles and School Achievement. University of Pittsburgh: Learning Research and Development Center, Studies No. 1, 2, 3.

1968a Formal Role Analysis and the Culture of Poverty. Paper Read at the Ohio Valley Sociological Society Annual Meetings, Detroit.

1968b Is Formal Role Analysis Generalizable to Low Income Groups? Paper Read at the Midwest Sociological Society Annual Meetings, Omaha.

1968c The Relation between Socio-cultural Styles and Orientation to School Requirements. Sociology of Education 41:201-220. 1969a Conceptual Styles, Culture Conflict, and Nonverbal Tests of Intelligence. American Anthropologist 71:828-856.

1969b A Social-psychological study of 450 Youth Development Center Students. (Mimeographed. )

Quarterly 10:19-28. : |

Cohen, Rosalie, Gerd Fraenkel, and John Brewer

1968 The Language of the Hard-core Poor: Implications for Culture Conflict. Sociological

Cohen | CONCEPTUAL RULE-SETS AND LEARNING ABILITY 57 Garretson, W. S.

1961 College as Social Object: A Study in Consensus. PhD dissertation, State University of Iowa.

In Press Behavioral Correlates of Self Concept. Chapter in a Forthcoming Book. Hartley, W. S.

1971 Report of a Special Education Project for Retarded Children. (Mimeographed.) Kagan, Jerome, Howard A. Moss, and Irving E. Siegel 1963 Psychological Significance of Styles of Conceptualization. In Basic Cognitive Process in Children. Chicago: University of Chicago Press, Society for Research in Child Development Monograph 86. Kuhn, M. H., and T. S. McPartland 1954 An Investigation of Self-attitudes. American Sociological Review 19:69-76.

Genetic and Environmental Components’ of Differential Intelligence

THOMAS G. GREGG : MIAMI UNIVERSITY , | PEGGY R.SANDAY

| DEPARTMENT OF ZOOLOGY

DEPARTMENTS OF ANTHROPOLOGY AND URBAN AFFAIRS ©

CARNEGIE-MELLON UNIVERSITY |

THE STUDY of differential intelligence, both within populations and between them, is fraught with difficulties. Yet, the subject is one of intense interest, as has been clearly demonstrated by the widespread press coverage accorded a recent article by Jensen (1969) and the barrage of response and counterresponse papers appearing in the professional journals. Much of this interest legitimately stems from a desire to discover the causes of differential intelligence so as to know’ whether compensatory education programs are based on the proper assumptions. In this paper we will first consider how the relative importance of genetic and environmental components of differential intelligence within a population are determined. In the light of this discussion we will then present an alternative hypothesis to the one suggested by Jensen that the mean difference in

IQ between blacks and whites has a genetic basis. |

Obviously, the causes of differential intelligence are genetic, environmental, or both. But~ — beyond this there seems to be little agreement. It seems incredible that anyone still doubts that | there are genetic factors involved. The heritability estimates cited by Jensen (1969), even if they are twice too high, should establish this point. It is fruitless to nitpick at the accuracy of the heritability estimates. It is equally absurd to deny that the environment plays a role in determining intelligence. Consequently, the question boils down to whether it is meaningful to ask what the relative importance of the genetic and environmental components is; and if so, whether their | relative importance can be determined. For simple qualitative biochemical traits, it can be seen that there are some—blood groups, for example—that cannot be altered no matter what is done to the environment. Others, such as the conditional lethals of Neurospora and the various microorganisms, can be completely alleviated by environmental manipulation; for example, an organism that cannot synthesize a particular amino

acid can be rendered completely normal if that amino acid is added to the medium. (The administration of insulin to diabetics is a similar situation, although in this case, the individual may not be completely normal.) Perhaps strict environmentalists, if any in fact exist, will seize upon

this latter point as justification for their views that environmental manipulation can achieve everything. However, in any organism these simple biochemical traits interact with each other and/or with

molecules introduced from the environment in numerous ways and with ever increasing

58 |

organizational complexity to produce the complex phenotypes that we usually observe. In other - words, for any given trait which involves a level of organization higher than the polypeptide itself (we might call the polypeptide, which is the direct product of gene action, the primary phenotype and the higher levels of complexity secondary phenotype, tertiary phenotype, etc.) there are many interrelated chemical reactions involved. Some of these can be directly altered by environmental

influences and some of them cannot. With respect to intelligence, the situation is further

Gregg & Sanday ] GENETIC AND ENVIRONMENTAL COMPONENTS 59 complicated by the fact that the individual’s IQ is somehow developed by interactions between sensory stimuli and the central nervous system. Furthermore, for IQ it seems quite likely that there are two fairly distinct levels of environmental input: (a) those factors involved in the physiological development of the nervous system and (6) sensory stimuli.

This greatly oversimplified description of the organizational complexity of an organism is sufficient to illustrate that it will probably never be possible to identify the points at which genetic

factors and environmental ones interact and what the extent of any specific interaction is. 7 Attempts have been made, however, to estimate the relative importance of the genetic and ‘environmental components. These have usually involved strategies that attempt systematically to . alter genotype or environment while holding the other constant. Many of these have made use of the concept of heritability. Heritability has been discussed extensively in general by Falconer (1960) and in connection with IQ by Jensen (1969) and Crow (1969). In brief, the total variance’. V, can be partitioned

into the genetic variance V|, and the variance due to environmental factors V. (ignoring interaction and error variance). Heritability, then, although it can be defined in other equivalent

ways, is the proportion of the total variance attributable to genetic differences in the population.

, Vi . Ve ,

Thus, heritability equals V_,/V,; and Z

— Vo ~~ Vo = I.

Heritability, despite its superficial simplicity, is by no means without pitfalls of interpretation,

and in pointing out certain ones, one is apt to fall into others. Jensen (1969), after a lengthy explanation of heritability, seems to conclude that a trait with a high heritability such as IQ is primarily determined by genetic factors and is therefore not much influenced by alterations in the environment. And although he never says so directly, this seems to be the basis for his hypothesis that there are genetic differences contributing to the mean difference in JQ between blacks and whites, and generally that environmental changes, above a threshold level, will have little effect on

IQ. Actually, he never makes it clear why a high heritability means that changes in the environment will have little effect on IQ. One line of reasoning could be the one just mentioned, i.e., high heritability means that the trait is primarily determined by genetic factors. A related, but

not identical, argument could be that since the proportion of the variance attributable to environmental variation is small, environmental manipulation does not have a very large share of

the variance to work with and thus cannot effect much change. Both of these arguments are

- fallacious. With regard to the first argument, it is clear that heritability is not a measure of the extent to which a trait is determined by genetic factors. Rather, it is a measure of the proportion of the total phenotypic variance which is attributable to genetic differences between individuals. These are not the same, as can be shown by expanding Jensen’s discussion of the constancy of heritability. If we take a hypothetical population with a certain amount of genetic variability and place it in a highly variable environment, heritability may very well be low, perhaps as low as 0.2, because of the variable environment and the consequently high V._. If we place the same population in a less variable environment, heritability will perhaps rise to 0.6. Finally, if we put it in a completely

uniform environment, heritability will become nearly one. Now, surely the genetic contribution to " the expression of the trait in the individuals of this population is not going to differ in these three situations. If heritability were a measure of the genetic contribution to the expression of the trait,

the genetic contribution would be different in each of the three environments; but it is not. Heritability only tells what proportion of the variance is due to genetic differences, not the extent to which a trait is determined by genetic factors. This point can be further illustrated as follows. In a wild population of mice, for example, there

will be genetic variability for most traits, more for some than for others, including the mouse equivalent of “‘intelligence.”” The environment for at least some of these will be reasonably stable “in the ecological niche inhabited by mice and they will have high heritabilities. If this population is

60 , RACE AND INTELLIGENCE [AS8 taken into the laboratory, the heritabilities will increase because of the even more uniform environment. If a pair of these mice is systematically inbred, they will produce a population in which all the individuals will be homozygous for the same genes. In this case the heritability for all traits will be zero because there is no genetic variability and thus no genetic variance. But surely no

one would conclude that because the heritability is zero there is no genetic contribution to the traits in question. Likewise, if there is no environmental variance, that does not mean that the environment has no influence on the expression of a trait. Still another way of looking at this problem is to suppose that all of the individuals in a population are homozygous at 95% of the loci affecting a particular trait, with differences existing only at the other 5%. The heritability may be high, yet 95% of the loci affecting the expression of | the trait are making no contribution to the heritability. Heritability is not a measure of the genetic contribution to any trait, and it seems obvious to us that one would need to know the magnitude of the genetic contribution to the trait itself, rather than to its variance, before hypothesizing that’. mean differences in its expression are genetic. ~

On the other hand, the reasoning with regard to IQ could be that since 80% of the variance is genetic and only 20% environmental, environmental improvements do not have a very large share of the variance to work on and so cannot be expected to accomplish much and therefore the trait is largely controlled by genetic factors. The trouble with this argument is that it does not allow for the operation of uniform factors in the environment that do not contribute to the environmental variance but do affect the expression of the trait. For example, a herd of dairy cattle kept under highly uniform conditions and fed only hay will have a heritability of one for milk production. But this does not mean that changes in the environment will not have any effect on the expression of the trait. The uniform addition of grain to the diet will dramatically increase milk production, which will still have a heritability of one.

If there are other lines of reasoning, they are not easy to see. In either case, it is entirely unwarranted to conclude that the genetic contribution to the expression of IQ is so large that the environment can have little effect. And it is this conclusion on which Jensen bases his hypothesis that the mean difference in IQ between blacks and whites is due partly to genetic differences. Some of these points have been elegantly discussed elsewhere by Crow (1969). Jensen’s statement is also based on the agreement among geneticists that groups which have |

been geographically or socially isolated from one another for many generations are practically certain to differ in their gene pools and consequently are likely to show differences in gene frequencies for given traits, including mental abilities. It is true that geneticists would agree with these statements, but the significance of such genetic differences is not well understood. One approach would be to examine these differences in light of Dobzhansky’s concept of the adaptive norm (Dobzhansky and Spassky 1963). Briefly, the concept is that in a particular population there | are a number of different genotypes that have similar or equal fitnesses. Fitness in this case would —

correspond to IQ. This concept can easily be extended to different populations so that even if. there were genetic differences between blacks and whites for IQ the mean expression of the trait

would not necessarily be different. For instance, there are racial differences in the relative frequencies of various blood constituents which are under genetic control, yet we do not assume that there is a mean difference in their ability to carry oxygen, produce antibodies and protect

. against disease, carry nutrients and hormones, produce enzymes of various sorts, etc. We do not claim that there could be no differences, but just that so far as anyone has discovered the blood of one race does not function better than that of another despite genetic differences. An alternative hypothesis to the one proposed by Jensen would be that the range of genetic variability is similar for blacks and whites and that the observed mean difference in IQ between the two groups can be attributed to qualitative differences in environment. This hypothesis is alluded to by Crow (1969:160) in his statements “‘that the environmental factors may differ qualitatively

in the two groups’’ and “that being white or being black in our society changes one or more aspects of the environment so importantly as to account for the difference.”’

There is a growing body of empirical evidence to support the assertion of qualitative environmental differences between blacks and whites. The research of Johnson and Sanday (1970) summarizes some of this evidence and empirically establishes a significant difference between

Gregg & Sanday ] GENETIC AND ENVIRONMENTAL COMPONENTS 61 blacks and whites in three urban poverty neighborhoods along several dimensions which anthropologists (Beals 1967) have suggested as suitable boundary criteria for delimiting cultural systems. Johnson and Sanday suggest that the reason for these differences lies in a central theme of survival which motivates the life styles of black people. The basis for this theme lies in a history of discriminatory practices which have destroyed for black people an orientation to opportunity and a belief in other people. The data of this study further suggest that the subcultures delimited for blacks and whites in an , urban poverty neighborhood may be generalizable to the nonpoor. The authors propose that in ~ spite of all the various ethnic groups found in America, there are two definable subcultures—one ) black and one white. This conclusion, they point out, depends upon a dynamic perspective. If the American culture is examined at any one point in time it looks like a conglomeration of ethnic subcultures. However, -

when viewed over a period of time these ethnic groups eventually blend into the overall structure adding their part to the cultural repertory of the White subculture [Johnson and

Sanday 1971:1389]. .

The black subcultural system is an adaptive phenomenon. It emerged and has persisted because of the adaptation Black people have had to make to being continually barred from strategic status positions and resources because of discrimination on the basis of color [Johnson and Sanday 1971:139].

: If this is the case—and it remains to be documented more thoroughly—then it is plausible to suggest that it is a cultural difference and not a deficit in ability which confronts the members of either subculture when attempting to achieve success within the framework of the other. These considerations lead Johnson and Sanday to suggest that the documented failure (cf. Jensen 1969) of compensatory education can be traced not to a genetic inferiority of the participants but to the predilection of the developers of such programs to an assumption of cultural deficiency and inferiority and the consequent internalization of these attitudes on the part of those who are

supposedly being helped. | The implications of the cultural-difference argument are numerous. For example, it cannot be said that blacks and whites of the same socioeconomic status or family structure type are in fact exposed to comparable environments. In a study of Negro adolescents, Henderson (1967) presents

data to show that ghetto youth tend to perform very much alike on such measures of social performance as educational achievement, law-violating behavior, aspirations, and perceptions of life chances. Henderson (1967:138) demonstrates that this similarity exists even when youth from nonpoverty families are compared with youth from poverty families. Moreover, youngsters from broken homes perform as well as do youngsters from intact homes. These facts led Henderson to conclude that family structure and socioeconomic status had no effect on social performance. The data of Johnson and Sanday show that when family structure is controlled and the racial samples are separated into poor and working-class groups, the differences between black and white still persist. The conclusion from both of these studies is that the significant variable. determining behavior for blacks is participation in the black subcultural system. Another implication of the cultural-difference argument relates to the above discussion of the

effect of a highly variable and a homogeneous environment on heritability estimates. As mentioned above, heritability estimates will be much higher in a homogeneous environment, since

not much of the variance will be accounted for by differences in the environment. Furthermore, in such a case, the total phenotypic variance will be smaller, since the proportion of the variance due to the environment is very low. It is interesting to note that the phenotypic variance for IQ in the black population is 60% as great as in the white population. If the lines of reasoning presented here

are correct, the heritability estimates should be correspondingly higher unless there are compensating sources of environmental variability that are as yet undiscovered. No such estimates are now available for the black population.

To illustrate these points further, let us consider a population of cows half of which are fed erain in amounts that are varied experimentally and the other half hay only or a uniformly small ~ amount of grain only. We would expect higher milk production with increase in the amount of

62 RACE AND INTELLIGENCE [AS8 grain for the first group and a correspondingly higher variance for this trait, largely traceable to the varying environmental conditions. On the other hand, there would be a smaller total variance for the

expression of the trait in the second group of cows, and the heritability estimate would be higher. | There is evidence that this analogy may be applicable to the expression of JQ in the black and white groups. Henderson (1967) stresses the uniformity of the black environment resulting from

the racial, physical, mental, and social isolation imposed by the white subculture. This homogeneity, according to Johnson and Sanday (1970), is further intensified by the neighborhood public school system which in urban America affects blacks and whites differently. Urban schools in the white subcultural system provide the means for ever expanding social contacts with heterogeneous class and ethnic elements as the student moves through the system. In the isolated urban black community, in contrast, the student is exposed to a homogeneous class and racial

context and geared to a limited set of expectations. Furthermore, there is the leveling effect exerted by the school indicated in the tendency for ghetto youth to perform alike on measures of social performance. ~

Although it can be established from this evidence that blacks are in a homogeneous environment and whites in a variable one, more intensive analysis must be made of whether. the environmental factors affecting the expression of IQ in the two populations are in fact uniformly negative in one and highly variable in the other. The work of Cohen (1969) on the relationship between primary group structure, cognitive style, and language structure suggests some of the cultural factors which might be involved.

In conclusion, it is agreed that genetic variability can be partially related to differential intelligence within populations. However, heritability estimates are not measures of the extent to which a trait is determined by genetic factors, but only of the proportion of the variance due to genetic differences. Before hypothesizing that the mean differences in the expression of a trait between populations are genetic, it seems to us that one would need to know the magnitude of the

genetic contribution to the trait within the populations. A further complication in

between-populations investigations is the influence of the environment. It cannot be assumed that two environments are comparable. In the case of blacks and whites, the evidence is that they are

definitely not comparable. Furthermore, there is the possibility that in one environment factors . exist which increase the mean expression of the trait, while in another environment these factors __ are absent or variable within a narrow range.

NOTE | |

} Variance, as defined by Hays (1963:177), is a measure of the degree of spread of a group of scores around the mean. The more the scores tend to differ from each other and the mean, the larger the variance. The variance will be zero when all the scores in the distribution are the same as the mean. Individual differences in IQ scores are represented as population variance in a phenotype

Vp: This total variance of the phenotypes is conceptualized in terms of the proportions of the

variance attributable to various genetic and environmental components (Jensen 1969.:33). REFERENCES CITED Beals, Alan R.

, 1967 Culture in Process. New York: Holt, Rinehart and Winston. Cohen, Rosalie A.

Crow, James F. .

1969 Conceptual Styles, Culture Conflict, and Non-Verbal Tests of Intelligence. American Anthropologist 71:828-856.

1969 Genetic Theories and Influences: Comments on the Value of Diversity. In Environment, Heredity, and Intelligence. Harvard Educational Review, Reprint Series No. 2. Dobzhansky, Theodosius, and B. Spassky 1963 Genetics of Natural Populations XXXIV: Adaptive Norm, Genetic Load, and Genetic Elite in Drosophila pseudoobscura. Genetics 48:1467-1485. Falconer, D. S. 1960 An Introduction to Quantitative Genetics. New York: Ronald Press. Hays, William L.

1963 Statistics for Psychologists. New York: Holt, Rinehart and Winston.

Gregg & Sanday | GENETIC AND EN VIRONMENTAL COMPONENTS 63 Henderson, Donald M.

1967 A Study of the Effects of Family Structure and Poverty on Negro Adolescents from the Ghetto. Unpublished Ph.D. dissertation, University of Pittsburgh. Jensen, Arthur E. 1969 How Much Can We Boost IQ and Scholastic Achievement? In Environment, Heredity, and Intelligence. Harvard Educational Review, Reprint Series No. 2. Johnson, Norman J., and Peggy R. Sanday

1970 Subcultural Variations in an Urban Poor Population. American Anthropologist 73:128-148.

On Creeping Jensenism a _

C. LORING BRACE —a| | FRANK B. LIVINGSTONE DEPARTMENT OF ANTHROPOLOGY

UNIVERSITY OF MICHIGAN | |

CONCERN about the meaning of the physical differences between human populations dates

back to before the dawn of written history, but it did not really become a major issue until the Renaissance, when the revolution in ocean-going transportation brought large numbers of diverse people physically face to face. The superior technology of the Europeans enabled them to coerce |

and exploit the peoples encountered, many of whom were forcibly uprooted and relocated as slaves. While one could argue that this was one of the most extraordinary examples of barbarism in .

the annals of “civilization,” it was justified at the time, not so much on the basis of race, but

because the people being enslaved were “heathens.” Actually the “Christianity” of the unprincipled and largely illiterate slaving crews was often a convenient fiction, and the real reasons

why the slave trade continued were greed and the force of firearms. ~ The phenomenon of the Christianized (and even literate) slave removed the initial rationale, but, needless to say, the institution persisted. Economics and the established social order in the American South assured its perpetuation while the Calvinistic fatalism of the North tended to maintain the status quo with little question. For example, in 1706 that godly Puritan, Cotton Mather, heartily supported the Christianization of ‘‘the Negro”’ on the one hand, while arguing on ~ the other that baptism does not entitle a slave to liberty (Osofsky 1967:389). Although Quakers publicly and repeatedly extended Christian principles to the extreme of condemning slavery from the latter third of the seventeenth century on, this did not become a matter of general concern

until the winds of ‘‘enlightenment” ushered in the Age of Reason, complete with elegant statements on the nature of man and human rights, and culminating in the American Revolution. | Backlash followed the excesses of the revolutions in Santo Domingo and France, and it was more _ than half a century before the momentum of the Enlightenment was regained and slavery was finally abolished—in name, at least. (For excellent and detailed historical treatment, see Jordan

1968 and Stanton 1960.) |

The intellectual legacy from late eighteenth century idealism, however, is apparent in the

continuing debate concerning the meaning of human physical differences. Clouding this debate has been another legacy of considerably less exalted origins. This legacy survives in the wretched social and environmental surroundings that continue to characterize the living conditions of Negroes in the United States. Quite recently there has been an explicit recognition of this situation—witness

the belated extension (in 1954 and 1969) of eighteenth century Constitutional guarantees to Americans of African ancestry—but, at the same time, there has also been a continuation of the attempts to justify bloc differences in human treatment that began when slavery was already an

accomplished fact. ,

Enforced inferior status—slavery—was initially justified by the heathen state of African peoples. All the other attributes of Negroes were automatically stigmatized and, although the justification changed through time, their association with inferiority has remained a continuing item of faith. Black skin color was regarded as the result of the curse placed on Ham and all his descendants. Negroes then were identified with the Biblical Canaanites, their servitude was considered justified by Noah’s curse, and their attributes were regarded as visible evidence of the Lord’s displeasure.

64 :

Brace & Livingstone | ON CREEPING JENSENISM 65 With the rise of a rational world view in the latter part of the eighteenth century, this became an increasingly unsatisfactory explanation to thoughtful men. Separate creations—separate and unequal—were suggested, although this was offensive to the faithful who preferred something

which remained compatible with the Biblical original pair. The result was a pre-Darwinian development of a form of evolution by means of a crudely conceived kind of natural selection (Smith 1965). Inevitably, however, vested interests of a social and political nature clouded all efforts at objectivity, as they continue to do. The record of published attempts to justify existing social inequalities on the basis of innate or biological differences extends unbroken from the Renaissance era of exploration and subsequent colonization (Jordon 1968) through the nineteenth century (Barzun 1965, Stocking 1968) to the present day. The association with events that maximize human misery (epitomized in the American Civil War and World War II for instance) is so clear that each new attempt to justify differential treatment of large numbers of human beings,

often prejudged en masse, should be examined with the greatest of care. |

We offer this cautionary preamble because yet another such attempt has been made, this time couched in the language of modern science, published in circumstances which tend to enhance its prestige, and given widespread if uncvritical publicity. The presentation we refer to is Jensen’s (1969a) monograph-length article in the Harvard Educational Review. Pointing up the obvious seriousness of its implications is the fact that the very next issue of the Review (Vol. 39, Spring 1969) contained responses from more than a half-dozen scholars. Many of the points raised are well-taken, but, in the haste of immediate reaction, documentation was incomplete, important aspects were missed entirely, and organization suffered.! Adding further to this unfortunate confusion is the treatment it has been given in the popular press. The discussion in the prestigious New York Times Magazine (Edson 1969), attempting journalistic impartiality, presents the various arguments and rebuttals as though they were all of equal probability. The result has been the widespread circulation of conclusions which are possibly pernicious, and certainly premature, to a

evaluation. _ |

readership which, though highly literate, is largely unable to make a reliable independent Seen in perspective, ‘‘jensenism, n. the theory that IQ is largely determined by the genes” (Edson 1969), is the extreme if logical outcome of the preoccupation which the field of behavior genetics has had with the “defeat”? of the environmentalist heritage of Watsonian behaviorism

(Hirsch 1967b:118-119). The extreme of the environmentalist position is best expressed in Watson’s (1924:82) famous dictum:

Give me a dozen healthy infants... and I’ll guarantee to take any one at random and train him to become any type of specialist I might select—doctor, lawyer, artist, merchant-chief and, yes, even beggar-man and thief, regardless of his... abilities... and race of his ancestors. Fulminations against this position by committed racists, who anathemize it as “‘equalitarian doctrine,” are well known (Putnam 1961). Objections to extreme environmentalism have also been

repeatedly offered by students of behavior genetics, one of whom, Hirsch, has variously characterized it as a “‘counterfactual assumption” (1961:480), a “counterfactual dogma”’ (1963), and a “counterfactual ... postulate” (1967a), based on ‘“‘fallacious reasoning”’ and “excessively anti-intellectual’’ (1967b). Both these positions, the racist and the behavior geneticist, represent reactions to the emotionally based humanitarianism in much of recent social science. Leaving the racists out of it for the moment, it is evident that the advances in behavior genetics

and ethology must be considered of prime importance among the recent major developments in biological, psychological, and anthropological science. This has led to the organization of many symposia and fostered the production of a series of popular books such as those by Lorenz (1966), Ardrey (1966), Morris (1968, 1969), and Tiger (1969). Questions being asked include “‘Why are men aggressive?” “‘Does man have a pair bond?’’ and “Is there such a thing as male bonding?”’ The

interest in basic human biology is apparent, and much of this new questioning is concerned with supposed “species-specific”? characteristics of man, although there is a tendency to postulate a genetic cause for human behavioral differences—Lorenz’s (1966:236) discussion of Ute aggression, _ for example. Given these recent trends, it seems inevitable that attention would be focused on intelligence differences and that genetic causation should be stressed.

66 ] RACE AND INTELLIGENCE [AS8 In the general picture, caution should be urged on two accounts, and Jensen’s work illustrates what can happen when neither problem is adequately considered. First, the distinction between

individual and population performance should be clearly perceived; and, second, if genetic | differences are to be the object of concern, thorough control of the environmental component of observed variation should be achieved.

Considering the first issue, one of the roots of the problem is the inability of Western science—and biological science in particular—to recognize and differentiate between individual and

populational phenomena. Certainly birth rates, death rates, or intelligence levels are the result of individual performances, but their variability among human populations is not primarily due to

individual genetic differences, however much these may be involved. Many of the recent discussions of incest, inbreeding, and sexual behavior demonstrate the same inability to differentiate populational and individual phenomena (Roberts 1967; Livingstone 1969). Ironically, Hirsch, one of the people most responsible for the trend of research which Jensen has carried to ,

something of an extreme, has apparently sensed the fact that the approach he has promoted is being carried too far and has recently articulated a brief critique which could be applied to

Jensen’s work (Hirsch 1968:42): .

What is the relative importance of genetic endowment and of environmental milieu in the development of the intelligence of an individual? The answers given to that question ... have nothing to do with an individual, nor are they based on the study of development. The answers have been based on the test performance of a cross-section of a population of individuals at a single time in their lives.

In his critical comment, Kagan (1969) in fact notes that Jensen makes no effort to resolve this issue.

Turning to the environmental component in observed behavior, we see that again Jensen has made little effort to grapple with the problem. Admittedly, the thrust of recent work in behavior genetics has been to discount the environmental contribution, but, again, Hirsch, who has been a major part of this thrust, has recently provided a warning against excesses in this direction. While he refers to the heredity-environment question as “‘a pseudo-question to which there is no answer”

(1968:42), he goes on to warn that

it should also be noted that one cannot infer from a high heritability value that the influence of ~ environment is small or unimportant, as so many people try to do [1968:48, italics Hirsch’s]. To illustrate the unwarranted extreme of what Medawar (1961:60) has called “‘geneticism,”’ Hirsch refers to the controversial pronouncements of William B. Shockley, Nobel laureate in physics. (For excerpts of Shockley’s speech, see Birch 1968:49, and for a responsible rebuttal, see Crow, Neel, and Stern 1967). Mention of Shockley in this regard is particularly important since Jensen was apparently much impressed by Shockley when he was visiting Stanford in 1966-67. The result was what has been called his “most unfortunate speech”’ illustrating “‘the dangers of

inappropriate use of both the concept of heritability and that of race by the biometrically unsophisticated” (Hirsch 1967a:434). Jensen’s speech, in turn, provided the background for the article which is the focus of our concern here.

The first half of Jensen’s article is a comprehensive review of quantitative genetics. He . concludes this review with the-statement (1969a:65) that the question of whether heritability estimates can contribute anything to our understanding of

the relative importance of genetic and environmental factors in accounting for average phenotypic differences between racial groups (or any other socially defined group) is too

complex to be considered here. )

Since heritability estimates are specific to the populations studied—at the time studied—and since they vary considerably with environmental circumstances, Jensen, as quoted, correctly expresses the problem and should have stopped there. He does not stop, however, and proceeds under the assumption that there is a definite intelligence heritability of .8. Not only is this the highest found,

but it is based on twin data, which are most unlikely to differentiate the environmental component.* This estimate he then generalizes to all humanity.

Brace & Livingstone | ON CREEPING JENSENISM 67 Despite his statement that the matter is “too complex,” his further discussion of racial differences apparently implies that the preceding review of quantitative genetics supports his view. It does not. Furthermore, we fail to see how, after pointing out that environment can change IQ by as much as 70 points, he can make the statement that “‘in short it is doubtful that there is any

significant environmental effect on IQ.”

For purposes of comparison, let us take the case of stature. As a “trait,” it is sufficiently complex to warrant the expression of doubts concerning simplistic treatment, although it is - somewhat less of a “typological reification” (Hirsch 1968:44) than intelligence. Treating them for

the moment as though they were comparable traits, it is evident that both are under polygenic control. Proceeding with this in mind, Kagan (1969), in his initial reaction to Jensen’s article, cites the difference in stature between rural and urban populations in Latin America to show the effect of environment on an inherited trait. Hunt (1969), on his part, makes casual mention of stature in colonial Jamestown, Plymouth, and during the American Revolution, noting the radical changes that have taken place since that time. Other examples are well known and documented, but perhaps the changes in Sweden and the Low Countries in the past 100 years constitute a better example (Chamla 1964). Stature certainly has a major genetic component. Estimates concerning its heritability vary widely, although they average about .5, comparable to the average for IQ despite what Jensen claims. For example, Kagan and Moss (1959) have found an average correlation of .43 between parents and offspring for IQ, and an average correlation of .36 for stature. In the past 100 years, or about four generations,

very little genetic change could have occurred, particularly when one considers the lack of evidence for strong selection. However, in that time, the stature of the average adult male in many European countries has changed 4—5 in., or almost two standard deviations. Since populational

differences in IQ are at most about one standard deviation, we do not see why anyone would

maintain that the same amount of nongenetic change could not occur where IQ is concerned—particularly when the trend of IQ increase is not only known but, in some cases, even greater than the trend for increase in stature. The principal reason for the observed increase in stature appears to be a change in nutrition,

particularly an increase in the amount of protein in the diet. A similar increase in stature is occurring in Japan (Kimura 1967), and there is a strong correlation between stature and protein

intake (Takahashi 1966). Recently evidence has been accumulating to suggest that mental development can be markedly influenced by nutritional inadequacy—particularly where protein-calorie malnutrition occurs during the period in development when the brain is growing most rapidly (Cravioto, DeLicardie, and Birch 1966; Davison and Dobbing 1966; Eichenwald and

Fry 1969).

Brain weight, amount of brain protein, and RNA increase linearly during the first year of

human life—all being directly proportional to the increase in head circumference. The amount of

brain DNA is regarded as a good indication of cell number and, although it largely ceases to increase at six months, it too maintains a direct relation to head circumference during the first year of life (Winick and Rosso 1969). With this in mind, it is of grim interest to note that in cases of severe malnutrition, head circumferences have been recorded that were two standard deviations

below the mean for normal children of the same age. Brain weight and protein were reduced proportionately, while DNA content was reduced at least as much and in some cases more (Winick

and Rosso 1969:776). In one instance, rehabilitation was tried on malnourished children and behavioral recovery was measured by the Gesell method. Children who were under six months of . age on admission retained their deficit, leading to the conclusion (Cravioto and Robles 1965:463) that “there is a high possibility that at least the children severely malnourished during the first six months of their lives might retain a permanent mental deficit.” In another instance, recovery of

head circumference following early malnutrition lagged way behind other aspects of growth

recovery (Graham 1968, esp. Fig. 3). :

The studies cited above deal principally with the consequences of malnutrition in Latin America, but the record from Africa is equally clear: small fetal and neonatal brain sizes among _ the starving people of Biafra (Gans 1969), decreased cranial circumference and reduced brain weight among the malnourished of Uganda (Brown 1965, 1966), reduced cranial circumference

68 RACE AND INTELLIGENCE [AS8 and lower intelligence test scores in the Cape Coloured of South Africa. In the latter case, the reduction in circumference and test score was in comparison with a control population, also of Cape Coloured, but one which was not suffering from severe malnutrition. The differences were statistically significant (P = < .01) but, interestingly enough, there was no significant difference between the parents of the two groups (Stoch and Smythe 1963, 1968).

One could argue that the works we have mentioned deal principally with extremes of malnutrition, and, in fact, Jensen does so, claiming that there is little extreme malnutrition in the United States. Yet with substantially more than half of the American black population living at or below the poverty level as defined by the US Department of Health, Education, and Welfare, with the shocking deprivation recently and belatedly brought to the attention of the US Congress (Javits 1969; Hollings and Jablow 1970), and with the obstacles to survival facing the American poor so graphically depicted by Coles (1969), it would be most surprising if malnutrition did not — contribute something to the lowering of intelligence test scores in American Negroes—all other , things being equal, which, of course, is not the case. ~

Before leaving strictly biological matters, we should note that deprivation need not be extreme for its consequences to show. Admittedly, these data are derived from studies on experimental animals rather than on human beings, but this can hardly justify ignoring their implications. Inadequate nutrition delays development of the myelin nerve sheaths in rats, and the deficit is not completely made up. The importance of this particular study is to be seen in the fact that the deprivation was that of the lower end of the “normal” range and did not constitute “‘starvation’”’ (Dobbing 1964:508). Demonstration of the reduction in cell number of other brain tissues following early deprivation is equally clear (Dickerson 1968:335; Dobbing 1968:195). Not surprisingly, the behavioral consequences are also apparent (Hichenwald and Fry 1969:646):

Protein deprivation in early life not only causes ... behavioral changes but also reduces the capacity of the experimental animal to learn at a later age. Furthermore, rats born of and

suckled by malnourished mothers are similarly deficient in their learning capacity.

: So far we have stressed the role of nutrition—particularly protein-calorie malnutrition—in the stunting of mental development. Vitamin deficiency, illness susceptibility, and chronic ill-health all contribute to a malnutrition-disease syndrome (see Scrimshaw and Gordon 1968) which, given nothing else, should certainly lower performance levels on intelligence tests. These factors alone - can go a long way towards accounting for the differences in the tested intelligence of the world’s

populations, but they constitute only a part of the nongenetic background of testable mental performance. However, strictly experiential factors can have an even more pronounced effect on intelligence test performance and may completely mask the nutritional and genetic factors. Obviously there are many problems associated with estimating the heritability of behavioral traits. Data on IQ tests derived from family studies do indicate a genetic component, although this may in fact be somewhat less than the heritability for physical traits. The heritable component is extremely difficult to separate from the nonheritable component in assessing the results of most tests of complex behavior, and it is apparent that Jensen really does not make the effort to do so. The cumulative interaction of particular types of experience with facets of biological maturation produces an elaboration that is extremely difficult to assess in terms of what percentage of which

part is represented in the end-product. This is what Hirsch meant when he referred to the nature-nurture problem as a “‘pseudo-question to which there is no answer,” but if Jensen expects to demonstrate the credibility of his conclusions, it is a question to the solution of which he must direct research efforts more carefully planned and better controlled than any that have yet been

undertaken or even proposed.’ )

Studies on the heritability of behavioral traits in Drosophila are frequently cited to bolster estimates on behavioral heritability in man, but, even ignoring the enormous phylogenetic gap,

recent research has shown the heritability of the oft-mentioned geotactic and phototactic responses to be quite low. Richmond (1969), for example, found the heritability of both to be less

than .2 in all cases and not significantly different from .0 in one instance. Dobzhansky and Spassky (1969) found realized heritabilities for these traits to be below .1. We should like to make it quite clear that we do not deny the existence of a genetic component

that contributes to differences in performance on IQ tests—within a single population, where

Brace & Livingstone ] ON CREEPING JENSENISM 69 conditions of early experience and education are relatively similar. The differences, however, are

less important than implied by Jensen.” For example, he has noted that there are significant correlations between the IQ scores of adopted children and their real parents, while correlations with their foster parents tend to be nonsignificant. However, he does not mention the fact that adopted children consistently display a substantially higher IQ than their biological parents. Skodak

and Skeels (1949) found that the average IQ of the real mothers was 86 while that of their children adopted into other families was 106—well over a whole standard deviation higher. Surely i this indicates that, with an improved socioeconomic background, one can accomplish in one generation a change that is greater than any difference between racial or religious groups in the United States. The overwhelming component of this difference is certainly environmental. In their review of behavior genetics, Spuhler and Lindzey (1967) come to much the same conclusion with regard to racial differences in IQ. While citing many cases of behavioral differences

among humans which have a known genetic basis, they show that there is a very significant _ relationship between IQ and educational expenditure. They conclude (1967:405): that ‘‘we do not

know whether there are significant differences between races in the kinds and frequencies of polygenes controlling general intellectual ability.” In our turn, we do not see how anyone would disagree with this statement, but would go further. We suggest that it is possible to explain all the

measured differences among major groups of men primarily by environmental factors, while noting, on the other hand, that it is not possible to provide genetic explanations which are evolutionarily plausible for most of these differences.

: Within a given population there certainly is a spectrum of the inherited component of “intelligence,” and there may be some association between this and certain demanding occupations, but from the perspective of biological evolution, the time depth of the professions in

question is so shallow that little change in the genetic structure of the population can have - occurred. Furthermore, Jensen’s reaffirmation of the time-honored assumption that there are average differences in innate intelligence between social classes is also without demonstrable foundation and is very probably incorrect. At the top end of the social scale in America, the initial

establishment of position may have had some relationship to ability, although demonstrable unprincipled ruthlessness was at least as important (Lundberg 1968). Once established, position is retained with little relation to the continuing presence of ability in the families in question and reproductive behavior is notoriously unrelated to the intellectual attributes of the partners chosen. At the bottom of the social hierarchy there is one outstanding factor that makes suspect any

claims concerning inherited ability. This factor is poverty. It is not unexpected that “in most settings there is a positive association between poor nutrition and poor social conditions” (Richardson 1968:355). And if this itself does not assure retardation in the development of mental ability, an atmosphere of social impoverishment certainly does. Inculcation into the ways of ‘‘the

culture of poverty” (Lewis 1966) does not train people to perform well on IQ tests. Nor has ability or its lack had much to do with recruitment into the ranks of the extremely poor. Mere possession of a black skin was sufficient until quite recently and, with the addition of certain

geographic provisos, still is.

One of Jensen’s basic assumptions is made explicit in the comment printed with his approval in The New York Times Magazine (Sept. 21, 1969, p. 14). In this he clearly regards “‘intelligence as

the ability to adapt to civilization,’ adding that “races differ in this ability according to the civilizations in which they live.’’ Building on this, he further assumes that “‘the Stanford-Binet IQ

test measures the ability to adapt to Western civilization,” an ability in which he claims American Negroes to be inferior to “Orientals,” with the clear implication that, as a blanket category, they are far less well-endowed than American whites. For an educated man to hold such beliefs is regrettable, but for a presumed “scientist” to be allowed to publish them in a popular journal without informed editorial supervision is an example of the unfortunate failure of intellectual responsibility. First of all, “Western civilization,” if this is indeed a valid category in this context, is largely a

product of the Industrial Revolution and has a maximum time depth of little more than two centuries. Even if natural selection had been operating at maximum efficiency during this time, it “would have been hard put to change a polygenic trait as much as a full standard deviation for an

70 | RACE AND INTELLIGENCE [AS8 entire population. In terms of actual reproductive performance, there is little reason to believe that the intellectually highly endowed were in fact favored to such an extent. If it is fair to make such

sweeping judgments, we can make a case for the fact that most of the labor roles which were _

created by and ensured the success of the Industrial Revolution—and, hence, Western : civilization—required relatively little learning and no creative decision-making on the part of their occupants. This, of course, is why child labor was practical until it was outlawed. In terms of the kind of folk knowledge and unwritten tradition necessary for survival, it is perhaps fair to claim

that the average European (i.e., peasant) of the sixteenth century lived a life that had more elements of similarity with that of the average West African than it did with that of the descendents of either one in the Europe or America of the twentieth century. Obviously in saying this we are making a value judgment that cannot be proven one way or the other, but, nevertheless, it would appear to square with the data of both anthropology and history _

rather better than Jensen’s suggestion that races differ in intelligence “according to the civilizations in which they live.” Considering the fact that, with a few numerically unimportant

exceptions, all human populations now live under conditions characterized by cultural adaptations—“‘civilizations”’ in Jensen’s terms—that are radically different from those of their lineal predecessors only a few thousand years ago (and often much less), it is reasonable to conclude that

no races are really adapted to the “civilizations in which they live.” | :

The time is not so long past when instructing Negroes in the mechanics of reading and writing was contrary to law in parts of the American South. Educational opportunities remain drastically

substandard, and there is scarcely a rudimentary form of the tradition in child-rearing, so characteristic of the middle and upper-middle classes, which promotes literacy as the key to wordly success. When used to compare groups with different cultural backgrounds, the Stanford-Binet IQ test is less a comparative measure of ability than an index of enculturation into the ways of the American middle class. Since Negroes have been systematically (see the account in Woodward 1966) denied entrance to the middle-class world, it is not surprising that their learned

behavior is measurably different from that on which the IQ test is based. Certainly before the results of IQ tests can be taken as indicating inherited differences in ability, some cognizance

should be taken of the effect of tester expectation on performance (Rosenthal 1966), or motivation in its various aspects (Katz 1967), and of the results of nonverbal tests where conceptual styles of the groups being studied are markedly different (Cohen 1969). Finally, Jensen’s assertion that intelligence or brain differences must exist among the “‘races”’ of

man is an argument by analogy which ends up assuming what he presumably was trying to demonstrate. As he notes, separate breeding isolates will very likely show differences on some genetic characteristics which will be due to the various evolutionary forces. In most cases, these differences, if at all considerable, will coincide with differences in selective forces to which the populations are subject. To conclude from this perfectly reasonable genetic statement, as Jensen does, that it is “practically axiomatic’ that two populations will be different in any characteristic

having high heritability (1969a:80) and, ergo, that the races of man differ in their genetic capacities for intelligence or in the genetic properties of their brains is simply a non sequitur. Certainly it is contrary to all we have learned from evolutionary biology. All human populations have 10 fingers, 10 toes, 2 eyes, and 32 teeth per individual. These all have high heritability and . some variability within human isolates, but are constant between isolates. This is primarily due to the operation of natural selection, a factor which Jensen deemphasizes. Behavior or brain function is obviously under the control of many loci, and, equally obviously, it is subject to the influence of natural selection. If differences exist at these loci among human populations, these differences would be correlated with differences in the forces of selection. These in turn would be reflected in the cultural and behavioral attributes designed to counteract _ them. Within any continent there are as many differences in cultural and behavioral adaptation as there are between continents. However, implicit and even explicit in much of behavior genetics is the assumption that cultural differences are caused by genetic differences. The anthropological findings that cultural differences represent responses to varying environmentally imposed selective forces are simply ignored. Selective force distributions do not neatly coincide as a rule. Some may covary in some areas,

Brace & Livingstone] ON CREEPING JENSENISM 71 some may show crosscutting distributions, and others may vary completely at random with respect to each other. Given this situation, we have elsewhere suggested that, in order to make sense out of human biological variation, the typological gestalt of the race concept be abandoned and human adaptation be studied trait by trait in the contexts where the relevant selective forces have been at work (Livingstone 1962, 1964; Brace 1964a, b). We cannot resist adding the comment that this

approach, if taken seriously, can completely defuse the potentially explosive situation which

Jensen has created. .

.- Jensen (1969a:89) cites the Harlows (Harlow and Harlow 1962) to the effect that if the average - IQ were lower and thus fewer geniuses were produced, then there would be fewer people to make inventions and discoveries and thus cultural evolution would have been slower. We suggest that, , just as mutation rate does not control the speed of biological-evolutionary change, neither does the frequency of the occurrence of genius have anything to do with the rate of cultural evolution. We can even offer a converse suggestion and raise the suspicion that levels of cultural complexity are inversely related to IQ. Survival takes less innate wit for the socially and economically privileged than it does for those to whom culture does not offer ready-made solutions for most of life’s problems. It is possible that the average level of intelligence is highest among populations where culture is least complex. Post-Pleistocene food preparation techniques, including, especially, pots in which boiling was easy and common, have rendered the human dentition of far less importance to survival than before. The sharp reductions in the Post-Pleistocene human face are concentrated in the dentition and have proceeded farthest in just those people whose forebears have been longest associated with “high civilization” (Brabant and Twiesselmann 1964:55). Is it not possible that supraregional political and economic organization increased the survival chances of any given individual without regard for his inherited ability? Why then should we not expect an average lowering of basic intelligence to accumulate under such circumstances? We offer this solely as an hypothesis for possible testing. Jensen, on the other hand, feels that failure to test the hypothesis that Negroes are intellectually inferior for genetic reasons may constitute “‘our society’s greatest injustice to Negro Americans” (1969c:6). Unless there is a latent racist bias to the kind of research Jensen feels is urgent,° it is difficult to see why the testing of the hypothesis we have outlined above is not considered at least of equal importance with the testing of its converse, and yet the

possibility is not even mentioned, let alone seriously entertained. Ironically, the possible consequences of our failure to take this issue seriously will be enormous for all Americans, but particularly for non-Negro Americans, i.e., “whites.” Knowledge of both cultural and biological dimensions is required for a full understanding of the human condition. The stress on ‘“‘geneticism”’ (to use Medawar’s word again) should be tempered by an insistence that the environmental component be thoroughly controlled. Certainly as much

of each. .

effort and sophistication should be devoted to this task as to comparative performance assessments. Jensen’s work is conspicuously lacking in this regard. As such, it is the logical antithesis to the old environmentalist thesis. Perhaps the synthesis will contain the reasonable parts

Finally, in the words of Jensen (1969:78), “If a society completely believed and practiced the ideal of treating every person as an individual, it would be hard to see why there should be any problems about “‘race’’ per se.” Unfortunately Jensen ignores this ideal in practice and, in the absence of adequate control, insists on treating a substantial portion of the American population as though a stereotype were sufficient and as though the individual could be ignored. In effect this . guarantees that there will continue to be problems about race per se and that Jensen and his like

will only intensify them. _ NOTES

1Since the preparation of this manuscript (for the November 1969 Annual Meeting of the American Anthropological. Association), another paper by Jensen has appeared (1969b) in conjunction with more thoroughly documented critiques in the summer issue of the Harvard Educational Review. Some of the points we raise are discussed in greater detail than in our ‘presentation, but since other important ones are not even mentioned we have decided to let our paper stand substantially as originally written.

72 RACE AND INTELLIGENCE [AS8 *For an elegant demonstration of the inappropriateness of Jensen’s use of twin data, see the critique by Light and Smith (1969). Fehr (1969) also shows the inaccuracy of Jensen’s use of twin ~ ‘data to arrive at his assumed heritability level. 3 Jensen (1969b) offers a rebuttal to the somewhat anecdotal accounts of Kagan and Hunt but, again, uses a single debatable account to generalize for all mankind. 4 This criticism of Jensen’s approach has been eloquently and forcefully made by Stinchcombe

(1969) and Deutsch (1969).

° The interaction of heredity and environment in the development of a trait has been brought into focus by Stinchcombe (1969), but an even more important point is made by Gregg and Sanday in their contribution to this present volume. They note that heritability figures for given traits will vary in inverse proportion to the similarity of the environments of the populations being : considered. This illustrates the generalization offered some time ago by Lerner (1958:63, italics his): ““The heritability of a given trait may differ from one population to another, or vary in the.

same population at different times ... strictly speaking, any intra-generation estimate of ~~

heritability is valid only for the particular generation of the specific population from which the |

populations and not of traits.’’

data used in arriving at it derive.”’ As Hirsch (1969:138) has phrased it, ‘“‘Heredity is a property of

6 The regretful comment made by a collaborator and admirer of Jensen’s experimental research is worth quoting here: “I believe the impact of Jensen’s article was destructive; that it has had negative implications for the struggle against racism and for the improvement of the educational system. The conclusions he draws are, I believe, unwarranted by the existing data and reflect a

consistent bias towards a racist hypothesis’ (Deutsch 1969:525). REFERENCES CITED

Bajema, C. J. .

Ardrey, R.

1966 The Territorial Imperative. New York: Delta. )

1963 Estimation of the Direction and Intensity of Natural Selection in Relation to Human Intelligence by Means of the Intrinsic Rate of Natural Increase. Eugenics Quarterly

10:175-187. Barzun, Jacques

Birch, Herbert G. Br 1965 Race: A Study in Superstition. New York: Harper Torchbooks.

7:11-84. Brace, C. L. ,| Brown, Roy E.

1968 Boldness and Judgment in Behavior Genetics. In Science and the Concept of Race. M. Mead, T. Dobzhansky, E. Tobach, and R. Light, eds. New York: Columbia University Press. Brabant, H., and F. Twiesselmann 1964 Observations sur l’Evolution de la Denture permanente humain en Europe Occidentale. Bulletin du Groupement International pour la Recherche scientifique en Stomatologie

1964a The Concept or Race. Current Anthropology 5:313-320. | 1964b A Non-Racial Approach Toward the Understanding of Human Diversity. In The Concept of Race. M. F. A. Montagu, ed. New York: Free Press.

1965 Decreased Brain Weight in Malnutrition and Its Implications. East Africa Medical Journal 42,:584-595. 1966 Organ Weight in Malnutrition with Special Reference to Brain Weight. Developmental Medicine and Child Neurology 8:512-522.

Chamla, M-C. ce

1964 L’accroisement de la Stature en France de 1800 a 1960; Comparison avec les pays d’Europe occidentale. Bulletins et Mémoires de la Société d’Anthropologie de Paris, Série 11, 6:201-278.

Cohen, Rosalie A. ) 1969 Conceptual Styles, Culture Conflict, and Nonverbal Tests of Intelligence. American Anthropologist 71:828-856. Coles, Robert

1969 Still Hungry in America. Cleveland: North American Library. .

Cravioto, J., E. R. DeLicardie, and H. G. Birch 1966 Nutrition, Growth and Neuro-Integrative Development: An Experimental and Ecologic Study. Pediatrics 38:319-372.

Cravioto, J., and B. Robles

1965 Evolution of Adaptive and Motor Behavior During Rehabilitation from Kwashiorkor.

American Journal of Orthopsychiatry 35:449-464. | ;

Brace & Livingstone] ON CREEPING JENSENISM 73 Crow, James F.., James V. Neel, and Curt Stern

158 :892-893. .

1967 Racial Studies: Academy States Position on~Call for New Research. Science

Davison, A. N., and J. Dobbing ; 1968 Myelination as a Vulnerable Period in Brain Development. British Medical Bulletin

Dickerson, J. W. T. | _ Dobbing, J. / 22:40-44.

1968 The Relation of the Timing and Severity of Undernutrition to Its Effect on the Chemical _ Structure of the Central Nervous System. Jn Calorie Deficiencies and Protein Deficiencies. R.

a A. McCance and E. M. Widdowson, eds. London: J. A. Churchill.

MIT Press. |

1964 The Influence of Nutrition on the Development and Myelination of the Brain. Proceedings of the Royal Society of London, Series B, Biological Sciences 159 :503-509.

1968 Effects of Experimental Undernutrition on Development of the Nervous System. Jn Malnutrition, Learning and Behavior. N. 8. Scrimshaw and J. E. Gordon, eds. Cambridge: Dobzhansky, T., and B. Spassky 1969 Artificial and Natural Selection for Two Behavioral Traits in Drosophila pseudoobscura. Proceedings of the National Academy of Sciences 62:75-80.

Edson, Lee |

Gans, Bruno 1969 jensenism, n. the theory that IQ is largely determined by the genes. The New York

Times Magazine, August 31, pp. 10-11, 40-41 43-47.

HKichenwald, Heinz F., and Peggy Crooke Fry

1969 Nutrition and Learning. Science 163:644-648.

1969 A Biafran Relief Mission. The Lancet 1969-I:660-665. Graham, G. G.

1968 The Later Growth of Malnourished Infants: Effects of Age, Severity and Subsequent

Diet. In Calorie Deficiencies and Protein Deficiencies. R. A. McCance and E. M. Widdowson, eds. London: J. A. Churchill. Hirsch, Jerry 1961 Genetics of Mental Disease Symposium, 1960: Discussion: The Role of Assumptions in the Analysis and Interpretation of Data. American Journal of Orthopsychiatry 31:474-480. 1963 Behavior Genetics and Individuality Understood: Behaviorism’s Counterfactual Dogma Blinded the Behavioral Sciences to the Significance of Meiosis. Science 142:1436-1442, 1967a ed. Behavior Genetic Analysis. New York: McGraw-Hill. 1967b Behavior-Genetic, or ‘‘Experimental,’’ Analysis: The Challenge of Science Versus the Lure of Technology. American Psychologist 22:118-130. 1968 Behavior-Genetic Analysis and the Study of Man. In Science and the Concept of Race. M. Mead, T. Dobzhansky, E. Tobach, and R. Light, eds. New York: Columbia University Press.

Hollings, Ernest F., as told by Paul Jablow .

1970 We Must Wipe out Hunger in America. Good Housekeeping, January, pp. 68-69,

Javits, Jacob : 144-146.

Hunt, J. MeV.

1969 Has Compensatory Education Failed? Has It Been Attempted? Harvard Educational Review 39:278-300.

1969 Hunger in America. Playboy, December, p. 147. Jensen, Arthur R. 1969a How Much Can We Boost IQ and Scholastic Achievement? Harvard Educational Review 39:1-123. 1969c Arthur Jensen Replies. Psychology Today 3:4, 6.

Jordan, Winthrop D. ° 1968 White over Black: American Attitudes Toward the Negro, 1550-1812. Chapel Hill:

University of North Carolina Press. 7

Kagan, Jerome S.

1969 Inadequate Evidence and Illogical Conclusions. Harvard Educational Review 39: 274-277.

Kagan, J. S., and H. A. Moss 1959 Parental Correlates of Child’s IQ and Height: A Cross-Validation of the Berkeley Growth Study Results. Child Development 30:325-332. Katz, Irwin

_ 1967 Some Motivational Determinants of Racial Differences in Intellectual Achievement. International Journal of Psychology 2:1-12.

74 RACE AND INTELLIGENCE [AS8 Kimura, K.

1967 A Consideration of the Secular Trend in Japanese for Height and Weight by a Graphic

Method. American Journal of Physical Anthropology 27:89-94. :

Lewis, Oscar

1966 The Culture of Poverty. Scientific American 215:19-25.

Livingstone, Frank B. :

10:45-61. ce Lundberg, Ferdinand | Stuart. | .

1962 On the Nonexistence of Human Races. Current Anthropology 3:279-281. 1964 On the Nonexistence of Human Races. In The Concept of Race. Ashley Montagu, ed.

New York: Free Press of Glencoe.

1969 Genetics, Ecology and the Origins of Incest and Exogamy. Current Anthropology

Lorenz, Konrad

1966 On Aggression. New York: Bantam. . 1968 The Rich and the Super-Rich: A Study in the Power of Money Today. New York: Lyle

Medawar, P. B.

1961 The Future of Man. New York: Mentor.

Morris, Desmond

1968 The Naked Ape. New York: McGraw-Hill. 1969 The Human Zoo. New York: McGraw-Hill.

Osofsky, Gilbert

New York:Carleton Harper and Row. | Putnam, Richmond, R. C. -

1967 The Burden of Race: A Documentary History of Negro-White Relations in America.

MIT Press. | |

1961 Raee and Reason: A Yankee View. Washington: Public Affairs Press. Richardson, Stephen A.

1968 The Influence of Social-Environmental and Nutritional Factors on Mental Ability. In Malnutrition, Learning and Behavior. N. S. Scrimshaw and J. BE. Gordon, eds. Cambridge:

1969 Heritability of Phototactic and Geotactic Responses in Drosophila pseudoobscura.

American Naturalist 103:315-316. Roberts, D. F. 1967 Incest, Inbreeding and Mental Abilities. British Medical Journal 4:336-337. Rosenthal, Robert

1966 Experimenter Effects in Behavioral Research. New York: Appleton-Century-Crofts. ,

Scrimshaw, Nevin S., and John E. Gordon (eds. ) 1968 Malnutrition, Learning and Behavior. Cambridge: MIT Press. Skodak, M., and H. M. Skeels

1949 A Final Follow-up Study of One Hundred Adopted Children. Journal of Genetic

Psychology 75:85-125. ,

Smith, Samuel Stanhope

1965 An Essay on the Causes of Variety of Complexion and Figure in the Human Species (reprint of the 1810 version). Winthrop D. Jordan, ed. Cambridge: The Belknap Press.

York: McGraw-Hill. 7

Spuhler, J. N., and G. Lindzey

1967 Racial Differences in Behavior. In Behavior-Genetic Analysis. Jerry Hirsch, ed. New

Stanton, William

1960 The Leopard’s Spots: Scientific Attitudes Toward Race in America, 1815-59. Chicago:

University of Chicago Press. Stoch, Mavis B., and P. M. Smythe

1963 Does Undernutrition During Infancy Inhibit Brain Growth and Subsequent Intellectual Development? Archives of Diseases of Childhood 38:546-552.

1968 Undernutrition During Infancy, and Subsequent Brain Growth and Intellectual

Development. In Malnutrition, Learning and Behavior. N. S. Scrimshaw and J. E. Gordon, eds. Cambridge: MIT Press. | |

Stocking, George W., Jr.

1968 Race, Culture and Evolution: Essays in the History of Anthropology. New York: Free Press.

Takahashi, E. 1966 Growth and Environmental Factors in Japan. Human Biology 38:112-130. Tiger, Lionel 1969 Men in Groups. New York: Random House. Watson, J. B.

1924 Behaviorism. New York: W. W. Norton. .

Brace & Livingstone] ON CREEPING JENSENISM 75 Winick, Myron, and Pedro Rosso 1969 Head Circumference and Cellular Growth of the Brain in Normal and Marasmic Children. The Journal of Pediatrics 74:774-778. Woodward, C. Vann 1966 The Strange Career of Jim Crow. New York: Galaxy Book. SUPPLEMENTARY REFERENCES Deutsch, Martin

1969 Happenings on the Way Back to the Forum: Social Science, IQ, and Race Differences

Revisited. Harvard Educational Review 39:523-557. Fehr, F. S. 1969 Critique of Hereditarian Accounts of “Intelligence”? and Contrary Findings: A Reply to Jensen. Harvard Educational Review 39:571-580. Harlow, H. F., and M. K. Harlow. 1962 The Mind of Man. In Yearbook of Science and Technology, pp. 31-39. Hirsch, Jerry

1969 Biosocial Hybrid Vigor Sought, Babel Discovered. Review of Genetics: Second of a Series on Biology and Behavior, edited by David C. Glass. Contemporary Psychology

14:138-139. Jensen, Arthur R.

1969b Reducing the Heredity-Environment Uncertainty: A Reply. Harvard Educational

Review 39:449-483. Lerner, I. Michael 1958 The Genetic Basis of Selection. New York: John Wiley. Light, Richard J., and Paul V. Smith.

1969 Social Allocation Models of Intelligence: A Methodological Inquiry. Harvard

Educational Review 39:484-510. Stinchcombe, Arthur L. 1969 Environment: The Cumulation of Effects Is Yet to be Understood. Harvard Educational Review 39:511-522.