Lawler, J.M. (1978). "Heritability" (pp. 133-158). In I.Q. Heritability, and Racism. New York: International Publishers.

Heritability

The history of IQ testing reveals the attempt to create an instrument that would measure innate and fixed intelligence. We have devoted the major part of this essay to an examination of the presuppositions of IQ ideology in order to explain how the semblance of fixed and innate intelligence has been derived primarily from the method of selecting test items and of interpreting scores. This effort is necessary for an understanding of just what IQ tests measure. In one respect, however, this effort may seem unnecessary. Jensen differs from his predecessors, Galton, Terman and others in the history of the biological theory of intelligence, in that he does not appear to assume the validity of IQ tests as a measure of innate intelligence. On the contrary, he pretends to prove this by use of the methods of 'population genetics' which were unavailable in the early days of IQ construction. Jensen pretends to have effected a theoretical revolution in giving a genetic interpretation of intelligence and in challenging the reigning "environmentalist" interpretations.

From innate IQ to IQ heritability

In fact the biological interpretation of intelligence not only was assumed from the beginning of the effort to measure intellectual differences, but has permeated the method of IQ construction. Nevertheless this biological interpretation has been challenged from various scientific directions. The rejection of eugenic theories and the biological interpretation of IQ scores grew with the revulsion against fascist thought in the 1930s and with the war against Nazi aggression in the 1940s. The newly formed United Nations sponsored international research to expose the fallacious and fundamentally vicious character of biological and racist theories of human society. It is a tribute to this collective effort of natural and social scientists, as well as to the general public rejection of Nazi racist theories and the untold destruction of lives which such theories justified, that [p.134] the biological interpretation of IQ tests can no longer be assumed. This is not to say that this belief does not still have roots in popular opinion as well as in theory, as we have noted in the beginning of this essay. There is still a widespread belief in IQ as a measure of one's basic capacity to think. But contrary beliefs prevent this tendency from resulting in the kind of conscious racist and nationalist attitudes that were deliberately cultivated in the twenties and thirties, and which were effectively challenged by the civil rights movement in the U.S. South during the late fifties and throughout the sixties.

Jensen's article in 1969 is aimed at reviving a biological interpretation of intelligence and of giving academic and scientific credence to the racist theory of the intellectual inferiority of Black Americans. For this revival, however, it was not sufficient for Jensen to report that whites score higher than Blacks on IQ tests. The mystique of IQ as a valid measure of innate capacity has been sufficiently challenged so as to make it necessary to prove validity on grounds seemingly outside of the IQ itself. Thus, in addition to reporting racial differences in IQ tests, Jensen adds that by using methods of "population genetics" it can be shown that IQ is 80% "heritable." From there, Jensen marshals a number of arguments which render "plausible" the "hypothesis" that the main explanation of the differences between IQ scores for Blacks and whites is to be sought in genetic differences.

The appeal to heritability analysis of IQ scores is one further step in the attempt to validate IQ as a measure of innate and fixed intelligence. It is another attempt to find some criterion outside of the operational definition of IQ which would demonstrated that it measures "what it is supposed to measure." And yet, as with other attempts of this sort, it rests on a priori presuppositions that need to be critically evaluated. The presuppositions of heritability analysis [p.135] are in fact very similar to those that we have examined in our study of IQ.

One obvious presupposition of the "heritability of IQ" is that it must begin with IQ scores. If IQ scores cannot be relied upon as measures of "general intelligence" then it follows that the "heritability" of IQ cannot be something meaningful. We think that our previous analysis has shown the essentially artificial and ideological character of IQ theory -especially as metaphysically and idealistically interpreted by Jensen. Nevertheless, if some further evidence is presented to validate IQ as a measure of innate intelligence, we should have to reconsider our earlier interpretations. To what extent, then, has "population genetics" come to the rescue of the much beleaguered IQ theory?

Heritability and heredity

The assertion that IQ is 80% heritable gives the impression that a large part of intelligence is inherited and owes relatively little to the "environmental" circumstances in which we have grown up, and there is not too much that can be done about changing the intellectual abilities that were confered upon us at conception. In other words, the assertion is taken to mean that "intelligence" is mostly, but not absolutely, innate and fixed for life. Thus the expression appears to reassert, with somewhat more modesty, the main theses of IQ ideology.

On closer inspection, however, the concept of "heritability" has a particular meaning in population genetics which fails to support this "common sense" interpretation. The "heritability" of a trait does not in fact mean how much of some trait (the "phenotype") is due to the genes (the "genotype"), as opposed to the environment. Nor does it refer to something that is necessarily fixed; the "heritability" of a trait can change drastically. In fact the common sense interpretation and the technical usage operate on completely different planes. As we will see, Jensen is perfectly aware of the technical usage of the concept of heritability and of the incorrectness of the above "common sense" interpretation. Jensen displays sufficient expertise to convince biologists that he knows what heritability in fact means -while at the same time the main thrust of his argument is to reinforce the "common sense" impression that it [p.136] is the measure of "how much of intelligence is determined by the genes." There is a sleight-of-hand, essentially similar to the one played in IQ theory, which confuses a technical expression, which has a peculiar relative meaning, with an intuitive belief that attributes an absolute, metaphysical meaning to the concept. The following examples will help to illustrate the use of the notion of heritability in biological theory.

Consider a situation in which there are two completely inbred lines of corn, line A and line B. In each line all of the seeds are genetically identical with one another (have the same genotype), but there is a genotype difference between the two lines. The seeds are planted in ordinary potting soil which gives a variety of different conditions for development. Since we know that in each line there is genetic identity, we know that the differences or variations in the height (the "phenotype"), say, of the corn in line A is due to differences in the environment of line A. Observed variations in the height of line B are entirely environmental in origin. Because none of the observed phenotypic differences (i.e., differences in height) in each line is due to differences in the genotype of the line, the "heritability" of each line is said to be 0%. Heritability is defined as the proportion of the total variation of a certain trait in a population which is due to genetic variation. It is the ratio of the genotypic "variance" over the total phenotypic "variance." In this example we know that there are no differences between the genes within each line. The technical formulation for variation is called the "variance" and is derived by a method which will remind us of IQ. We will return to this shortly. Since there is no genetic variation or differences, our formula for the heritability (H) of each line will be 0/t ("t" being the total variance) or 0%. The total variance is defined as the sum of the genetic variance and the environmental variance (t=g+e). We will return to this general formula for closer scrutiny.

While the heritability of both lines A and B is 0%, lines A and B are nevertheless genetically different from each other. Suppose that we observe a difference between the average height of line A and that of line B. We would probably be safe to assume that the cause of this difference between lines is due to the genetic difference between A and B. This leads to the paradoxical conclusion that while the "heritability" of each line is 0, the explanation of the differences between the lines is entirely genetic. Thus, there is no logical connection [p.137] between the heritability figure for a given "population" and the cause of the real differences between two populations. In this it is also similar to IQ, which is entirely relative to the population on which it is standardized. This point is important on technical grounds because Jensen, while aware of the technical limitations of heritability, attempts to explain the difference between Black and white populations from the heritability figure that is said to obtain within each population. T. Dobzhensky points out that contradiction in Jensen's work between his awareness of the technical limitations of heritability and the conclusions he attempts to draw which overstep and even contradict these limitations:

Jensen (1969), after recognizing explicitly that the heritability of individual differences within a population cannot validly be used as a measure of the heritability of the population means [of different races or classes], tries to do just that [1973, p. 21].

Jensen seems to try to justify such contradictions as part of the creative effort of a pioneer in a field:

Though I always heeded expert advice on purely factual and technical matters, I usually kept my own counsel on matters of interpretation and judgment.... There are always differences among investigators working on the frontiers of a field. They differ in their weighing of items of evidence, in the range of facts in which an underlying consistency is perceived, in the degree of caution with which they will try to avoid possible criticisms of their opinions, and in the thinness of the ice upon which they are willing to skate in hopes of glimpsing seemingly remote phenomena and relationships among lines of evidence which might otherwise go unnoticed as grist for new hypotheses and further investigations.

Another experiment will bring out more clearly the peculiar "relativity" of the concept of heritability, as well as the thinness of the ice on which Jensen is skating when he uses this concept to boost IQ theory. Suppose that the seed is taken from a sack of an "open pollinated" variety of corn. Here there is a great deal of genetic variation. This time, however, we plant quantities of this seed in two carefully controlled environments, each one containing within it chemically identical soil and nutrients. But environment B has exactly half of the amount of nitrates as in environment A. In [p.138] addition, zinc, an important trace element, is removed from the soil of environment B. Because the soil in environment A is perfectly uniform (or as close to uniformity as is possible) all of the variations we observe when the corn reaches maturity will be due to the genetic variations between the corn. The same is true of the variations we observe in environment B. Applying our heritability formula, g/g+e (where g is the genetic variance and e is the environmental variance) we arrive at a heritability of the corn in environment A which is 100%. Could we conclude from this something about the characteristics of the corn from the same batch in environment B? According to the common sense, metaphysical interpretation, 100% heritability means that all of the characteristic (such as height) is determined by heredity, nothing by environment. It would seem to follow from this that wherever we plant such highly heritable corn we would get the same results. The property should be entirely determined and fixed by the genes. In fact, of course, the corn in environment B would be quite different from the corn in environment A, despite the fact that in both cases the heritability of the corn is 100%. Paradoxically, this time the cause of the difference between the average heights observed in the two environments is entirely environmental.

From this example it is clear that the heritability of a trait is relative to a given environment. It is not an estimate of the portion of that trait that is due to heredity alone, as opposed to environment. The metaphysical distinction between heredity and environment, which conceives of these as separate and juxtaposed entities, only wreacks havoc with our attempt to understand heritability. Since a heritability estimate is relative to a given environment it says nothing about what will become of the same individuals under environmental conditions different from the one(s) in which heritability has been estimated; it says nothing about the possibilities of change under other circumstances.

More fundamentally, in our opinion, it is important to understand that heritability tells us nothing about the particular characteristic itself under any circumstance. A heritability figure tells us only something about the variations that occur of that particular trait in a "population"; it tells us only to what extent the variations are due to genetic variations (and even this is done in a very limited way, as will be clear later). The confusion that easily arises in attempting to understand what such a statement means stems from [p.139] a natural tendency to treat a variation or a difference as though it were a definite entity. This illusion results in part from the fact that differences are expressed in terms of definite numerical quantities. Heritability theory, including the technical method for establishing variance, attempts to treat variation as something in itself, irrespective of the real features of the thing that varies. Thus, in our previous example, 100% heritability told us nothing about the quality of our two samples of corn. The fact that we could measure the real height of the corn in each group enabled us to note that one group of corn was taller than another. This observation rested on the fact that we could tell how tall each stalk of corn was in itself. Heritability analysis, however, tells us nothing about the real features of the characteristic we are measuring. It deals strictly with observed differences, measured in relation to the average. Where there is no variation in the particular trait, even though this be genetically necessary, the heritability of the trait would be zero. (Consider, for example, the heritability of "one-handedness" among humans. Since there is no variation of this trait in a population, its "heritability" is 0, although it is completely caused by "heredity.") This abstraction of variation as a seemingly independent reality is something that heritability methods have in common both with IQ and with statistical "intercorrelations."

Jensen's "sociology"

A final look at our second experiment will bring out the degree of caution that must be exercised in determining heritability estimates under conditions which are not absolutely controlled. In his [1976] article from which this example is borrowed, Richard Lewontin asks us to suppose that the composition of the soil in environment B is not known to the investigator. What is observed is a lower average height in the corn grown in this soil. Upon analysis of the composition of the soil, the experimenter discovers the absence of nitrates, but fails to detect the missing zinc trace. He then repeats the experiment after doubling the amount of nitrates in soil B, and finds that, although the corn now reaches a higher average growth, it is still not equal to the corn in soil A. Where a Jensen might skate out on thin ice and proclaim that the cause of the differences is probably genetic, a closer study would reveal the inaccuracy of this conclusion. [p.140]

This example is meant to show how much caution must be applied to the far more complex and less perfectly understood case of human development. But here we do not want to plead ignorance. Jensen makes due acknowledgment of possible future discoveries that might explain Black-white IQ differences environmentally. However, where he pretends to take account of environmental differences to explain IQ performance, taking differences in gross socioeconomic status into account, he fails to consider an element that is known to everyone -the four hundred year history of racism and the special oppression of Black people in the U.S. It is not a matter of a scientist who is ignorant of a difficult to detect trace element, but of an assumption that a flagrant and absolutely central fact of U.S. history can be treated as negligible in explaining differences in IQ or academic performance.

Heritability analysis as applied to human development assumes that it is possible to equalize environment by the use of crude socioeconomic indicators. Thus, if all the individuals in a particular school come from roughly the same kind of working class environment, the differences in intellectual performance would be "explained" by the genetic factor that is left once we have crudely dismissed the environment as an explanation of differences on the grounds that all of the families have roughly the same income, educational background, housing facilities, etc. This concept is stretched to the limit in the meritocracy theory which assumes that since environmental opportunities become equalized for everyone in the U.S., the observed differences are increasingly due to genetic differences. The leap from our chemically controlled soil for the growth of corn to equality of opportunity for the growth of our children may leave some of us gasping for air.

We wonder what sociologists Jensen has been reading in his effort to explain differences in IQ who omit among their "socioeconomic indices" the special character of racism against Blacks in a country that was built in large part on the slave exploitation of Black people who were both systematically terrorized and consciously kept in a state of illiteracy. And enormous superprofits are still made through the substantial wage differential that continues to be maintained from doubly and triply exploited Black and other minority workers. A study of the history of the struggle of Black Americans shows that fear and ignorance were the tools of their [p.141] brutal domination -not natural attributes of an inferior race. Roy Brown thinks that Jensen's "tentative conclusion" that

'Heredity... plays some role in the heavy representation of Negroes in America's lower socioeconomic groups' [is] unbelievable when one considers the fact that absolutely nothing is said about the extreme deprivation that blacks have endured -300 years of the cruelest slavery known to mankind: 100 years of barbaric servitude, murder, lynching, burning, and intimidation superimposed with an arrogant savage con game. There was literally no intention of treating blacks as human beings, but rather, they were to be exploited and kept in servitude by any and all means, legal and illegal.

Lack of real consideration of this central fact of U.S. history, as well as the deep-rooted, inhumane and devastating character of continuing racial oppression, shows how feeble has been Jensen's attempt to find "environmental" explanations for IQ differences understood not as indicators of innate capacity but as indirect signs of suppressed abilities. The fundamental role that racism plays in the United States cannot be reduced, in vulgar materialist fashion, to "socioeconomic indicators" that lump together social and national groups with qualitatively different historical characteristics. Despite equally savage treatment, the history of Native Americans cannot be equated with that of Blacks, to say nothing of Taiwanese immigrants (whose higher scores and low socioeconomic status gives Jensen cause for further genetic hypothesizing). The national character and forms of oppression of the peoples mentioned by Jensen cannot be reduced to a scale of gross living standards, however significant these might be.

Historical materialism is not disproved by the lack of full explanatory value of the crude sociological categories with which Jensen identifies the "environmentalist" position. Marx in fact explicitly criticized the mechanistic environmentalist theories of education developed by enlightenment and utopian socialist thinkers who held that "men are product of circumstances" and forget or could not yet recognize that "circumstances are changed precisely by men and that the educator must himself be educated." We have already seen that the concept of an environment as something external to the people who make it stems from the real separation of individuals from the historical conditions of life and thought. [p.142] It is based on a system which "reconnects" the individual with the necessary instruments of human existence only on the condition that this expand capital at the going rate. The limitations of liberal "enlightenment," at the ideological basis of programs such as Headstart, stems in part from a failure to recognize the extent to which economic exploitation and superexploitation continue to stifle the development of the working population, white and Black. But far from drawing anti-enlightenment or anti-environmentalist conclusions from such limitations, Marx deepened the concept of the environment and the materialist theory of knowledge by showing that education is intimately linked with the self-development of the majority of the people struggling for full control over the conditions of their physical and intellectual existence. It was just such a struggle that led to the reforms embodied in the Headstart program, and only the enlargement of such struggle will stop the growing assault on the educational rights of children and adults, both white and Black.

Heritability and widthability

A further example should help to clarify more precisely the limited and essentially pragmatic meaning of heritability, and the inability of this concept and its methods to deal with the real interaction of heredity and environment. This is true both on the strictly biological level and in the broader question of the relation between biological laws of evolution and the specifically social laws of history.

An example from the domain of geometry, reproduced by Howard Toppoff (1974), makes the same point that Lewontin does regarding the heritability of corn. There are particular features of this example, however, which enable us to question more closely the assumptions of heritability estimates. Suppose that there are two sets of football fields, A and B. The lengths of all of the fields of set A are 100 feet while all of those in set B are 50 feet. The widths vary in each set, however. In the first, the widths are 48, 49, 50, 51, and 52 feet; in the second, they are 23, 24, 26, and 27 feet. [p.143]

To determine the "variance" of the width, we find the mean or average width in each set, subtract each individual width from the mean, square these remainders, add them together, and divide by the number of cases. In both sets we arrive at a variance of 2. The variance of the length of course, is 0 -since the length does not vary. If we follow the "additive" model for heritability to determine the "widthability" (the ratio of the variance of the width over the total variance) we will use the following formula: W = Vw/V1 + Vw (V1 = Variance of the length and Vw = variance of the width, while W = "widthability.") Accordingly W = 2/0 + 2 = 1.0. Since there is no variation in the length of any field, the widthability in both sets is 100%.

This example is likely to be somewhat disappointing since although it makes the same point as in the example of the [p.144] heritability of corn, it seems intuitively to be even more trivial. This is true not only because variance in width, as with variance in the height of corn, gives no indication as to the fact of the important and real differences between sets of objects, whether corn or rectangles, but because the general formula itself appears to be completely arbitrary.

What reason is there to define the "total variance" as the sum of the variance of the length and the variance of the width, especially if this is supposed to represent the variance of the "phenotype" or observed phenomenon? In fact, the real "phenotypic variance" is the variance of the areas of the sets of fields, which has not been calculated. To determine the variance of the areas, however, it is not only necessary to have the variance of the widths and the variance of the lengths, it is necessary to have the "absolute measure" of the lengths and widths themselves. We need to multiply length with width in each case, determine the mean area, square the differences from the mean, add them up and divide by the number of cases. The variance of the area in set A is 20,000 sq. ft., and in set B it is 5,000 sq. ft. Assuming now that "total variance" means variance of the areas, our widthability formula changes considerably. W = Vw/V(1xw) = 2/20,000 or .0001 for set A, and 2/1270 or .00158 for group B.

The same operation could be made where the "phenotype" variance was considered to be the variance in the perimeters of the fields. Our formula for W in that case would be W = Vw/V2(1+w). In this case we might double the number of widths, since two widths in fact vary in each case. However, since we should also probably have to double the number of cases in the denominator, Vw would continue to be 2 in both sets. The variance in the denominator would be the same in both sets, this time, however, giving a figure for W in each case of .25.

What is the significance of these calculations? In the first place they call into question the formula for heritability, especially the formula for "total variance," suggesting that this formula cannot be adequately represented by our knowing only the variance of each factor. In addition, we must know the actual dimensions of each factor on an "absolute" scale, and the type of relations which these factors have with each other. In the determination of both area and perimeter these figures are known; in the first dimension the factors have a multiplicative, and in the second an additive relation to each other. [p.145]

Suppose that we knew neither how long each side was, nor what relation each had to the other. All that we were to know was that the width had a variance of 2 and the length had a variance of 0. This is the situation that we find in Jensen's use of population genetics, where we know only variances or "deviation scores" but not absolute values. Jensen writes that heritability deals with "differences among individuals and not with some absolute amounts of some attribute" and that "an absolute scale, though preferable for certain purposes, is nonessential for heritability analyses so long as we think of the phenotype values merely as deviation scores. Nearly all psychological test scores are only deviation scores."

Knowing only the variance of the length and the variance of the width, we in fact know little about the variance of the phenotype, except that it varies in some unknown way not totally unrelated to the variance of the width. We may know that since only the width varies, the "phenotypic" (area or perimeter) variance must be a function of that variance, and not of a variance of the length. We would be entirely wrong, however, to define the phenotypic variance as merely the sum of the variance of the length and variance of the width. This definition is only a result of the poverty of our knowledge -for all we know, by our hypothesis, is that Vw=2 and V1=0. It would be disastrous to multiply these figures or divide them, since the result would be zero and we would end up with nothing at all. It is unlikely that we should think of subtracting the figures since we intuit that they have some kind of "positive" relation to each other. Jensen suggests a functional model, P (phenotypic variance) = f(G,E) (is a function of the variance of the genotype and the variance of the environment). He says that this is potentially valid ("like all models") but is not simple, and besides, the additive model is verified in agricultural genetics. We will return to agricultural genetics. But for the moment it is clear that the functional model is inadequate for perimeter or area as long as we do not know what the functional relation is and what the absolute quantities are whose variances alone are known to us. Thus, in the absence of any other information, we would have to make the best of the formula for phenotypic variance as given in heritability estimates. This may in fact be useful in the practical conditions of agricultural breeding, but then it should not be improperly stated as a definition of the variance of the phenotype, suggesting a deeper theoretical understanding than is in fact warranted. [p.146]

How much or how?

Jensen devotes a section of his book Educability and Group Differences to a criticism of "Heredity, Environment and the question 'How'" by Ann Anastasi. W.F. Overton bases much of his criticism of the "additive model" in the heritability formula on Anastasi, arguing that explanations in biology have moved from questions dealing with "which" factor was determining or "how much" of each, heredity and environment, produced the phenotype, to the question "how" heredity and environment interact to produce the phenotype. Overton argues that between the first two questions and the third there is a fundamental difference in general conceptual frameworks -the additive model versus the "interactive" model.

In defending "additivity," Jensen argues that the developed formula for phenotypic variance includes any interactive or "multiplicative" effect. The elaborated formula for phenotypic variance, according to Jensen, is P = G + E + GE, where "GE represents G x E, i.e., the interactive or multiplicative effects of genetic and environmental factors." Does Jensen's GE account for the "multiplicative effect" as shown in our example of area variance? A comparable effect in that example to what Jensen means by G x E would occur were the variance of the width to produce a change in the variance of the length. Of course this does not happen, and such a "multiplicative effect" is not the kind that takes place when a variation in the width produces a variance in the area because of the relation of the absolute width to the absolute length. Although Jensen writes of G x E, he means that variance of G produces a change in the variance of E.

In fact there can be no "multiplicative" or even "additive" effect such as seen in our examples of area and perimeter variance unless we know the absolute measure for both factors and "how" they in fact interact. Knowing neither, Jensen is right in saying that the "simplest" thing to do is to add the genetic variance, the environmental variance, and any mixed variance where we can't readily distinguish the two. This would not equal the phenotypic variance (outside of simple quantities where such an addition would in fact produce the phenotype variance), but it would somehow affect it. [p.147]

In this state of general ignorance we might hope that the little that we do know might have some practical usefulness. Jensen points to agricultural genetics as "verifying" the additive model. Thus if we know that the egg-laying capacity of chickens is highly heritable, "genetic selection rather than environmental manipulation is likely to yield the most rapid results" in changing the "phenotypic values of the trait." But the "truth" of any pragmatically verified theory is limited to the conditions of the particular practice. The fact that a formula such as the above is "replicable" does not give it full theoretical validity. This is not achieved until we have a causal explanation of how the process takes place by knowing what the genetic activities are and what are the environmental circumstances that "interact" in the development of the trait. Were the egg-laying capacity of chickens 100% heritable, this only means that variations in this capacity have in the past not been affected by variations in nutrition given to the animals, amount of sunlight, musical entertainment, etc. It also means that good results have been attained by selectively breeding high egg-layers. This "high heritability" means that breeding has worked in the past, while nothing else has, so we should concentrate on that. Farmers knew this long before "heritability estimates" were used to represent it. Heritability figures may give a more precise estimate of how much attention should be devoted to breeding, based on the wider experience of many agriculturists than an individual might estimate by intuition. Thus we do not mean to belittle the degree of value such estimates have, but only to show their purely empirical nature. Heritability is relative to a given form or range of environmental conditions. Egg-lying is highly heritable only given an established state of environmental conditions. And at any moment, an "environmental" change might prove vastly superior to any breeding method. Jensen is precise when he writes,

The proportion of variance indicated by 1-h2 [i.e., the environmental component of the variance], if small, does in fact mean that the sources of environmental variance are skimpy under the conditions that prevailed in the population in which h2 [heritability] was estimated. It means that the already existing variations in environmental (or instructional) conditions are not a potent source of phenotypic variance, so that making the best variations available to everyone will do relatively little to reduce individual differences. This is not to say that as yet undiscovered [p.148] (or possibly already discovered but as yet rarely used) environmental manipulations or forms of intervention in the learning or developmental process cannot, in principle, markedly reduce individual differences in a trait which under ordinary conditions has very high heritability.

In both instances, however, we are dealing with a purely empirical procedure. That is, we cannot explain how the cure is achieved. We only know that herb "x" has been effective in the past, and it therefore makes sense to stick with it. Thus, when we learn how the genes controlling egg-laying in fact operate we may find that they do so by affecting the production of a certain hormone which leads to high egg-laying only when a certain combination of nutrients are assimilated. Artificially producing the hormone and/or providing an ideal diet may produce a much higher egg-laying capacity than could be achieved by the old selective breeding techniques. At this point "heritability" would drop for the chicken population in question, since environmental variation of hormones and nutrition now produce the main differences in egg-laying capacity. But instead of saying this, and inevitably implying the existence of a mysterious entity in the chickens, we could also say that the farmers have stopped concentrating on breeding techniques for producing high egg-laying chickens. The heritability of the chickens has become itself a function of the practical, technological, man-made environment of the chickens.

Interaction, development and the metaphysics of variation

This detour into the heritability of corn and chickens and the "widthability" of football fields was necessary to clarify the precise and very limited meaning of the concept of heritability so that when we hear that "IQ is 80% heritable" we have a clear understanding of the meaning of this phrase -whether it is true or not. It has nothing to do with the real interaction of genes and environment- just as "widthability" has nothing to do with the real relations of the actual length and width or with the variations of the areas or perimeters. Heritability does not refer to the role that heredity plays as opposed to environment. It presupposes a given and unknown interaction of heredity and environment. For humans, it assumes a given interaction of biological laws of development and sociological laws whose [p.149] real mode of interaction is unknown. Jensen himself makes all this clear, so as to gain the assent of the specialists in the field. Attacking "interactionists," who he charges are merely environmentalists in disguise, Jensen discards the argument that genes and environment interact and mutually influence each other in a complex way which cannot be gathered from heritability studies. In his 1969 article he dismisses this position as all-or-nothingism; i.e., either we must know exactly how the genes operate in the overall development of the organism, or we can say nothing about the relative importance of the genes. Jensen argues in 1973 that such "interactionist" critics do not understand the concept of interaction as used in population genetics (i.e., the "multiplicative" relation GE which we have examined above.) But their confusion is

even more a failure to distinguish between (a) the development of the individual organism, on the one hand, and (b) differences among individuals in the population. To say that a growing organism, from the moment of conception, 'interacts' with its environment is a mere truism... and repeating the assertion that the individual is the result of 'the complex interaction of genetic and environmental factors' is simply stating the obvious. What the population geneticist actually wishes to know is what proportion of the variation in a particular trait among individuals is attributable to their genetic differences and what proportion is attributable to differences in their environment.

These remarks explicitly confirm our detailed analysis of the abstract character of both IQ and heritability studies which say nothing about real historical development and attempt to treat variation as something in itself. Jensen's criticism of interactionists in biology is probably correct, and his "clarification" is accurate. But this is not to say that it is very meaningful. We have tried to analyze the limited and pragmatic meaning which heritability figures may have. Real biological scientists may be excused for assuming that heritability pretends to be a scientific concept and should be treated scientifically. Jensen takes for granted the "truism" that there is a complex interaction of genes and environment, and is not interested in a scientific analysis of the real process of development. However he is so immersed in the abstract world of "deviation scores" and of working in the dark with variations, without knowing anything about what it is that varies, that he [p.150] seems to think that his "clarification" should settle matters, and that the distinction between development and variation and their separation has any reality other than in the metaphysical world of abstract concepts. And even there such abstractions continue to slip back and forth between the limited technical usage and the "pioneer" interpretations of Jensen. Carrying his rock of Gibraltar out onto the thin ice of metaphysical and idealistic interpretations, Jensen seems to think that it is perfectly natural to treat pure variations as definite realities with obvious meaning. It is this slipperiness of Jensen's footwork that enables him to indulge in technical refinements which disclaim that he is talking about "intelligence itself" or anything to do with real development, and then conclude that intelligence in fact does not develop. Throughout his work there is a "sleight-of-hand" in which the real questions of human development are "finessed," while abstractions, with the meagerest residue of reality clinging to them, are passed off as the real thing.

This sleight-of-hand is nowhere more obvious than where Jensen applies heritability to human education, seeing the heritability of intelligence as the basis for understanding "educability." Thus in a passage previously cited, Jensen makes all the necessary refinements for applying heritability to human education. High heritability means that "already existing variations in environmental (or instructional) conditions are not a potent source of phenotypic variance...." It is not clear what kind of instructional variations are meant here, but presumably he means that the actual programs to change the school system so as to improve the scholastic performance of low performers have not been very successful. Using the heritability model, since such environmental variations do not affect existing variations very much, these variations must be due to genetic variations. But how thoroughgoing have been the environmental variations? A horticulturist may try to change the poor performances of some of his roses by giving them more water, when the basic problem comes from the clay soil they are growing in. Without a real knowledge of environment, how can one assert that the environment to which Jensen refers are now drying up, with the help of his argument that we have tried and failed. Our gardener who stopped watering his roses because of similar thinking would [p.151] soon lose his roses altogether. Thus if environmental variation is understood superficially, variations will be attributed to genetic differences which are in fact due to deeper structural differences in the environmental changes.

Jensen guards himself against criticism on technical grounds stemming from the fact that heritability says nothing about changeability under different environmental circumstances: "This is not to say that as yet undiscovered (or possibly already discovered but as yet rarely used) environmental manipulations or forms of intervention in the learning or developmental process cannot, in principle, markedly reduce individual differences in a trait which under ordinary conditions has very high heritability." We have underscored Jensen's parenthetical remark and the words "in principle" (in other citations above the underlining was Jensen's own) because they point to the technical significance of the concept, while implying that in practice such marked changes will "probably" not come about. In any case, this assertion completely contradicts the main thrust of Jensen's work which is that environmental changes will never affect a trait so solidly embedded in the individual as "g." It shows how little the concept of heritability can theoretically support the idea of innate and fixed IQ. However, it constitutes one more "argument" in the accumulation of arguments which Jensen uses to make this concept plausible.

Mendel and evolutionary interaction

H. Topoff, who defends an "interactionist" position and questions the validity of heritability studies as offering real scientific comprehension of the developmental process, carries his criticism back to Mendel's own experiments:

Imagine for a moment that you were assisting Gregor Mendel as he conducted his pioneering crosses of inbred, homozygous garden pea plants. Mendel crossed plants having red flowers with those of the white variety; all of the offspring (the F1 generation) had red flowers. Mendel's conclusion was that red flower color is inherited. Suppose you had asked Mendel how the hereditary factors (which we now call genes) produce flower color during the development of any one individual plant from the seed to the mature organism. Mendel's reply could only have been, "I don't know." [p.152]

To say that a particular trait is inherited in this context is not to say that environmental factors played no role in its development. It is rather to say that the variation of certain, usually obvious, environmental conditions, does not affect the distribution of a trait in the given population. Thus, argues Topoff, Mendel raised his plants under equal conditions of light and water, and in spite of this "equal" environmental treatment, the color of his flowers showed pronounced variability -some being white, and some red. A deeper understanding of the complex conditions of development makes this concept of "environment" appear superficial. Although light and water are certainly the most pragmatically accessible environmental conditions, other factors from the very beginning of fertilization, in the seed itself, are external to the chromosomes carrying the genes and being an interactive process, the knowledge of which in developmental genetics brings out the deep penetration of "environment" in the very functioning of the genes. Topoff writes that "The only valid conclusion from Mendel's experiment is that light and water were not responsible for the variability in flower color. Studies of heritability can only point out that certain factors are not responsible for the observed phenotypic variability."

The fact that Jensen refers to his work as part of a "Mendelian revolution" makes it necessary to understand more precisely the "field of action" (to quote Engels) of Mendel's discoveries. It is often pointed out that Mendel's experiments were published in 1865, only six years after Darwin's Origin of Species, and remained basically unrecognized until 1900. This remark is sometimes made to suggest that evolutionary theory developed without any knowledge of genetics. The opposite notion is also true: genetic theory developed without any real connection with evolutionary theory. In fact, Mendel's theory of the genes as remaining unchanged throughout generations makes it essentially impossible to understand evolutionary change. It was only when it was understood how genes themselves could change that genetics could become reconciled with evolutionary theory. Their stability is only relative, not absolute, as was implied by Mendel's work.

There is a clear Mendelian bias in the concept of "heritability." In his breeding experiments, Mendel was interested in the variations that occurred in certain sharply contrasted traits of his pea plants. Thus while the first generation that resulted from the crossing [p.153] of read-flowering plants with white-flowering plants consisted in all red-flowering plants, in the next generation, based on the self-fertilization of the previous generation, on the average one out of four plants was white, while the rest were red. Further self-fertilization of these plants showed that the white plants always bred white, and one-quarter of the red plants always bred red. The remaining fifty percent of the plants, which were red, produced red and white plants in the ratio of three to one -just as did the first generation reds. Mendel concluded that certain "unit characters" (the genes) determined flower color, and that these were of two types, one dominant and one recessive. These variants of the gene are called alleles, and are commonly symbolized as "A" for the dominant gene and "a" for the recessive. These two types of the gene that affects flower color in pea-plants combine in three possible combinations, according to the laws of chance. These "genotypes" are type AA, type aa, and the hybrid type Aa. The first two types will breed true, when self-fertilized, while the third will produce all three types again, in the proportion of AA, 2Aa, aa. Since the dominant gene determines flower color, both the pure dominant and the hybrid genotypes will produce red flowers. Thus the phenotype, red flowers, can be produced by one of two genotypes.

Essential to this analysis is the concept that variations in certain traits will occur in definite ratios, based on the assumption of two types of the gene that separate and recombine over the generations. When such ratios appear in fact and follow the genetic model, it can be said that all the variations are genetically produced and that flower-color is 100% heritable.

Suppose that after he carefully conducted his experiments and made his brilliant deductions, Mendel, dejected at the lack of receptivity to his ideas, allowed his garden to grow wild. When this garden was rediscovered thirty-five years later, suppose that most of the flowers were white, and only a few red ones could be found, mainly in inaccessible areas. This finding contradicts the ideal Mendelian ratio. After some search, it was discovered that the cause of this distortion was a particular species of bird that was attracted to the red-flowing plant, but left the white flowering ones alone. An environmental agent had intervened to "select" some of the plants for destruction. The actual variation of flower color among the pea plants in this particular area can be conceptualized as the product of [p.154] genotypic variation plus some environmental selection that reduces the "ideal" heritability of the particular trait. "Total variation" can therefore be conceptualized as the result of these two factors. Note that low heritability here applies to a trait which is still "genetically determined." "Heritability" refers to the distribution of the trait in a population, not to its causes in the individual. Both white and red flowers are still produced by an interaction of genes and environment whose nature is not in question.

All of this of course leaves unanswered the question regarding "how" the genes operate so as to produce the particular color variations. The pea-plants of course grow up in a definite environment, which is essential to the formation of plant color. Gross variations of this environment, up to a certain limit, do not affect variation of plant color. Light, water, temperature, soil composition, etc., are all necessary for the growth of the plant and for its coloring process. Variation in color, however, does not depend on variation in these factors, and is linked with genotype differences.

The hypothetical bird that destroys red-colored plants is a special type of environmental intervention that appears to disturb the normal genetic-environmental equilibrium in which the environment factor is discounted only so far as providing an explanation of color variation. But this genetic-environmental equilibrium has its own history; pea-plants have not always existed. They evolved in the course of natural evolution through transformations of other species and under the decisive influence of environmental selection. In other words general environmental conditions are necessary for the evolution and normal growth of pea-plants. The independence of the genetic factors is therefore only relative, and has to be explained by previous natural history. Heritability theory abstracts from this "ancient history" and is concerned with how special environmental factors modify the variations that are predictable from knowledge of the genetic composition. The fact that "environment" is regarded as reducing heritability presupposes that ideal heritability is taken as a starting point. Such a conception, which distinguishes genotype effects under normal conditions of development and environmental alterations of a special kind, is valuable within a definite practical field of activity and analysis. Like most such antitheses, once they become absolutized and turned into "metaphysical" absolutes, they become barriers for further scientific development. Such a conception [p.155] is applicable within relatively short time spans, during which the special environmental factor (e.g., the bird) can be conceived of as adding its effects to those of the genotype and detracting from the full expression of the genotype under general environmental conditions. From the point of view of the evolution of the species, however, the environmental "Darwinian" factor still stands out as the overall determining factor in the explanation of natural history. As a result of Mendel's discovery of the gene and further developments of genetic theory, evolutionary theory has been enriched and concretized, but not refuted.

The above examples of heritability presuppose either a definite knowledge of the genotype or of the environment as a basis for determining heritability. In many cases, probably in most instances of importance to agriculture, genetic structure is unknown or imperfectly known. The trait in question, such as milk productivity, weight, etc., is not the result of a single gene or a combination of relatively few genes. Where the trait results from the action of many genes it is said to be "polygenic" and estimates of heritability become more approximate. In ordinary breeding methods, the breeder selects above average qualities which may appear randomly in a given population. He then interbreeds the individuals so selected, in order to produce stock that only exhibit the desired properties. The new generation generally does not do as well as the selected parents, falling somewhere between the average of the original population and the average of the selected parents. Were the offspring to exactly repeat the characteristics of their parents, these characteristics would be considered 100% heritable. A simplified formula for determining heritability under these conditions is to determine the extent to which the selected parents are above average for the given population and then the extent to which their offspring is above the original average. Heritability is then defined as the ratio of the "gain in selection" of the offspring over the "selection differential" of their parents. Thus if the parents average twenty pounds above the average for their particular population and their children average only ten pounds above average, heritability of that trait is 10/20 or .50. Essentially it is argued that the variation of the trait in the original population is a result of genetically determined variation plus environmental variation. Since environmental conditions are kept similar for the selected parents and their offspring, the [p.156] offspring are said to express the extent to which their parents are genetically above average. This explanation is a simplification of a more complicated situation, involving the possibility, e.g., that a combination of recessive genes in the parents may produce the relatively poorer showing of their offspring. Since this would be a genetic effect, its probability would increase the estimate of heritability. On the contrary, suspicion that the selection sample was getting special environmental treatment would diminish the heritability estimate. This entire experiment would of course be relative to a particular environmental context, which would be as controlled or equalized as possible. Heritability is seen to be still relative to environment. Furthermore, we are dealing with variations from the average or population means, and our heritability figure has to be evaluated as having meaning only in this context. Finally, we do not explain how the actual genetic structure produces the final result. Agricultural science has in fact made major strides precisely by leaning less on gross heritability estimates, and more on understanding the precise mechanisms of gene-environment interaction -where it is not possible to juxtapose heredity and environment as two separate factors. Far from being an empty truism, the concept of the gene-environment developmental process provides the key to major advances in agriculture. Heritability estimates can serve a useful, but limited function in this study.

The social environment and human activity

Thus even where breeding experiments produce marked phenotype variance, at most we can say is that the immediate cause of the variation is "hereditary"; but since this effect operates only in the complex gene-environment interaction, it is not possible to say that the ultimate or determining cause is "hereditary" -for breeding is always relative to a certain response to environmental conditions. Such a clarification is itself likely to be misleading if we were to apply it to human development, where, as we have seen, the polar relation of genes and environment, which determine the biological process, cannot provide a scientific basis for the study of real human development. Biological processes are integrated into a socioeconomic process whose basic determining factors are the interaction of productive forces and relations of production in the unity-of-opposites [p.157] of human society and nature. Even where criticisms are levied against the misuse of the concept of heritability, and the gene-environment developmental process is explained, it is still misleading to apply this "interaction" to the process of human development, or to suggest that IQ differences can be looked at in this framework. John Hambley writes of the "environmentalist" response to the "genetic" argument that "the tacit idea of manipulation of the environment to produce the 'appropriate' phenotype, while perhaps appropriate in agriculture, represents shortsighted paternalism in the human context." More deeply, this approach implies a population that is only passively related to an "external" environment, and not the active determiner of its own development. The concept of an "enlightened" educator or class of technocrats who can mold the population according to an ideal preconception of life is the Skinnerian, behaviorist foil against which heriditarians construct their arguments that the real determinants of development are buried "within" the individual. Marx rejects this antithesis between a "contemplative materialist" theory of knowledge, on the one hand, and an activist, but idealistic theory of knowledge on the other. Thus in his first thesis on Feuerbach Marx writes:

The chief defect of all hitherto existing materialism... is that the thing, reality, sensuousness, is conceived only in the form of the object, or of contemplation, but not as sensuous human activity, practice, not subjectively. Hence, in contradistinction to materialism, the active side was developed abstractly by idealism -which of course, does not know real, sensuous activity as such.

The debate that Jensen carries on with "egalitarian environmentalists" reflects a similar philosophical antithesis. Whereas the "environmentalists" -generally, behaviorists- argue that learning is the result primarily of external environmental inputs, Jensen argues that their programs have failed to produce any significant variation in the results of their educational efforts. The cause of variation, he argues, should therefore be sought in the "inner," i.e., genetic, part of the individual. What is missing from both analyses, to follow Marx, is social activity, human "self-changing," which today has the main feature of overcoming the historically developed and transient mode of separating individuals from their underlying and necessary social relations and from the main material and intellectual instruments of existence. To confine the argument to [p.158] deciding between two sides of the separation between the individual and the "environment" is merely to reflect the effects of the underlying social structure, and to build speculative bridges from one side to the other. The real connection between the individual and his environment, involving a theoretically more essential understanding of the "environment," consists in one's practical, social activity which is increasingly impeded by the transient mode of production in which the means of life are privately appropriated even while they become more and more directly social. A deeper understanding of the real connection between the individual and the environment as social practice, Marx writes, ultimately has a revolutionary direction.

*See also Lawler's (1978) "Introduction"; and his critique of Anne Anastasi's views on the "Reliability and validity of IQ tests".

Anastasi, A. (1958). "Heredity, environment, and the question "How'?" Psychological Review, 65, 197-208.

Dobzhensky, T. (1973). Genetic Diversity and Human Equality, Basic Books, New York.

Hambley, J. (1976). "Diversity: A Developmental Perspective" in In N.J Block & G. Dworkin, eds., The IQ Controversy. Pantheon Books, New York.

Jensen, A. (1969). "How much can we boost IQ and scholastic achievement?" Harvard Educ. Rev., 39, 1-123.

Jensen, A. (1973). Educability and Group Differences, Harper & Row, New York.

Lewontin, R. (1976). "Race and Intelligence" In N.J Block & G. Dworkin, eds., The IQ Controversy. Pantheon Books, New York.

Overton, W.F. (1973). "On the Assumptive Base of the Nature-Nurture Controversy: Additive versus Interactive Conceptions," Human Development, 16, 74-89.

Toppoff, H. (1974). "Genes, Intelligence and Race," in E. Tobach et al., The Four Horsemen, Racism, Sexism, Militarism and Social Darwinism, Behavioral Publications, New York.