"The g Factor" is Jensen's (1998, 1999) most comprehensive statement about the origins of differences in human intelligence. The book's impact may unfortunately be diminished because of his detailed attention to race differences. It is g's fundamental importance in predicting a wide variety of real-world outcomes and its covariation with a number of biological and cognitive indices that should be the focus of debate about general intelligence.
2. The first is simply that Jensen's persistence over the years in focusing his considerable talents on the legacy left by Charles Spearman's discovery of general intelligence has its down-sides. Critics as well as those in broad agreement with Jensen may not find many new ideas here. To my eye at least, Jensen's views on g, the first factor or first principal component extracted from a battery of complex problem-solving tasks, have not changed much since he first took up the topic, and he has published an imposing body of evidence to substantiate them. What he does here is to bring together the considerable evidence regarding g and provide the reader with an extended description of what I will call g's 'fundamentality.' In this regard, the chapter entitled, 'The Practical Validity of g' is indispensable. Jensen traces here the matrix of predictive validity correlations into which g enters, a body of evidence that is deep and impressive. Critics of g theory often act as though they were unaware of the extended nature of these connections, but much of this evidence has been available for some time. There can no longer be much doubt, at least from an empirical standpoint, that the g score extracted from an intelligence battery is the single most predictive feature of an individual that can be measured, if the intent is to predict later occupational and socioeconomic status, among a family of other important outcomes. Unfortunately, if critics have not been convinced by these facts by now, I doubt that anything in "The g Factor" will sway them.
3. My second concern about the book's impact stems from Jensen's long-standing interest in race differences, specifically black/white differences. "The g Factor" clearly reflects that interest. Jensen devotes two chapters to this topic under the rubric of population differences in g, and these two chapters alone constitute slightly less than one-third of the book. If race-related material in other chapters is factored in, the topic of race differences occupies a dominant portion of the book's material. This is an enormous commitment of the book's resources, one that will, I fear, allow critics to dismiss this work as racially tinged, as they have some of Jensen's earlier contributions.
4. The race-difference arguments offered in this work consist of a set of premises that lead Jensen to what he terms the 'default hypothesis,' the idea that all of the available evidence implies that at least a portion of race differences in intelligence is genetically based. The accuracy of the premises that lead to this conclusion is questionable, but before elaborating on those reservations Jensen's premises should be set forth in some detail to see the nature of his argument. The first premise focuses on hominid evolution. Jensen argues that Homo sapiens branched from the African Homo erectus line approximately 100,000 years ago. From there over the course of the next 70,000 years or so they spread throughout Europe, Asia, and what has become the rest of the populated world. In that time, breeding isolation occurred, and with the inevitable mechanisms of mutations, natural selection, and related migration effects, appreciable genetic changes occurred. The result was that the non-African Homo sapiens became somewhat distinct lines, or races, to follow Jensen's use of that term, though ones with 'fuzzy' boundaries.
5. This foundation allows Jensen to conclude not only that the average genetic distance between African and all non-African populations is appreciable, but that these genetic differences must perforce have included brain differences, given the centrality of brain function to evolutionary fitness and the concomitant selection pressures that must have attended human evolution. At this juncture, none of this evidence bears directly on race differences in intelligence, but Jensen then raises the following argument concerning the outcome of this genetic distancing:
"In migrating from Africa and into Europe and Asia, they (Home sapiens) encountered highly diverse climates. These migrants... had to forage for sustenance under the highly different conditions of their climatically diverse habitat... This necessitated the development of more sophisticated techniques for hunting large game, requiring vocal communication and cooperative efforts... It seems plausible, therefore, that the behavioral adaptations of a kind that could be described as complex mental abilities were more crucial for survival of the populations that migrated to the northern Eurasian regions, and were therefore under greater selection pressure as fitness characters, than in the populations that remained in tropical or subtropical regions." (pp. 435-436)
6. Jensen attempts to bolster this hypothesis by factoring in the literature on group differences in head or brain size, and concludes that there is a reliable rank-ordering of the three most studied populations, with East Asians ranked first, followed by Europeans and Africans. He goes so far as to estimate what portion of the approximate 15-17 point IQ difference between blacks and whites is due to brain size -- about 6 points, he concludes. A final facet of this argument, and an important component of the default hypothesis, contends that whatever factors produce within-group differences in intelligence, they are the same as those that produce between-group differences. Given that the heritability of intelligence, at least as measured by standardized tests, is substantial, the default hypothesis implies that black/white differences contain a genetic component as well as an environmental one.
7. In attempting to evaluate this line of reasoning, it should first be accepted that there is indeed something to be explained. Race differences, particularly black/white differences in IQ, are appreciable in magnitude, at least one standard deviation or more, and have been remarkably resistant to change over the entire time that such measurements have been taken. They appear to have been little meliorated by the vast changes in culture and opportunity in the past 75 years. Critics of Jensen's arguments contend that these differences would be eliminated if only the tests were not racially biased, if schools were not biased, if our culture were not biased, or perhaps if Jensen and others who measure these differences were not biased. And anyway, they argue, race itself is such a permeable construct as to be useless for empirical study.
8. These arguments are for the most part nonsense. The tests appear to predict equally well for blacks and whites to the point that if one equates on g score (e.g., imagine a black person with a total IQ score -- a reasonable g surrogate -- of 100 and a white person with the same score), income and other socioeconomic differences are largely diminished, though not to the point of disappearance (see Herrnstein & Murray, 1994). As for the ability of schooling to overcome race differences: g, as represented by total IQ score, is quite difficult to change, and there is no evidence that intensive schooling can produce the sort of malleability in IQ that is demanded by environmentalists (Locurto, 1991a). The best evidence concerning the impact of environmental changes comes not from preschool interventions but from adoption studies. These studies are profoundly affected by design and interpretative issues, but even if one looks beyond these caveats, the best evidence, coming from studies of extremely contrasted environments, indicates a possible change of 10 to 12 points (Locurto, 1990). I have argued elsewhere (Locurto, 1991b) that the extent of this malleability is nearly sufficient to account for most of the black/white IQ difference and leave a remainder of little practical or theoretical interest. It should be added for present purposes, however, that this malleability is hard won: it comes after many years of living in quite favorable circumstances after having been born into quite modest circumstances. One must assume that the average difference in living conditions between blacks and whites corresponds to this sort of profound environmental difference to consider the black/while difference as largely environmental in origin.
9. These caveats aside, the premises composing Jensen's default hypothesis do not add up in my view to a conclusive demonstration of the origin(s) of race differences. The particular out-of-Africa thesis which Jensen puts forth imagines a one-way ticket that leads to genetic isolation, whereas other current approaches to human evolution posit a sort of 'in-and-out of Africa' approach where migrating early hominids may have periodically returned to Africa (Bownds, 1999). What is most speculative about Jensen's approach -- and he is by far not the first to put it forth -- is not how many round-trips our early ancestors may have made, but what we can make of the differing environmental conditions they may have faced and how these differences map, if at all, on to later intelligence differences. I can't imagine that one can accurately deduce differing selection pressures and survival challenges with any accuracy for either African or migrating hominids. Perhaps the migration itself was initially forced by the severity of living conditions, or by the inability of those who migrated to compete successfully with those who remained. If Jensen's conjectures in these regards are unsound, inferring resultant intelligence differences between these two groups from this foundation is also unsound.
10. The next link in the chain, head/brain size differences between races, has also received critical attention (e.g., Kamin & Omari, 1998), although it should be stated that the weight of evidence appears to favor some, though quite limited differences between races of this kind. But to extrapolate from these modest differences to g differences suggests a linearity between head/brain size and g that I think is unsupported by available evidence. If there is a brain size difference, where is that difference centered? And if that question can be answered, can we then map those differences on to anything connected to intellectual functioning? I suspect not.
11. The last link in the chain is Jensen's claim that the factors that engender within-group differences in g also constitute the factors producing between-group differences. This hypothesis has the satisfactory aspect of simplicity. According to the alternative (which Jensen calls the dual hypothesis), whereas genetic and environmental factors produce within-group differences, only environmental factors are involved in between-group differences. The problem with the default hypothesis, as with its alternative, is that there is simply no way to demonstrate the correctness of either one. Jensen appeals to various data sets taken from transracial adoptions and other 'racial admixture' studies, but the interpretative problems in evaluating this sort of work are insurmountable (Locurto, 1990). Take for example the classic study of transracial adoptions by Scarr and Weinberg (1976) which principally contrasts three groups of adopted children: children of two white parents, children of two black parents, and interracial children. Jensen attempts to divine corroboration for the default hypothesis by comparing the mean IQ levels of these three groups with each other, with regional norms for IQ, and with the IQs of the adoptive parents and their biological children. Unfortunately, these comparisons are confounded by several factors, especially for the groups of black and interracial children that constitute some of Jensen's important comparisons with respect to the default hypothesis. For one, the pattern of the data in this study clearly indicates that the interracial children were adopted into more favorable families. They were also adopted earlier, at an average age of nine months compared with 33 months for the children of two black parents. That finding in turn indicates that the interracial children spent more time with their adoptive families before testing and were subjected to fewer pre-adoptive placements. Given these problems alone, comparisons between the interracial and black children in this study are useless for determining genetic influences on between-group differences.
12. The problems with demonstrating the origins of race differences lead most investigators to be wary of using them to infer basic principles about human intelligence. Add in the political costs for pursuing this interest and most investigators deliberately steer their work clear of this issue. Jensen has been unyielding in his interest in race differences, and he alone fully understands the price he has paid for this interest. My fear, as I stated earlier, is that his prodigious attempt to support the default hypothesis will undermine what should otherwise be a highly significant book. As someone interested in human differences in intelligence, I am especially concerned that the focus on race differences will allow critics to ignore this work. Instead, they should be forced to confront g by the depth of Jensen's arguments about what I called earlier the fundamentality of g.
13. G and Reification. G-theory stipulates that the general factor extracted from intelligence batteries represents a generalized problem- solving ability, one that is invoked in the solution to a wide variety of intellectual tasks. One criticism of this approach has come repeatedly from Gould (1981) and others, who argue that investing the first extracted factor from an intelligence battery with so much power represents the error of reification, that is, the false equating of a psychometric entity with a real mechanism. In theory, reification is indeed a potential problem: it is not unusual to find a first factor of some magnitude in almost any battery of tasks, and that factor need not represent any real entity.
14. There are two points to make about the problem of reification. The first is that Jensen appears to be well aware of it and he offers no simple-minded definitions of g. In fact, he takes pains not to define it in any simple reductionistic or reified way:
"It is wrong to regard g as a cognitive process, or as an operating principle of the mind, or as a design feature of the brain's neural circuitry. At the level of psychometrics, ideally, g may be thought of as a distillate of the common source of individual differences in all mental tests, completely stripped of their distinctive features... In this sense, g can be roughly likened to a... computer's central processing unit. At the level of causality, g is perhaps best regarded as a source of variance in performance associated with individual differences in the speed or efficiency of the neural processes that affect the kinds of behavior called mental abilities." (p. 74).
15. The second point concerns Gould's principal claim that g does not correspond to any real, that is, biological, entity. He argues in The Mismeasure of Man that 'to vindicate either view [g theory or non-g theory] some legitimate appeal must be made outside the abstract mathematics [of factor analysis] itself... All combatants made appeals to biology and advanced tenuous claims, but no concrete tie has even (sic) been confirmed between any neurological object and a factor axis (1981, p. 310; parentheticals added).' Gould repeats this claim verbatim in the 1993 and 1996 re-issues of his book. The repetitions, without modification, are unfortunate, given the advances that have been made in the ensuing 20 years in marking g using physiological/biological measures, including EEGs, averaged evoked potentials (AEPs), glucose metabolism, among others. Their impact on the study of human intelligence has to date been more basic-science oriented than practical, and there are numerous issues to be resolved about the nature of these relationships (e.g., Verleger, 1999). At the least, the wide variety of substantial correlations into which g enters with these physiological measures should give one pause before making rigid statements about the lack of evidence.
16. These links between g and physiological/biological measures can be extended to work on the relationship between g and a number of elementary cognitive tasks (ECTs) such as simple and discriminated reaction time and inspection time (e.g., the time required to distinguish between two lines of different length). The latter correlations are themselves substantial. Taken together, this 'g nexus' strongly suggests that g has life beyond the factor matrix and is not simply an example of the error of reification. While it is premature to conclude much more than this, the fact that g appears to be related to measures that take it far from its paper and pencil test origin is consistent with the idea that it is more than a psychometric artifact. It appears instead that, as Jensen argues, g is related in some general way to speed and/or efficiency of information processing.
17. G and Modularity. Despite the extensive evidence for the presence of some sort of general ability and its correlation with a wide variety of cognitive, physiological and real-world measures, critics often contend that the proper conceptualization of mind should focus instead on specific abilities, or 'modules' to follow Fodor's (1983) original use of this term. This view has gained particular credence among evolutionary psychologists, who view brain organization as the result of a series of specific selection pressures that have produced a corresponding series of specific adaptations. Adaptations, in other words, are problem specific.
18. General solutions of the sort represented by g would, in this view, fail to guide an organism to correct adaptive solutions and would be extremely costly (see Buss, 1999, pp. 51-52, for an elaboration of this argument). It appears, unfortunately, that a correlate of this view is the attempt to render g not only evolutionarily implausible but also trivial. Pinker (1997) contends, for example, that 'Nothing could be farther from my subject matter (i.e., the evolution of mind) than a comparison between the means of overlapping bell curves for some crude consumer index like IQ' (p. 34; parenthetical added). Why IQ (read g) is a 'crude consumer index' is unexplained. Similarly, Bownds (1998) also refers to IQ as 'crude' -- twice, in fact, in a section devoted to the primacy of specific vs. general solutions:
"IQ tests are a crude estimate of intelligence in the same way that pulse, blood pressure, and body temperature provide an overview of general health. A physiologist or physician can proceed to tests of liver function, heart function, and so on, but the ability of a psychologist to dissect the components of intelligence is much more crude." (p. 169)
19. If Bownds's analogy were correct, g would indeed be a crude index of intelligence, but there are two problems with Bownds's argument. The first is the presence of the impressive g nexus. Differences in socioeconomic status reflect systematic g differences, as do differences in education level, or performance on ECTs and on a variety of physiological measures. Rather than being crude, g appears to be sensitive to detecting differences in these other domains. The analogy with heart rate or blood pressure is simply incorrect.
20. Second, there is in principle no conflict between measuring differences in specific cognitive abilities -- the psychological equivalent of Bownds's heart and liver functions -- and also measuring g. This question of specific vs. general abilities is part and parcel of Gould's attempt to refute g through the argument of reification. It is also central to Gardner's (1983) effort to argue for the existence of independent multiple intelligences. These ideas, it should be noted, have a long history in psychology. Decades before Gould and Gardner made their arguments, Thurstone (1947) had set forth his multiple factor model of intelligence. The model included eight primary abilities that were posited as distinct entities not connected through a general factor.
21. The subsequent development of a hierarchical model reconciled the differences between specific vs. general factor approaches by demonstrating that there were indeed specific factors, much as posited by Thurstone, and that these specific factors were themselves correlated, thereby revealing g. As such, g was not antithetical to Thurstone's primary abilities but was instead a powerful organizing factor for more specific lower-order factors. Similarly, the modern incarnation of Thurstone's theory in the from of modularity is not itself incompatible with a general intelligence factor. The modules, if you will, have something in common. Or, stated differently, they are integrated by an overarching process that ensures their functionality in generating adaptive behavior. If this view doesn't sit easily, imagine its alternative -- a set of functionally unrelated modules attempting to produce adaptive behavior. It is only a short step from this argument to imagining that individual differences in this organizing process, and, hence, in the appropriate application of the modules, would be central to all sorts of performance.
22. Rather than being crude, modern hierarchical analysis bespeaks a kind of elegance that is associated with maturing theories in science. It is interesting that critics of Jensen's approach seldom mention hierarchical models, possibly because the reconciliation of general vs. specific factors is troubling for simple-minded criticism based on modularity. They also fail to mention the now-elaborate g nexus of correlations which is troubling for any simple-minded arguments based on reification. Critics also do not mention that one of psychology's enduring motivations has been a kind of 'physics envy', in which informal statements about psychological phenomena eventually yield universals that are ultimately transformed from verbal statements to the more certain language of mathematical models grounded in replicable measurement. G theory offers just such a progression, and its universality, while not fully established, now looms much larger than the arguments of critics.
Bownds, M. D. (1999). The biology of mind: Origins and structures of mind, brain, and consciousness. Bethesda, Md: Fitzgerald Science Press.
Buss, D. M. (1999). Evolutionary psychology. Needham Heights, MA: Allyn & Bacon.
Fodor, G. (1983). Modularity of Mind : A Monograph on Faculty Psychology. Boston: MIT Press.
Gardner, H. (1983). Frames of mind: The theory of multiple intelligences. New York: Basic Books.
Gould, S. J. (1981). The mismeasure of man. New York: W. W. Norton.
Herrnstein, R. J. & Murray, C. (1994). The bell curve: Intelligence and class structure in American life. New York: Free Press.
Jensen, A. (1998). The g Factor: The Science of Mental Ability. Praeger
Jensen, A. (1999). Precis of: "The g Factor: The Science of Mental Ability" PSYCOLOQUY 10 (23). ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1999.volume.10/ psyc.99.10.023.intelligence-g-factor.1.jensen http://www.cogsci.soton.ac.uk/cgi/psyc/newpsy?10.023
Kamin, L. J. & Omari, S. (1998). Race, head size, and intelligence. South African Journal of Psychology, 28, 119-128.
Locurto, C. (1990). The malleability of IQ a judged from adoption studies. Intelligence, 14, 275-292.
Locurto, C. (1991a). Beyond IQ in preschool programs? Intelligence, 15, 295-312.
Locurto, C. (1991b). Sense and nonsense about IQ: The case for uniqueness. New York: Praeger.
Pinker, S. (1997). How the mind works. New York: Norton.
Scarr, S. & Weinberg, R. A. (1976). IQ test performance of black children adopted by white families. American Psychologist, 31, 726-739.
Verleger, R. (1999). The g factor and event-related EEG potentials: Book Review of Jensen on Intelligence-g-Factor. PSYCOLOQUY 10(39) ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1999.volume.10/ psyc.99.10.039.intelligence-g-factor.2.verleger http://www.cogsci.soton.ac.uk/cgi/psyc/newpsy?10.039