The Nature of Intelligence

The Novartis Foundation, an international science and education charity located in London, regularly hosts symposia that deal with cutting edge issues in science; appropriately, the organization often deals with topics related to evolution and psychology.“The Nature of Intelligence” meeting, held in late 1999, convened to address questions at the intersections of evolutionary psychology and behavioral genetics.The book by the same name (edited by Gregory Bock, Jamie Goode, and Kate Webb) consists of the papers read at that conference, as well as transcripts of the fascinating discussions and conversations that took place between the participants.

While the topics covered range widely over issues such as whether or not “general intelligence” (known as ‘g,’ first postulated by Sir Francis Galton in 1869) exists to the modularity of mind, the primary question addressed by the authors revolves around a fundamental tension: one would predict that a trait so strongly adaptive as g would, owing to intense selection for it, show only weak individual variation within the population.Evolutionary psychologists such as John Tooby and Leda Cosmides have argued for this position.Behavioral geneticists, on the other hand, note that there is ample empirical evidence that intelligence is highly variable as well as highly heritable.How are we to reconcile this tension?Answering that question requires addressing a host of assumptions and hypotheses, many of which are surfaced in the book.I’ll spend most of the remainder of this review discussing each paper in the volume.Ultimately, they are very satisfying, but what is most provocative is the inclusion of transcripts of the discussions that took place; here, the “give and take” of science can be examined first hand, and the reader gains an appreciation for the fallibilism and humility that undergirds the pursuit of scientific knowledge.These interludes punctuate each paper and are a delight to read.

After Michael Rutter sets the stage for the conference in his introduction, David Lubinski begins (in “Intelligence: success and fitness”) by defending the scientific evidence for the existence of g, arguing that it is a very robust construct.It has “…achieved the status of differential psychology’s most central dimension” (p. 17).General intelligence does not merely co-vary with other measures of abstract learning and work performance; rather, he states, its role is primarily causal.

Arthur Jensen argues in his essay “The g factor: psychometrics and biology” that we should not expect to find a cognitive module for “general problem solving”; rather, g as a construct cuts across the various cognitive functions of the brain, and is instead highly correlated with very basic brain properties such as brain size, reaction time, nerve conduction velocity, metabolic rate, etc.Jensen approvingly quotes Charles Spearman, a pioneer in the investigation of g, who noted in 1927 that the final understanding of g “…must come from the most profound and detailed direct study of the human brain in its purely physical and chemical aspects” (p. 46).

Next, Ian Deary’s essay “Psychometric intelligence differences and brain function” makes the case for a multilevel exploration of g’s instantiation in the brain, arguing that investigation at the psychometric, cognitive, psychophysical and physiological levels will prove useful; the real “explanatory toil must come from linking individual differences at different levels of description” (p. 71).

Britt Anderson shifts gears slightly, examining “The g factor in non-human animals.”He argues that there are direct correlates of g in other animals besides people; given that we can directly intervene in brain development in these creatures in a way that we can’t in humans, “genetic contributions to g can be directly tested by modifying gene expression and determining the…benefits” (p. 79).We can move beyond the correlation/causation debate by using animals as test cases for hypotheses about the heritability, and neural substrates, of g.

Randolph Neese argues in his essay “Natural selection, mental modules and intelligence” that the debate regarding whether g is one trait or many has no commonly agreed-upon answer.In part, this is because knowing functional demands gives only modest guidance as to the structures that mediate those functions.Rather than look for mental modules that will constitute g, we should instead recognize the mixed modular/holistic nature of the mind, and gain what explanatory purchase we can upon the problem using concepts from evolutionary psychology.

Nathan Brody defends g’s conceptual hegemony in “g and the one-many problem: is one enough?”Brody thinks one concept is enough: “g is a heritable component of intelligence that accounts for approximately 50% of the covariance among diverse measures of intelligence…the covariances…are, in large measure, attributable to common genetic influences” (p. 122).He defends the explanatory necessity of g in the domain of intellect, rebutting the argument that it does not deserve pride of place in any general theory of intelligence.

In “General intelligence and the definition of phenotypes,” Douglas Detterman acknowledges the empirical support for g, but nonetheless argues that it is not the “perfect phenotype,” as it is probably a statistical abstraction that correlates a number of separable basic cognitive processes.While we know g is important, we still don’t know what it is, and we won’t know what it is until we have a better grasp of its phenotypic characteristics.

David Houle’s interesting essay “Is there a g factor for fitness” argues that fitness components have a variable but positive relationship with overall fitness (captured by an ‘f factor’), in much the same way that the factors that constitute g have a variable but positive relationship with it.His conclusion is that we should not expect fitness to automatically increase with increases in g; using the analogy of body size and growth rates, where increases in either of those factors are not automatically correlated with fitness, we can expect that in certain environments more g will not lead to more fitness.Indeed, it seems likely that current estimates of the relationship between g and fitness “…are neither likely to illuminate our evolutionary history, nor enable us to predict our evolutionary future” (p. 157).

J. Michael Bailey’s essay “How can psychological adaptations be heritable?” is a review of the nascent literature linking psychological traits to fitness and heritability.The most successful examples of such linkages, Bailey argues, are in the field of sociosexuality, where one finds the “…most successful evolutionary account to date of human psychological variation.”

In a similar vein, Andrew Whiten argues in his essay “Social complexity and social intelligence” that the Social (or “Machiavellian”) Intelligence Hypothesis—that primate intelligence came to exist because of the demands of primate social life—has been steadily gaining empirical support over the last 20 years.Whiten’s thesis is that domain specificity may not be an appropriate characterization of social intelligence; rather, our social intelligence abilities may be parasitic upon a cluster of advanced cognitive abilities that all ultimately rest on our capacity for second-order mental representation.The fact that we can “think about thinking” is critical—while second-order representation might subserve social intelligence, it is nonetheless a general purpose capacity that contributes to other important primate cognitive skills as well.

In “IQ gains, WISC [Wechsler Intelligence Scale for Children--a common IQ test] subtests and fluid g: g theory and the relevance of Spearman’s hypothesis to race,” James Flynn’s statistical work indicates that human groups can make “massive fluid g gains in a period too short to accommodate radical change in the speed and efficiency of neural processes” (p. 202).Flynn explores the impact this might have on debates about IQ and race, noting that verification of the Spearman hypothesis(that black/white IQ differences will tend to be largest on tests with the greatest g-loading) will not allow us to settle the “race and IQ” debate.

Andrew Pomiankowksi, in his paper “Mutation, selection and the heritability of complex traits,” argues that genetic variation in intelligence is probably maintained in the same way that variation in complex sexual ornaments used in courtship displays has been maintained.Contrary to expectation, comparative data demonstrates that sexual ornaments have higher additive genetic variation than similar non-sexual traits.This is probably because selection for extreme phenotypes (“my, what lovely l-o-n-g tailfeathers you have”) results in “condition-dependent expression of sexual traits.”That is, the number of genes that contribute to variation between traits is itself subject to evolutionary change—higher genetic variance is expected in traits affected by more genetic loci (as intelligence probably is).

In “The quantitative and molecular genetics of human intelligence,” Peter McGuffin argues that, whatever it is, g has a genetic component, and that current molecular genetic research can help us identify genes influencing intelligence.He concludes that new research techniques, such as detection of single nucleotide polymorphisms on microarrays, will help us better map quantitative trait loci on the human genome; “this together with the completion of a draft version of the human genome sequence which has recently been announced means that genes contributing to g may be identified (as opposed to just located) within a matter of years” (p. 253-4).

Finally, in his essay “Sexual selection for indicators of intelligence,” Geoffrey Miller argues that sexually selected characteristics share the same general diagnostic features as human intelligence.This gives us reason to think that intelligence is sexually selected; Miller proposes nine “novel, testable predictions about human intelligence derived from sexual selection theory,” including that the g factor will prove to be subordinate to general fitness factors (recall the “f factor” discussed earlier in Houle’s essay), and that mate choice mechanisms should favor courtship behaviors with high g-loadings.

So where does this leave us with regards to the question that motivated the conference?The several papers that touched directly on the issue favor the idea that since g’s expression will be a function of a multitude of genetic loci that we should expect such expression to be more highly variable despite obvious selection pressures; but even this idea received constructive criticism during the meeting.In the end, Michael Rutter’s closing remarks are indicative of the range of topics considered in these essays and the variety of viewpoints represented.As he notes, “it would be foolhardy…to attempt any kind of ‘bottom line’ message from the meeting.Even so…we agreed on the value of both evolutionary psychology and behavioural genetics.Both must constitute parts of the overall research portfolio to be brought to bear on the question of the nature of intelligence….” (p. 286).These remarks are generous and ecumenical; still, the conceptual breakthroughs that will enable paradigm shifts in the study of intelligence will mostly likely come from new methods of integrating the inputs from these two fields of study, and from deciding how they should be supplemented so as to compensate for their particular strengths and weaknesses.

This volume is an excellent introduction to the state of play in the relationship between intelligence and genetics; someone looking for “end-of-science”-style monolithic statements about the field should look elsewhere, but those looking for interesting and provocative science and philosophy about both fields should start with “The Nature of Intelligence.”

© 2001 William Casebeer.

William D. Casebeer is an Assistant Professor of Philosophy at the US Air Force Academy in Colorado Springs, Colorado.His research interests include ethics, cognitive philosophy, philosophy of mind, and philosophy of biology.

This book is also available from BN.com: Click here.