It is clear that cognitive abilities decline with damage to various areas of the brain, but we have very little knowledge of the effect of the most basic variable, size, on brain functioning in normal individuals. Some linguists (e.g., Bickerton 1990) have argued that, although language was made possible by our large brain, size does not have a direct influence on ability once a certain critical mass is obtained (i.e., either the circuit is there or it isn't).
On the other hand, it may be that the number of circuits devoted to a task does relate to ability. (Raz, et al. 1987), after showing that a very simple auditory recognition task correlates strongly with IQ (r= 0.68, p< 0.05) suggest that redundancy of neural components aids in the resolution of a detection system. They note that there are only about 2,000-4,500 hair cells in the cochlea of the ear that fire in response to vibration at different frequencies, but that this increases to 10 million neurons in the primary auditory area of the cortex (Green and Wier 1984). Larger numbers of neurons and organized in a hierarchical fashion would be the most plausible way to increase the resolution of noisy, imperfect signals (Raz, et al. 1987). Greater resolution at one level means faster information processing at the next higher hierarchical level. As discussed in the previous chapter, Calvin (1982, 1983a) points out that by geometrically increasing the number of neurons (or neural units) operating in parallel, neural timing accuracy is increased. He suggests that throwing might have been the initial impetus to increased brain size, after which language (and perhaps other cognitive abilities) was able to utilize the increased sequential timing abilities that this increase allowed. Both of these models suggest ways in which an increase in neural tissue may relate, directly or indirectly, to individual differences in cognitive abilities.
As indicated in chapter 2, the pre-frontal cortex in humans is about twice as large as would be expected given a primate brain of that size (Deacon 1988b).[1] Our brain as a whole is several times larger than one would expect given our body size. This means that the increase in size of the prefrontal cortex accounts for much more of the total increase in brain size that occurred during human evolution than other areas of the brain. The frontal lobes are known to be crucial in all sorts of "higher level" cognitive abilities (e.g. "planning"), but they also play a key role in memory of sequential order. Human patients with prefrontal cortex lesions are unable to remember information regarding the temporal aspects of their environment. They cannot plan and execute a complex set of motor movements, program a set of activities in correct temporal order, or remember the order of experiences (Fuster 1985, Milner et al. 1985, 1991, Struss and Benson 1986, Squire 1987). Frontal cortex damaged patients cannot remember the order of presentation of pictures or words but they can remember that the items were previously presented (Milner 1971). They also show difficulties ordering words into sentences and detecting grammatical errors (Novoa and Ardila 1987).
What is perhaps most interesting is that this same dissociation between item memory and order memory (with memory of sequential order localized in the frontal lobes) has been demonstrated in monkeys (Petrides 1991, Squire 1987) and even rats (Kesner and Holbrook 1987, Kesner 1990). The fact that the prefrontal cortex appears to be specifically involved in memory for serial order in species as far removed from humans as rats suggests that this specialization is very old (primate-rodent common ancestry dates to ~65 MYA; Sarich 1985). It appears that, at least in some respects, the human brain has not undergone a drastic restructuring, nor has it evolved a unique set of functional sub-units, but rather it has simply emphasized certain basic cortical functions that had evolved millions of years earlier and are apparently widespread.
Humans excel at understanding (and make use of) cause and effect relationships. This fact may be the most important reason humans are so adept at manipulating the environment (social or otherwise) to their advantage. In fact, one could argue that the process of reasoning is nothing more than an analysis of cause and effect relationships. But in order to utilize cause and effect relationships, the organism must be able to remember which actions cause which effects: i.e., the organism must be able to remember a specific temporal sequence. Given the bias in brain size increase in the frontal lobes during human evolution, they are a natural place to look for behavioral correlates to individual differences.
This may also have important implications for the evolution of syntax (Schoenemann and Wang 1996). Although Bickerton (1990) argues that syntax is not simply word order, language communication necessarily occurs through a serial channel (Pinker and Bloom 1991). While it is true that some languages, e.g., Latin, make less use of serial order in their syntax, ALL languages nevertheless display some form of word order constraint. Syntax must involve rules which translate a multi-dimensional idea or set of relationships into a single dimension. Word order is commonly used to convey these relationships. Without the ability to pay attention to sequential order, it would be impossible to learn most languages.
As discussed in chapter 2, there is a wide range of variation in brain size (and presumably in the other neuroanatomical features outline above) in modern humans. A priori, this variation might or might not be related to behavioral differences. If it is not, it would mean that the benefits to larger brains are not behavioral. Alternatively, one might argue that brain size was indeed adaptive in the past, but is no longer adaptive in modern populations (Martin Daly, personal communication). Perhaps behavioral differences are not related to amount of brain tissue available to the individual beyond a certain critical brain size.
The problem with this scenario is that it ignores the evolutionary costs associated with larger brain size. If increasing brain size has less and less influence on behavioral ability, while at the same time it has greater and greater associated costs, the balance point between increasing costs and decreasing benefits would necessarily occur significantly prior to the point at which brain size no longer has any meaningful behavioral implications. For this hypothesis to work, therefore, it is necessary for there to be no additional evolutionary cost associated with increasing brain size beyond the very same critical threshold at which brain size becomes irrelevant to behavior. If this were not the case, selection would operate to limit brain size to the critical threshold and thereby dramatically restrict brain size variability around this critical threshold. Since there does not seem to be any good reason why we should expect the evolutionary costs to have a critical threshold to begin with (let alone the same threshold as behavioral ability), this hypothesis must be seen as post-hoc and inherently unlikely. The most reasonable starting assumption must be that brain size variation has behavioral implications even within modern humans. This is of course entirely an empirical question, and the research presented below is a further investigation of this possibility.
The abilities generally subsumed under the rubric "intelligence," (e.g., the ability to solve problems, to generalize concepts from specific events, to think analytically, to use language, to learn, and so forth) all seem to relate, at least loosely, to the behavioral differences between humans and other animals. However, exactly how we should measure intelligence is an extremely contentious question. The fact that we have not found a test that everyone agrees perfectly measures intelligence has led some to argue that our concept of intelligence does not actually exist in the real world. However, it is more likely that the concept of intelligence is so contentious not because intelligence does not exist, but simply because it is so very important to us: those who are high in intelligence have clear advantages in our society (see, e.g., Herrnstein and Murray 1994). In any case, it is instructive to look at what evidence we do have with regard to the correlation between correlates of intelligence, measured in admittedly imperfect ways, and neuroanatomical variables.
First, we should look at the evidence from other species. It has been known for some time that species differ with respect to various cognitive domains. Some of these domains clearly do not depend on brain size. Echolocation in bats, for example, is an extremely sophisticated neurological adaptation (Jerison 1985). Upwards of 200 ultrasonic impulses are emitted (and an equivalent number of echoes have to be processed) per second at crucial times, allowing bats to create a three-dimensional map of their environment based on the characteristics of the returning echoes (Griffin 1958, Dawkins 1986). However, bats accomplish this without large overall brain sizes (Jerison 1985).
On the other hand, some types of behavior do appear to correlate with brain size. Because of difficulties in finding behavior patterns which are directly comparable across broad groups of species, Macphail (1982 ) has argued that it is not possible to make any judgments about the relative "intelligence" of various groups of species. However, this is not the relevant question. Given that brain tissue is very expensive (for the reasons outlined in chapter 3), we should expect that there be some benefit to it. Whatever behavioral differences correlate with brain size differences across species are probable clues to the benefits to increasing brain size. In other words, it matters not whether we can make any meaningful judgments about relative intelligence between species, as long as we find some behavioral correlate with brain size. We can then call this behavioral dimension intelligence, or anything we want.
Rensch (1956) attempted to circumvent the problems that Macphail (1982) has outlined by comparing closely related species which vary in brain size. He points to a number of instances in which species-typical behavior patterns (for example, mating displays) tend to be more complicated in larger species (with larger brain sizes) than in closely related smaller species. The same sort of effect has been shown for a number of closely related bird species in general learning capacity (number of different visual cues for which associations can be learned), as well as the length of time that such cues and there learned associations can be remembered (Rensch 1956). He also showed that larger brained ungulates (i.e., elephants) displayed better discrimination learning than smaller brained ungulates (such as zebras; Rensch 1957). These generalities have many individual exceptions, however.
Miller and Tallarico (1974) attempted to extend Rensch's (1956, 1957) argument to perceptual problem-solving, involving detour behavior, in small and large brained species in the family Columbidae (ring doves and pigeons, respectively). Even though ring doves have brains roughly half as big as pigeons in absolute terms (ring doves: 1.13 g; pigeons: 2.38 g), there was no difference in any of the detour-learning measures between the species. The significance of this finding is simply that detour behavior is not related to brain size in these species. As pointed out above, the question should be, what behavioral differences are there between these species? This was not addressed by the study (which understandably focused on one type of behavior).
Riddell and Corl (1977) showed that species differences in a variety of learning tasks correlated strongly with brain size (and relative brain size). While the number of data points was small (e.g., the largest comparison, involving learning sets, included only 11 species), it does suggest that there are general, cross-species behavioral dimensions which are associated with brain size.
Behavioral correlations have also been found at the level of specific brain structures. Devoogd et al. (1993) demonstrated a significant association across 41 species of oscine birds between the size of the brain nuclei related to song production and the size of the song repertoire displayed by the species. Species with larger HVC ("high vocal center") nuclei in relation to overall telencephalon volumes tended to use a larger variety of songs. The correlation was r= 0.56 (p< 0.001), and was independent of phylogenetic relationships among the birds (though this is a dubious evolutionary explanation in any case[2]).
Studies have also shown that the size of the hippocampus (a structure in the medial temporal lobe) is related to the degree to which spatial knowledge is a key component of a species survival. In the study mentioned in the discussion of spatial ability differences between the sexes, Jacobs et al. (1990) report that the polygynous meadow vole (Microtus pennsylvanicus) shows significantly greater sexual dimorphism in hippocampus volume (relative to overall brain volume) than does the monogamous pine vole (Microtus pinetorum). Overall brain volume differences, however, apparently show a different pattern. Jacobs et al. (1990) do not provide the raw data, but it is possible to derive from their figure 1 the necessary numbers to calculate average brain volumes for males and females in each species. While meadow voles have larger brain volumes overall than pine voles (~630 mm3 vs. ~500 mm3, respectively, for average of male and female means), male brains were only 1.6% larger than females in meadow voles but were ~8 % larger in pine voles. Jacobs et al. (1990) state that neither of the within-sex differences in brain volume were statistically significant, however. This indicates that overall brain size is not related to spatial processing in these species.
Comparisons of the bannertail kangaroo rat (Dipodomys spectabilis), which hoards seeds in its own burrow, with Merriam's kangaroo rat (Dipodomys merriami), which hoards food in spatially scattered locations, show that the hippocampus is significantly larger in this latter species (Sherry et al. 1992).
The association of hippocampal volume and spatial ability has also been shown in birds. Black-capped chickadees store food in spatially distributed cache sites, but their ability to recover cached food can be disrupted by aspiration lesion of the hippocampus (Sherry et al. 1992). Comparisons of food-storing with non-food-storing passerine bird species, both from North America and Europe, have shown that food-storers have hippocampi more than twice the size expected for birds with their telencephalon size and body weight (Sherry et al. 1992). Since the food-storers are not more closely related to each other than they are to non-food-storers, this finding cannot be explained as a case of phylogenetic conservatism (which, again, is a dubious evolutionary explanation in any case).
However, the exact nature of the spatial ability presumed to be mediated by the hippocampus in these birds is not clear, since studies of hippocampus size and migratory path length (which obviously requires some form of spatial ability) have shown no association (Sherry et al. 1992). Studies of homing pigeons suggest that the type of spatial ability mediated by the hippocampus has to do with the ability to use familiar local cues for orientation. Most pigeons with hippocampal lesions never reach home, even though they orient correctly when released. Furthermore, the hippocampus of homing pigeons is larger than that of pigeons that have not been breed for homing ability (see references in Sherry et al. 1992).
Finally, there have been two interesting within-species studies on this question, both done on rats. The first involved looking at the brain sizes of two different lineages of rats subjected to 11 generations of selection for either enhanced maze-running ability or decreased maze-running ability (Hamilton 1935). The resulting "maze-bright" and "maze-dull" groups differed by more than ~3.5 standard deviations in maze-running scores. Significantly, these two strains differed more in brain size than in any other anthropometric variable: ~2.5 standard deviations (compared to, e.g., ~1.2 standard deviations in body weight). More recently, Anderson (1993) has shown in rats that a general factor derived from performance on a set of behavioral tasks correlates r= 0.48 with brain size.
If brain size has a causal effect on behavior, we should find that damage to the brain causes decreased behavioral ability, and further that the degree of damage is at least loosely associated with the degree of behavioral deficit. Such findings would not prove that the amount of brain tissue in normal individuals is behaviorally relevant. of course. We know that there are structural features of the brain besides overall size that are critical to behavioral function. One could potentially have all the key structural pathways and networks, yet have fewer neurons subserving each area. Brain damage might be causing behavioral deficits by knocking out or disrupting pathways, with a greater amount of damage affecting a greater number of pathways, even though, under this hypothesis, the total number of neurons is not relevant. However, while an association between the degree of brain damage and the degree of behavioral deficit does not prove that brain size is causally related to behavior, the opposite finding, if generally true, would suggest that brain size is not causally related to behavior.
In fact, the results of a number of studies of different types of brain damage and disease, including Williams syndrome, Down syndrome, Alzheimer's, multiple sclerosis, and schizophrenia, are consistent with the idea that a decrease in the amount of neural tissue is associated with behavioral deficits. These studies use MRI or related techniques to estimate the degree of brain damage or brain size.
In general, mentally retarded children have smaller heads than average (Broman et al. 1987, Hack et al. 1991). Jernigan and Bellugi (1990) found that both Williams and Down syndrome patients had significantly lower overall brain volume than normals of roughly the same ages. Both of these patient groups display significant mental retardation, with IQ's averaging ~50 (3.3 standard deviations below the mean; Bellugi et al 1990). Estimating from the reported voxel counts in Jernigan and Bellugi (1990)[3], the 6 Williams syndrome patients had an average brain size of 1058 cm3 and the 3 Down syndrome patients averaged 991 cm3. By contrast, the 14 normals in the study averaged 1326 cm3. These differences are highly significant, though there was a difference in average age of about 3 years between the normal (mean age 19) and both patient groups (mean age 15.8 for Williams and 16 for Downs syndrome patients). However, brain size in normal 13 to 15 year olds is already more than 95% of the mean brain size for normal 19 to 21 year olds (Dekaban and Sadowsky 1978). In comparison, the Williams syndrome patients in Jernigan and Bellugi's (1990) study averaged less than 80% of the normals, and the Down syndrome patients averaged less than 75% of the normals. Thus, there is an association in between brain volume and mental retardation, though this is of course not proof that the brain size decrease is actually causing the IQ deficits.
Studies of Alzheimer's patients have demonstrated correlations between different neuroanatomical components and behavioral dimensions. Kesslak et al. (1991) found correlations between MRI determined hippocampal volumes in Alzheimer's patients and a number of cognitive tests, including Mini-Mental State ("general cognitive ability") (r= 0.90), Smell Identification Test (r= 0.73), Olfactory Match-to-Sample (r= 0.74), and Visual Match-to-Sample (r= 0.27). They also found correlations between MRI determined parahippocampal volumes and the same tests (Mini-Mental State, r= 0.91; Smell Identification Test, r= 0.66; Olfactory Match-to-Sample, r= 0.73; and Visual Match-to-Sample, r= 0.35). These correlations are inflated because of pooling of dichotomized groups (8 Alzheimer's patients and 7 normals) but still indicate a strong relationship between hippocampal volume and behavioral ability in this disease. In another Alzheimer's study, Rusinek et al. (1991) found a correlation between "global dementia score" (a measure of the degree of severity of dementia) with the percentage of brain volume that is grey matter (r=-0.55,p<0.01), in 14 ambulatory AD patients
Similar findings have been found for multiple sclerosis patients. Swirsky-Sacchetti et al. (1992) found correlations in 40 MS patients between "Lesion Area" in various regions (as assessed from MRI scans by neuroradiologists) and several cognitive tests (corrected for age and education effects). Specifically, left frontal lesion area correlated with Semantic Memory (r= 0.48), % semantic memory retained (r= 0.50), % figural memory retained (r= 0.50), Controlled Oral Word Association Test (r= 0.47), and the Wisconsin Card Sorting Test (r= 0.65). Right frontal lesion area correlated with the Boston Naming test (r= 0.55). Left parietal/occipital lesion area correlated with Verbal Learning Memory (r= 0.49) and the Hooper Visual Organization Test (r= 0.53). Lastly, right parietal/occipital lesion area correlated with Figural Memory (r= 0.62), Recognition (r= 0.54), Digits forward (r= 0.41) and Oral Symbol Digit (r= 0.52).
Callanan et al. (1989) found correlations in 48 MS patients between "MRI total lesion score" and the Wisconsin Card Sorting Test (r= 0.18, p<0.05), and a test of auditory attention (r= 0.20, p<0.05). However, other abilities including IQ deficit, verbal memory, visual memory, and visual attention did not showing significant correlations . Franklin et al. (1988) correlated "whole brain lesion scores" for 60 MS patients (mixed sex) and various behavioral measures including: "Trails A" (r= 0.47, p< 0.0001), "Trails B" (r= 0.43, p< 0.00?), "Verbal Learning" (r= 0.36, p< 0.005), "Non-verbal Learning" (r= 0.35, p< 0.01), "Visual Naming" (r= 0.34, p< 0.01), "Verbal Memory" (r= 0.31, p< 0.01), "Figure Copying" (r= 0.31, p< 0.01), "Symbol Digit Modalities" (r= 0.29, p< 0.05), and "Numerical Attention-Time" (r= 0.28, p< 0.05). Reischies et al. (1988) report correlations in 46 MS patients between the extent of periventricular lesions and overall degree of psychologic disturbance of the current mental state (as assessed by psychiatrists; r= 0.49, p< 0.001).
At least one study of schizophrenia has reported correlations between the gray matter volume deficits and behavioral symptoms. Zipursky (1992) segmented grey and white matter for various subregions of the brain in 22 schizophrenics (and 20 normals) and found a significant correlation within the schizophrenics between the percentage of the intracranial volume that was grey matter and the degree of emotional withdrawal (as indexed by the Brief Psychiatric Rating Scale Withdrawal-Retardation factor; r= -0.47, p< 0.05). Schlaepfer et al. (1994) reported that schizophrenics had significantly less gray matter volume in their heteromodal association cortex (i.e., dorsolateral prefrontal, inferior parietal including the supramarginal and angular gyri, and the superior temporal gyrus).
In addition, Rushton and Ankney (1996) review seven studies of clinical populations which correlated MRI-derived volume estimates (four studies estimated overall brain size, two used a subset of the total set of slices, and one estimated frontal lobe size) with various IQ tests (though one study used educational level as a proxy). The N - weighted mean correlation was r= 0.22. Thus, there is substantial evidence from clinical populations that behavioral deficits are associated with smaller brain size and/or neuroanatomical damage or disease.
The empirical studies of the relationship between brain size and behavior have focused almost exclusively on the concept of general intelligence. The idea that more intelligent people have larger brains has a long history. The earliest attempt to test this was carried out by the inventor of the product-moment correlation: Karl Pearson. Pearson (1906) compared the grades of 1011 university students to linear measurements of their cranial size (controlled for body size) and found a correlation of only 0.11, which was nevertheless statistically significant. Wickett et al. (1994) reviewed 25 studies involving 39 independent samples and obtained an N-weighted mean correlation of r= 0.194 (total N=51,931) between IQ estimates and head circumference. Van Valen (1974) pointed out that if brain size and intelligence were truly functionally related, then the correlations between IQ and head dimensions would necessarily be attenuated because head dimensions are imperfect measures of brain size and IQ is an imperfect measure of intelligence. In other words, if brain size and intelligence really are correlated, then as more accurate measures of brain size and intelligence obtained on a group of subjects, the size of the correlations would necessarily increase.
Several recent studies have shown that brain size estimated in vivo using MRI or CAT scan data correlates with differences in IQ test scores at about r= 0.42, which is more than twice as large as the average found for studies using head circumference as a proxy (Willerman et al. 1991, Andreasen et al. 1990, 1993, Raz et al. 1993, Harvey et al. 1994, Wickett et al. 1994, Egan et al. 1994, 1995, Pearlson 1989; see Rushton and Ankney 1996 for a review). One of these studies (Willerman et al. 1991) included correlations between IQ and two different measures of brain size: 1) outside circumference of the skull, and 2) brain size derived from Magnetic Resonance Imaging (which provides the best estimate of brain size in living humans). The circumference vs. IQ correlation was 0.17 while the MRI vs. IQ correlation was 0.35. These studies suggest that MRI-derived brain size estimates explain about 4 times as much of the variance in IQ than does head circumference, which is further support for the reality of brain/behavior correlations in modern humans. These findings confirm Van Valen's (1974) estimate and strongly suggest that relationships between true brain size and true "intelligence" are likely to be significantly attenuated. It is as yet unknown whether certain subdivisions of the brain are more highly correlated with IQ than others (cortical as opposed to sub-cortical regions, for instance).
However, just as cranial size is an imperfect estimate of true brain size, it is also known that "intelligence" is imperfectly sampled by IQ tests. Not only do IQ tests vary in reliability, but they also reflect aspects of mental ability that were not directly selected for during human evolution (but which we have increasingly put to use in more recent times). It is possible that certain basic properties of the brain were selected for (e.g. the ability to remember in detail the serial order of past events which would increase the ability to determine cause and effect relationships) which then may have led to a general size increase, thereby opening the door for the development of other cognitive abilities (e.g., language).
There is, to my knowledge, only one published study of the relationship between several different behavioral dimensions and brain size. Susanne (1979) correlated circumference of the skull with 5 different psychometric tests (Raven's progressive matrices, and tests of vocabulary, arithmetic, geometrical volume recognition, and elementary and technical knowledge) on a sample of 2,071 Belgian male conscripts. All the correlations with brain size were highly significant (p<0.001). The highest correlation to brain size was with a composite sum of all 5 psychometric tests (r= 0.242), followed by the arithmetic test (r= 0.233), the vocabulary test (r= 0.226), the test of elementary and technical knowledge (r= 0.199), Raven's progressive matrices (r= 0.185), and the test of geometrical volume recognition (r= 0.165). What is perhaps most interesting about this study is the fact that the composite of all these tests was the most highly correlated with brain size (although the statistical significance of the differences in correlations is not given). The implication is that brain size is more highly correlated with the ability to do well on all these tests than it is with any single behavioral dimension. However, since a factor analysis was not run on the data to separate out g (i.e., that portion of the individual variance that is common to a set of tests) from other behavioral dimensions tapped by each test, and since circumference was used instead of in vivo brain size, these results are difficult to interpret.
There is also evidence that variability in corpus callosum cross-sectional area in normal individuals correlates with cognitive abilities. Hines et al. (1992) found two major morphological principal components (corresponding essentially to anterior and posterior portions of the corpus callosum) and three major cognitive principal components (verbal fluency, degree of lateralization of language, and visuospatial ability) in a sample of 28 normal individuals. They found that the posterior callosal factor correlated at r= 0.55 with the verbal fluency factor (p< 0.01) and r= -0.32 with the language lateralization factor (p< 0.05). The anterior callosal factor correlated at r= 0.37 with the visuospatial factor (p< 0.05). However, one well known test of spatial ability - mental rotation (Vandenberg and Kuse 1978) - did not correlate significantly with any portion of corpus callosum cross-sectional area. This indicates that at least some neuroanatomical components are also correlated significantly with cognitive ability.
Lastly, it should be pointed out that there appear to be differences in brain anatomy that relate to differences in specific abilities. At the cytoarchitectural level the degree of dendritic arborization (i.e., number of higher order branches of the dendrites of neurons) there are differences between the left and right hemispheres, and between opercular (Broca's area) and precentral (motor strip) areas (Scheibel 1984). Since the opercular region in the left hemisphere is known to be important in speech processing, this suggests that there are cytoarchitectural differences which are associated with behavioral differences, at least within the cortex. It should be noted, however, that the sample size of this study was very small (N = 8). Jacobs et al. (1993b) report that a number of aspects of dendritic branching in Wernicke's area differ by educational level and by sex. Those with more education had greater total dendritic length, greater mean dendritic length (average length of each dendritic segment), and more dendritic segments overall than those with less education. That is, those with more education had richer and more dense dendritic branching. Extrapolating to other cortical areas, one can suggest a general manner by which brain size would correlated with behavioral ability. The extent to which the findings reported by Jacobs et al. (1993b) are causes or effects of more education is not known. Jacobs et al. (1993b) also show that females had richer and more dense dendritic branching on average than males. Given that females also do better on various tasks of verbal fluency (Kimura 1992), this finding is also consistent with the idea that local differences in dendritic branching correlate with functional differences in behavior.
Scheibel (1984) further reports unpublished data from Oscar and Cecile Vogt on cortical layers (lamina) in two brains from individuals with known talents. One, from an artist with a lifetime capacity for eidetic imagery, had a lamina IV in his primary visual cortex (Brodmann area 17) that appeared twice as large as ungifted controls. The other, from a musician with absolute pitch, had a lamina IV from Heschl's gyrus (primary auditory cortex) that was at least twice the size of the controls. Scheibel reports that these individuals did not appear to have more neurons, but apparently had greater dendritic branching. While two cases are not enough to conclude anything with confidence, these reports are consistent with the suggestion that neuroanatomical differences (even at the level of the neuron) are associated with behavioral abilities.
Further, Schlaug et al. (1995) reported that musicians with perfect pitch have greater leftward asymmetry of the portion of the cerebral cortex which directly subserves music-related abilities (i.e., the planum temporale, a portion of the Wernicke's speech area). Again, these studies indicate that neuroanatomical variability can have behavioral correlates.
There are two conditions in which brain size significantly different than normal reportedly has no behavioral consequence. The first involves midgets, who are claimed (Lenneberg 1967) to have brains roughly the size of modern pongids, yet have "normal" language ability. The word "normal" is in quotations because what is actually claimed is that they have normal ability for children (Lenneberg 1967). However, it is not clear on what basis these claims are made. There are, as far as I am aware, no detailed studies of language ability, IQ, or any other cognitive dimension in midgets.
One study looked at benign megalencephaly in children, in which head circumferences are greater than the 98th percentile for their sex and age. Lewis et al. (1989) compared nine such children to their siblings, who had head circumferences within two standard deviations from the mean for their age and sex. The megalencephalic children were screened for diseases known to increase brain size abnormally, and were given CT scans to rule out such conditions as hydrocephally, etc.. They found no significant differences in IQ, vocabulary, or other tests of cognitive function between the megalencephalic children and their normal siblings. However, there were a number of possible problems with the study. First, brain size was not directly determined: only head circumference was used. Second, specific data on the siblings was not given. Presumably the siblings tended to have larger head circumferences than average as well, even if they were within two standard deviations of the mean, such that the differences between the megalencephalic children and their siblings in head circumference might not have been very large. Third, the average age of the subjects was only 8.6 years for the megalencephalic children and 7.9 years for the sibling group. IQ in younger subjects shows some instability when compared to IQ in 18 year olds (Neisser et al. 1996). Fourth, and most importantly, the megalencephalic children showed other problems that indicate some sort of neurological problem. They performed significantly worse than the sibling group on a test of motor proficiency, four of the nine had a history of speech problems, five had a history of motor problems, and two had a history of attentional problems. If these problems are due to an undiagnosed neurological problem, this might also negatively affect scores on the cognitive tests, causing them to be lower than they might otherwise be. If brain size does correlate with behavioral ability in general, these subject's test scores would reflect a trade-off between neurological problems (depressing them) and higher brain size (increasing them). Thus, this study provides only weak evidence against the idea that brain size is correlated with better cognitive test scores.
[1]For some of the neuroanatomical discussion in this chapter, it may again be useful to refer to appendix E, which contains line drawings of the major external features and landmarks of the human brain. These drawings are reprinted from DeArmond, et al. (1989), and are used by permission of the publisher: Oxford University Press.
[2]As noted in chapter 2, the fact that a feature is found across vast groups of animals does not thereby indicate that the feature is fixed and not adaptive, or that the group sharing the feature should only count as one data point for statisical purposes (which is the rationale behind a number of recent proposals for comparative methodologies).
[3]In the
study, slice thickness was 5 mm, and pixel size was 0.9375 mm2. Slices were
2.5 mm apart. The volumes reported here were estimated by multiplying
the relevant voxel totals by 6.59 mm3 (7.5 mm x
0.9375 mm2),
then dividing by 1000 to obtain cm3.
Copyright 1997 by Paul Thomas Schoenemann