Selective changes in brain region size between taxonomic groups are called grade shifts (Barton and Harvey, 2000
). Primate suborders exhibit a number of grade shifts in brain and brain region size. Haplorhine primates (i.e., new-world and old-world monkeys and apes) exhibit a disproportionately enlarged brain and a disproportionately enlarged isocortex relative to that of strepsirhine primates (i.e., lemurs, lorises, galagos; and ; [Barton and Harvey, 2000
], [Finlay et al., 1998
] and [Stephan et al., 1981
]). Both haplorhine and strepsirhine primates exhibit a disproportionately enlarged isocortex relative to many other mammals. However, within each primate suborder, the size of the isocortex is extremely predictable when regressed against the rest of the brain ().
Comparative developmental studies in primates and other mammals show that the grade shifts just described in isocortex size may arise due to selective alterations in the timing of developmental schedules. Comparative analyses of isocortex generation and development showed that haplorhine primates (i.e., rhesus monkeys, humans) selectively delay isocortical neurogenesis compared to rodents (i.e., rats, mice, hamsters, spiny mice, guinea pigs; [Clancy et al., 2000
], [Clancy et al., 2001
], [Clancy et al., 2007
] and [Finlay et al., 1998
]). Relative delay in isocortical neurogenesis entails that the isocortical progenitor pool population will multiply exponentially relative to other nondelayed structures and the isocortex will expand in neuron number and size. Note that we have described at this point two separate, but logarithmically additive, ways of increasing relative cortex size. Increase in duration of development alone to produce a larger brain, with rate of cell production unchanged, automatically increases the relative proportion of the isocortex. The primate isocortex expands disproportionately still more, adding to the fundamental nonlinearity of allometric scaling an increase in the dedicated precursor cell pool for the cortex, by delaying isocortical stem-cell cessation with respect to the schedule established in rodents and insectivores, and thereby producing a “grade shift.”
Some intrinsic difficulties on the use of residuals and ratios in allometric studies In a series of studies beginning in the early 1990s, Dunbar and his colleagues showed that the relative size of the isocortex or brain size and residuals derived from a linear regression of brain and body size positively correlate with group size or related measures of group size in primates ( [Dunbar, 1992
], [Dunbar, 2009
], [Dunbar and Shultz, 2007a
], [Dunbar and Shultz, 2007b
] and [Pérez-Barbería et al., 2007
]). In parallel with statistical practices in the field, the first studies looked at basic regressions between two variables. The next set of studies used more elaborate multiple regression techniques and phylogenetic contrasts to eliminate the statistical problem of non-independence of taxonomic relationships. Recent studies attempted to determine aspects of temporal emergence of the correlated features examined using discretized variables in conjunction with extensive phylogenetic analyses (Pérez-Barbería et al., 2007
). As statistical analyses flourish, it is rare to see any representation of primary data, and the basic “visual” sense of the strength of association, magnitude of results, or amount of variation has tended to fade. In this chapter, following the historical progression of the analyses described, we will plot the basic data relating brain, body, and group size in primates, then add in taxonomic variability, and finally consider the range of variation and a few of the measurement issues in group size, but will go back and plot the basic data on which these claims are established. We should emphasize that we do not contest that there is a relationship between relative brain size, and (possibly) relative isocortex volume and social complexity, generally speaking. The sophistication of the statistical analyses is undoubted. What we do contest are the basic assumptions of the techniques, the causal relationships implied, and the claim that the relationship between social competence and relative brain size, compared to any of a number of other measures of behavioral complexity, is unique.
First, we describe problems with statistical comparisons between groups involving basic allometric relationships between brain parts, to set out very basic issues, which antedate the social brain hypothesis. A number of studies examining the potential mechanisms underlying species-specific adaptations or developmental disorders have focused on the relative sizes of parts of the brain. The initial problem (not a problem of the social brain studies) is “relative to what”? For example, suppose it is shown that the relative size of the frontal cortex is greater in autistic individuals relative to healthy individuals ( [Carper and Courchesne, 2005
] and [Courchesne et al., 2011
]), even correcting for a somewhat greater brain size in the autistic group by taking a ratio of frontal cortex to brain volume overall. If individual variation in humans follows primate brain allometry, increases in brain size will produce an even greater increase in the proportion of cortex, and frontal cortex will be a greater proportion still (). “Correcting” for brain size by taking a simple ratio of frontal cortex to brain size between two groups with differing brain sizes will invariably demonstrate relatively more frontal cortex in the group with the larger mean brain size, but this is simply a predictable outcome of the underlying allometry and no indication of any unusual hypertrophy or pathology of the frontal cortex. Although this problem plagues a number of comparative studies in which two species are compared, or brains with a developmental disorder that are compared to normal brains, fortunately, for studies of primate brain evolution, we have ample information to be able to predict the different allometries of various brain divisions.
The use of residuals derived from allometric equations relating the size of two structures is a common method to compare brain region size across species. In the case of the social brain hypothesis, the finding that residuals derived from the linear regression between brain and body size correlates with group size must account for grade shifts in brain size between haplorhine and strepsirhine primates, and when that is done, a significant statistical relationship remains (Pérez-Barbería et al., 2007
). The brains of haplorhine primates are disproportionately enlarged relative to those of strepsirhine primates. However, brain and body size strongly covary within haplorhine and strepsirhine primates. Fitting a linear regression through brain and body size in primates (haplorhines and strepsirhine primates) would fit a linear regression with a different slope and intercept than those obtained by fitting two separate linear regressions through the brain and body of haplorhine and strepsirhine primates (). Returning to the basic data, we look at the amount of association between relative brain size and social group size in these two taxonomic groups. In the case of the social brain hypothesis, the brain versus body residual values obtained for both haplorhine and strepsirhine primates correlate more strongly with group size than does the brain versus body residual values derived from separate linear regressions for haplorhine versus strepsirhine primates (Pérez-Barbería et al., 2007
). However, the correlation coefficients derived from the residuals of brain to body size in both scenarios are surprisingly low, and it would appear that something about the grade shift in relative brain and cortex size between these two taxa is accounting for most of the effect. Further analyses from the same laboratory group and others considering phylogenetic contrasts, other behavioral measures, and more elaborate statistics generally demonstrate a statistically significant but clearly very small effect in residual change in brain size ( [Barton, 1993
] and [Dunbar, 1993
Fig. 5 Residuals derived from a linear regression through the brain size and body size are correlated against group size in primates. In one scenario, residuals are derived from a linear regression through the brain and body size of both haplorhine and strepsirhine (more ...)
We return to the basic data in of the regressions of primate body, brain isocortex volume, and “isocortex ratio” with group size (taken from Dunbar, 1992
). The social brain hypothesis posits that isocortex (either volume or ratio) and group size are positively correlated (isocortex volume residualized with respect to both body size and brain volume). Of interest here is the relationship of body size to social group size, which opens the possibility of many other causal routes between brain size and group size than the ones mentioned here—for example, niche, range size, and type of food consumed.
Further, consider other behaviors demonstrated to vary with brain size: “innovation” in the wild, successful invasion of new territories, residual mortality when corrected for body size, and laboratory measures of learning ability. These all correlate with each other and with relative brain size ( [González-Lagos et al., 2010
], [Lefebvre and Sol, 2008
], [Lefebvre et al., 2002
] and [Reader et al., 2011
]). The strength of the association between isocortex size and group size that would be left after partialling out capability to innovate, or general learning ability, seems unlikely to be significant. Put another way, we would suggest that virtually any reasonable measure of cognitive or behavioral complexity—working memory grammatical sequence learning, innovation and so on—would show the same relationship to relative brain size.