Differences in evolutionary patterns and rates of general cognitive ability compared to neuroanatomical indicators in the primate phylogeny

Utilizing a sample of 174 primate species from all superfamilies we tested the following predictions based on recent findings involving coefficients of additive variance in the homininan lineage: (1) evolutionary rates for... [ view full abstract ]

Utilizing a sample of 174 primate species from all superfamilies we tested the following predictions based on recent findings involving coefficients of additive variance in the homininan lineage: (1) evolutionary rates for inter-specific general cognitive ability (G) will be faster than those for neuroanatomical indicators (NIs - which are frequently used as proxies for G in multi-species comparative analyses), (2) G will be less phylogenetically conserved, and (3) evolutionary models fitted to G and to NIs will present at least moderately distinct parameter values. We obtained species means for G and for the most-used NIs in comparative studies: neocortex ratio (neocortex size/size of rest of brain), neocortex residualized (and phylogenetically residualized) against body size, brain size likewise residualized both ways, absolute brain size, and absolute neocortex size. Evolutionary rates were significantly faster for G than for NIs residualized against body size than for absolute NIs and for body size. Phylogenetic signal (PS) was significantly different from zero and from unity for G and for residualized NIs using Pagel’s lambda but was smaller for G, and was equal to unity for absolute NIs and body size. Using Blomberg and colleagues’ K, PS was nonsignificantly different from zero only for G, suggesting G is not importantly more similar among closely-related primate taxa than among distantly-related ones, whereas K surpassed unity only for absolute NIs and body size, indicating phylogenetic conservatism. To test prediction three we analyzed the fit of each variable to the Brownian-motion, lambda, acceleration, stabilizing selection, and Early-Burst models. Although relative AIC model weights for G did not differ strongly from those for body size-corrected NIs (with the lambda model as best fit), G presented a delta parameter-estimate around ten times larger, indicating stronger rate acceleration. The relative model weights for absolute brain and neocortex sizes greatly differed from the other variables however, and strongly conformed to the patterns presented by body size (Early-Burst and Brownian-motion models fitted best). Finally, G presented selection regime shifts in different nodes from NIs. Predictions were largely corroborated.