The results of this study raise a number of important questions, which in turn suggest specific directions for future research. Answers to these questions will ultimately get us closer to full understanding of hominid evolution.
First, it is important to replicate the within-family design of the study, preferably with greater sample sizes, with males as well as females, and with subject populations that display a broader range of ability and/or brain size. Secondly, it would be useful, as was pointed out in the last chapter, to tap a number of cognitive abilities not directly tested in this study, specifically: Better measures of social ability, tests of breadth, depth, and complexity of memory (involving the complexity of semantic relations, in some form), and tests of serial order memory. Thirdly, some method of controlling for any within-family effects between siblings should be used. Ideally, identical twin reared in separate homes would be used. The correlation between one twin's cognitive test score and the other's brain anatomy would be a direct estimate of the genetic correlation between these to variables. To the extent that within-family effects (such as those caused by sibling rivalry) are randomized, this would be ideal. One could check for such effects by looking for differences between older and younger siblings (relative to the twins' adoptive siblings), similar to the methods used in this study to suggest some form of sibling interaction.
Another direction to pursue, which can be done directly without any further data collection, is to investigate other specific neuroanatomical structures for potential association with the cognitive variables included in this study. Neuroanatomical variables for which it would be possible to quantify with presently available software tools include: corpus callosum size and shape (which is a subject of current interest in hominid neuroanatomical evolution), frontal lobe cortex volume, cerebellum size (given the gross association between cerebellum size and linguistic ability seen in Williams syndrome patients: Jernigan and Bellugi, 1990), cortical surface area (including areas such as the frontal lobe), planum temporale, hippocampus (though this may not be feasible given its small size relative to the resolution of the scans), and sub-cortical structures such as the basal ganglia.
Also of interest would be to do similar studies with functional brain imaging: one could investigate the extent to which total area of brain activation on a certain task varied with brain size, instead of measuring only brain anatomy. This would get at how the brain might use extra tissue, as opposed to the more basic question of whether more tissue is associated with increased behavioral ability. It appears that increased ability on a task is associated with decreased cortical activation (e.g., Haier et al. 1988, Haier et al., 1992). Furthermore, brain glucose consumption appears to be inversely related to overall brain size (Yoshii et a. 1988). This can be reconciled with the idea that amount of neural tissue is important only if the extra neural tissue aids in learning (and then becomes unnecessary once this is accomplished), or, as discussed above, it results in greater richness and complexity of cognition or memory.
Another functional imaging study of importance would be to address the question of what areas of the brain are activated by processing increasingly complex syntax. The syntax test invented for this study, which compares two types of sentences identical in every way except their syntax, would lend itself beautifully to a functional imaging study. Brain activation caused by the two different sentence types could be subtracted from each other to determine the difference (if any) in functional areas activated by the two types of syntax. This might reveal a great deal about how language is processed in the brain, and ultimately, how language evolved.
Finally, it would be valuable to construct mathematical models of brain size cost/benefit tradeoffs, including basic empirical cross-species data such as the apparent increase in metabolic cost, decreased number of offspring per unit time, etc.. The idea would be to get some idea of the necessary evolutionary value of the benefits, and might give us a clearer picture of exactly how strong selection would have to be to accomplish the changes observed over the three million years of hominid brain size increase.
Copyright 1997 by Paul Thomas Schoenemann