In my last post, I wrote that the part two of my introduction to the cognitive science of religion was to soon follow. Since no one has, or likely will, ever read it, I’m postponing that promise so that I might write about something else that I’ve been reading reading about lately.
I’ve been going back through the piles of articles that I’d printed or downloaded before I lost access to the University of Minnesota’s journal subscriptions. One article was a Behavioral & Brain Sciences target article from 2010 by Michael L. Anderson reviewing several recent theories of neural reuse, including the author’s own ‘massive redeployment hypothesis’ (MRH). I had previously written a term paper on a different neural reuse hypothesis, Stanislas Dehaene’s ‘neural recycling hypothesis,’ so the topic was one with which I had some familiarity, though it had been a few years since I had read his papers. Since the recurring theme of this site is the intersection of evolution, cognition, and culture, I thought I’d take the time to roughly outline Anderson’s MRH and some of the evidence that he presents, and perhaps in a different post I’ll do the same for Dehaene’s neuronal recycling hypothesis which has been valuable in the explanation of the development of written language and higher mathematics.
Psychologists interested in many aspects of human cognition have had to confront a difficult problem: how do brains acquire new abilities? Neurologists have shown for well over a century and a half how damage to specific brain areas prevent people from doing certain things. For example, regardless of the keenness of one’s eyes, if a person has damage to the primary visual cortex in the far back of the brain, they will lose their sight in the parts of the visual field to which that cortex maps. Commonalities like these have left psychologists with little doubt that brain functions map on to specific brain regions, as opposed to the older idea that the whole brain can be used for every task, such as Karl Lashley’s ideas of the law of mass action and the principle of equipotentiality in the early 20th century. The difficulty, however, comes when we try to understand how the brain can come to do new things.
In evolutionary terms, how is it that we can explain the introduction of new abilities? Take language as a prime example; how is it possible that language could ever have been developed? Language is associated with one of the more complex networks of activated brain regions when observed through fMRI scans. One could propose that a miraculous genetic mutation, or suite of mutations, led to the construction of brain regions primed to process language. The contents of a language are learned, however, so that first individual to develop language processing abilities by happy accident would have had no language to speak. Without any use for those brain regions, the chances are high that those mutations would have been weeded out by natural selection.
Neural reuse hypotheses offer an explanation as to how this process occur. They are predicated on a model of the brain in which complex brain behaviors are made up of many smaller computational processes that come together to make up those processes. This is known also as modularity, which is an analogy from computer science and artificial intelligence in which complex processes are broken down into smaller and more manageable processes. This is more than just a convenience, since in many cases it would be impossible to program those complex processes without such deconstruction.
In general neural reuse hypotheses propose that when a new cognitive function is developed, either on an evolutionary time scale or during one’s life, the brain predominantly reuses many of the existing micro-processes as it can to carry out the new action rather than rely on brain evolution to create new brain areas. These micro-processes are drawn upon from all over the brain, which complicates the idealization that the brain is divided into larger sections that each perform a certain kind of function (e.g., the occipital lobe carries out the processes involved in vision and the primary motor cortex carries out the processes involved in motion). This is conceptually similar to the idea of exaptation in evolutionary biology, in which one structure (perhaps skeletal or an organ) is changed into a new structure.
As an example, Anderson points out that although Broca’s area has historically been strongly associated with language, it is also activated during fMRI scans of people performing many different tasks, such as movement preparation, action sequencing, action recognition imagery of human motion, and action imitation. Broca’s area’s involvement in linguistic tasks has long been a prime example of the clear-cut mapping of the brain, yet in practice, it appears to be involved in a number of other tasks that are unrelated to language.
Anderson’s paper outlines four specific recent theories of neural reuse, but I just want to summarize Anderson’s own and give some of the evidence that supports it. His theory is new enough that the evidence remains slim. However, it is an interesting idea, and if true, will be of great importance in understanding the evolution of the brain.
The MRH is a simple idea that proposes that when the brain attempts to acquire a new ability, it draws as much as possible from its existing micro-processes scattered around the brain. He validates this idea with an analogy from computer programming. Software developers have the fortunate ability to repeatedly link to (or “share”) an individual substructure from different parts of the program without those parts interfering with each other. This makes the program much more efficient, since certain sub-processes to not have to be repeatedly rewritten and stored independently in memory. This does not preclude the brain from evolving new structures, it just states that as much as possible the brain will try to make due with what it already has.
Anderson lists three predictions that follow from this proposition. First, brain regions should tend to be activated during many different tasks in different domains (perception, attention, decision making, etc.), as opposed to having brain regions dedicated to a specific type of function (such as the idea that Broca’s area only did language). Second, as time passes since a brain region evolved, it should be increasingly ‘redeployed’ to perform different functions. The older the structure, the more tasks in different domains with which it should be associated. Lastly, the more recently a cognitive ability has developed, the more numerous and widely dispersed we should expect its associated brain regions to be. The later a function is developed, the more likely that the micro-processes will already be available in the brain to perform the new function, however widely they are spread across the cortex.
Anderson offers a few of his recent studies as evidence to support his three predictions. He supports the first prediction by citing a study in which he reviewed 1,469 subtraction-based fMRI studies (activation levels during a control trials are subtracted from activation levels during the experimental trials to show that the remaining activation is associated with the task) of eleven different task domains. These cognitive functions were diverse, including action execution, action inhibition, action observation, vision, hearing, attention, emotion, language, mathematics, memory, and reasoning. Using a system in which the brain is divided into 66 cortical regions – or 33 regions from each the left and right hemispheres – he found that each region was activated on average by tasks in nine of the eleven different domains listed. Those data, however, appear to be distributed through three studies of his that he cites, since none of them include the statistics that he reports in BBS. Even more surprisingly, he cites unpublished data that show that his results were the same even if the cortex is divided into 1,000 regions – 500 in each hemisphere – rather than 66.
It is possible, however, that it is not neural reuse that is scattering brain functions around the brain. There is a possibility that those new micro-processes along with their brain regions – which may be tiny – are evolved anew every time, and that there is no real logic behind where they are placed. Though there have been great developments in fMRI technology, its spatial mapping is still rough and cannot distinguish between activation patterns by different neural networks that are packed closely together, especially since they may be linked by the same blood vessels. This is unlikely, since that would lead to a lot of waste that natural selection tends not to tolerate, though we don’t understand very well the forces that govern the evolution of the brain.
The second prediction was addressed by citing a study in which he attempted to correlate the age of brain regions with how extensively they are associated with tasks in different domains – a proxy for their rate of neural reuse. The problem he had to overcome, however, was that there are several opposing ideas of the ages of different brain regions. He attempted to overcome this is by declaring that, in general, the regions in the back of the brain tend to be older than the regions in the front. I assume that he is drawing upon the fact that the frontal cortex in humans has proportionally increased much more in size than any other cortical region, making more of the frontal lobe ‘newer’ than any other. He found that there was a correlation between a brain region’s position on an axis that runs from front to back and how much that area appears to multitask. This was a surprising correlation, in fact, since most scientists would have probably predicted that the greatly expanded frontal lobe would have been activated by more diverse tasks than the occipital lobe.
This is a rough approximation of age, however, and its results are arguable. Although the frontal lobe has expanded to a greater extent than the other lobes, all of them have undergone a large expansion since the last common ancestor (LCA) with chimpanzees. It is estimated that the LCA with chimpanzees had a brain size that is approximately identical to chimpanzees – around 400 to 500 cubic centimeters, whereas the modern human average is around 1200 cubic centimeters. Even in the least expanded regions, there has considerable increase. Given this threefold overall increase, it may not be reasonable to say that any region is older than any other, since all regions are significantly new.
Lastly, he approaches the third prediction by offering some confirming evidence from one of his recent studies. There does seem to be a relationship between how recently brain abilities developed and how many brain regions they appear to use. Language showed the largest and most widespread pattern of activation followed, in order, by reasoning, memory, emotion, mental imagery, visual perception, action, and attention. This is striking because many scientists might have suspected, as I did, that the older cognitive abilities would activate more widespread regions, since (1) the brain regions used for newer abilities would have to form around the brain regions used for the older abilities, forcing the older ones out into widely distributed areas, and (2) as older functions improved, their brain regions would have to fit in wherever possible, resulting in a more widespread pattern of activation.
On the other hand, some cognitive functions are more difficult than others, and may require more brain regions to be active in order to perform its task. This descending order of functions may just represent a descending order of complexity. Or, as time went on, the regions associated with older abilities may have had enough time to become more locally concentrated, since there are costs to long distance communication between associated regions.
As of yet, these data still aren’t enough for us to know if the MRH is correct, and I hope that it continues to be investigated more thoroughly in the future. Were the MRH found to be correct, scientists would have a much better understanding of how modular functions are organized in the brain, which has been a contentious topic for decades in cognitive neuroscience.