Why did some organisms switch from relying just on reflexive—i.e. purely perceptually-driven—interactions with the world to also employing the tools of representational decision making? What adaptive and other benefits does the reliance on representational decision making yield? Today, I sketch aspects of the answers to these questions; for more details, see chapters 4-6 of my book.
To set out an account of the evolution of representational decision making, it first needs to be noted that it is widely acknowledged that representational decision making comes with a number of cognitive costs: it tends to be slower than reflexive decision making and require more cognitive resources such as concentration and attention. Given this, what benefits does representational decision making yield that can, at least at times, outweigh these costs—and thus explain its evolution?
When it comes to the evolution of cognitive representational decision making, the key point to note is that it allows organisms to go from a table of reflexes that is likely to contain much redundancy to one that is much smaller. Purely reflexive organisms will often need to associate many different perceptual states to the same behavioral response in order to behave adaptively. (This is a key insight of Sterelny’s 2003 account of human cognitive evolution.) Representational decision makers, though, can avoid this redundancy: they can determine the appropriate behavioral response to the world by first “chunking” the information received from their perceptual systems, and then reacting to combinations of these “chunked” states. In turn, this brings two kinds of adaptive benefits to these organisms. First, it can make it easier for them to adjust to changes in their environment: organisms do not need to alter the large number of associations between perceptual states and behavioral responses, but merely the (generally) fewer associations between representations of the state of the world and behavioral responses. On the other hand, cognitive representational decision making can streamline an organism’s cognitive and neural system: reliance on chunked information can enable the organisms to make more efficient use of its neural resources.
Something very similar is true when it comes to the evolution of conative representational decision making. However, instead of streamlining a table of behavioral responses, it enables organisms to avoid the table altogether, and just compute the response to the situation. This, too, can make it easier to adjust to changed environments and helps organisms to streamline their cognitive and neural systems.
Given all of this, when is there an adaptive advantage in cognitive and / or conative representational decision making? Abstractly speaking, this will be the case when the costs of representational decision making—losses in decision making speed and increased use of cognitive resources—are outweighed by its benefits—easier adjustment to changed environments and increased cognitive and neural efficiency. What does this mean more concretely? It suggests that representational decision making is adaptive in cases where adaptive behavioral responding cannot be made dependent on just a few perceptual states, where organisms can compute their behavioral response relatively easily, and where the environment changes relatively frequently. Examples of such cases are certain kinds of social, spatial, or causal environments. For example, consider the classic challenges of great ape social living. In these environments, the choice of which coalition to join (say) is an adaptively important decision associated with a large and complex set of perceptual contingencies that needs to change frequently (e.g. with changes in the composition of the group), but where this decision can often be relatively leisurely computed by attending to surprisingly few variables, such as coalition size and one’s place in the social hierarchy.
Interestingly, this conclusion of which environments favor representational decision making matches the implications of other, related accounts—such as those of Sterelny or Millikan. However, here, this conclusion is arrived at very differently from what is the case in these other accounts: on my account, the evolution of representational decision making is driven not by enabling organisms to do things that purely reflexive organisms cannot do—rather, it is driven by it enabling organisms to do the same things better. In short: despite appearances to the contrary, representational decision making is—in certain circumstances—more, not less, efficient than reflexive decision making. In the next and final guest post, I will bring out some implications of this conclusion.
 See Sterelny, K. (2003). Thought in a Hostile World. Oxford: Blackwell.
 A. Whiten and R. W. Byrne, eds., Machiavellian Intelligence Ii: Extensions and Evaluations (Cambridge: Cambridge University Press, 1997).
 Ruth Millikan, Varieties of Meaning (Cambridge, MA: MIT Press, 2002).