Good habits, better heuristics

Posted on March 31, 2008


Here’s an abstract for a paper that I will present a couple of times this year, including at the European Conference of Analytical Philosophy in Krakow and at the Naturalism and Bounded Rationality workshop in Granada:

Dual process theories of rationality that argue for the co-existence of mutually complementary analytical and intuitive cognitive processes have been fairly popular recently. Although the models vary, the theories often appear to be an attempt to combine a fairly traditional view of rationality as the ability to use logic with something like the cognitive heuristics that Kahneman and Tversky’s empirical programme revealed as exceedingly common. These theories hold a definite attraction, offering as they do a way to retain something of a much valued view of rationality in the face of difficult empirical results. Indeed, some might argue, the dual process approach has a long and distinguished philosophical history if Hume’s distinction between reason and habit is understood in these terms. Unfortunately for dual process theorists, the arguments behind Hume’s rejection of what he called reason are just as powerful today as they ever were. What is more, dual process approaches do not actually offer any more than can be provided by a fully developed heuristics approach such as the bounded rationality account originated by Herbert Simon and further developed by others such as William Wimsatt. This account goes beyond the very simple examples studied by Kahneman and Trevsky as well as Gigerenzer to develop a model of how heuristics can combine and develop in an open-ended fashion to be able to deal with the kinds of problems that the dual process theories take for granted as beyond the reach of heuristics.

If this view of heuristics is correct then they can be understood as the habits that Hume, ultimately, placed his faith in. Importantly, the bounded rationality approach makes it reasonable to expect success from these limited means even while Hume’s problem stands, putting a universal method out of our reach.