ABSTRACT

INTRODUCTION In this chapter we consider how the rational analysis of the selection task that we have discussed in the last five chapters relates to an alternative approach developed by Dan Sperber, Francisco Cara, and Vitorio Girotto (Sperber, Cara, & Girotto, 1995) in an article entitled Relevance Explains the Selection Task. They present a set of new experiments that they attempt to explain in terms of Sperber and Wilson’s (1986) Relevance Theory, which we discussed in Chapter 5. In this chapter, we suggest that the notion of expected information gain that is central to our rational analysis can be thought of as giving a quantitative explanation of the meaning of “relevance” in the context of this task. Thus, we suggest that a relevance-based account of the selection task is not necessarily an alternative to our rational analysis, but can be viewed as entirely compatible with it. We therefore reconsider the experimental data that Sperber et al. argue favour their relevance account, and show that the data can be modelled successfully using information gain. Hence our conclusion, echoing the title of Sperber et al.’s article, that information gain explains relevance, which, in turn, explains the selection task.