ABSTRACT
Chapters 2±5 illustrated the complex and rather contradictory patterns of evidence about the mindreading abilities of children, adults and non-human animals, which provided the justi®cation for a hard look, in chapter 6, at just what we mean by mindreading, and how such mindreading might ever be achieved by real cognitive systems. In chapter 6, I argued that human adults have two kinds of cognitive process for mindreading ± ``low-level'' processes that are cognitively ef®cient but in¯exible, and ``high-level'' processes that are highly ¯exible but cognitively demanding. And, equally importantly, I suggested that we also have a large amount of social and socially constructed knowledge, and a range of other cognitive processes that do not entail mindreading as such, but which underwrite a large proportion of everyday social cognition. I then sketched how these abilities might be organized, and the general kinds of question that this should lead us to ask. In this ®nal chapter I want to pursue the implications of the model developed in chapter 6 in more concrete terms, before comparing it brie¯y with existing theoretical alternatives.
