Situation, Reason and the Extended Agent

Findings in cognitive science show that behaviour is regularly influenced by situational features. It has been maintained that because agents would reject these features as reasons, they do not act freely and agentially when so influenced. I argue that this view is mistaken, in that it (i) rests on a conflation of three distinct grounds on which to reject reasons; and (ii) takes too short-term a view of what constitutes an episode of agentiallysignificant behaviour.