Abstract
In his 1988 review of On the Plurality of Worlds (Lycan [1988]), William Lycan argued that what he called Lewis's 'mad-dog modal realism' (also 'rape-and-loot modal realism' and 'nuclear-holocaust modal realism' - I suspect that some reference to the supposed extremity of Lewis's position is intended) rested upon an unanalysed modal notion. Lycan accepted that actualists all seemed to be stuck with such unanalysed notions (adding that his own was the notion of compatibility as applied to pairs of properties), but argued that Lewis's notion of worlds was also a modal primitive: 'World' for him has to mean 'possible world', since the very flesh-and-bloodiness [which relieves him of the sort of abstraction indulged in by actualists] prevents him from admitting impossibilia. (Lycan [1988], p.46) Lycan's main concerns in this review go back to his earlier paper 'The Trouble with Possible Worlds' (Lycan [1979]), and are taken up again in his PAS paper: The ruling out of impossible worlds is a serious liability [...] For semantics needs impossible worlds. Though standard modal logics may trade just in possible states of affairs, the semantics of conditionals must deal with inconsistent beliefs. (Lycan [1991], p.224) He goes on to claim that the actualist has no problem with impossible worlds. An impossible world is just - e.g. - a set of propositions (one of which happens to be inconsistent). (loc.cit.) Whatever the truth of this in principle, most actualists have either explicitly or implicitly excluded possible worlds from their theories.* It is true, nevertheless, that Lewis has a clear problem with the very idea of worlds at which logically incompatible propositions are true. Lycan attempts to exploit this as follows.