It is immensely difficult to give a philosophically adequate account of mental imagery. Peter F.R. Haynes (Mental Imagery, this Journal 6 [1976), pp. 709–19) objects to the standard accounts, and offers one of his own which avoids the standard difficulties. Unfortunately it in turn seems to lapse into incoherence.

Haynes rejects Cartesian accounts which would make images private objects in non-physical space. He also rejects current alternative views: both Rylean or behaviourist ones; and also intentionally complex ones, which assert that the relevant terms change their meaning (so that “red” means something different when applied to red balls and red images). He thinks that J. J. C. Smart's “topic neutral” approach may supply “the basis for a correct, intentionally simple, intuitively satisfying account of mental imagery” (p. 707). But an adequate account must also be neutral, as Smart's is not, between different theories of mind (pp. 707–8). In addition to avoiding these difficulties an adequate theory has certain other desiderata; in particular Haynes places valuable emphasis on the need to explain the fact that we can sometimes mistake images for real perceptions or, more real perceptions for images.