Introduction. Whatever its underlying causes, even the description of the phenomenon of thought insertion, of the content of the delusion, presents difficulty. It may seem that the best hope of a description comes from a broadly cognitivist approach to the mind which construes content-laden mental states as internal mental representations within what is literally an inner space: the space of the brain or nervous system. Such an approach objectifies thoughts in a way which might seem to hold out the prospect of describing the ''alienated'' relation to one's own thoughts that seems to be present in thought insertion.1 Method. Firstly, I examine the general structure of cognitivist accounts of intentional or content-laden mental states. I raise the general difficulty of explaining how free-standing, and thus world-independent, inner states can still have bearing on the outer world. Secondly, I briefly examine Frith's model for explaining thought insertion and other passivity phenomena by postulating a failure of an internal monitoring mechanism of inner states. I question what account can be given of non-pathological cases and raise two specific objects. Results. Cognitivist accounts of the mind face a general, and possibly insuperable, challenge: explaining the intentionality of mental states in non-intentional, nonquestion-begging terms. There have so far been no satisfactory solutions. Cognitivist accounts of passivity phenomena in terms of a failure of internal monitoring face two objections. Firstly, accounting for non-pathological cases generates an infinite regress. Secondly, no account can be given of the paradoxical nature of utterances of the form of Moore's paradox: ''it is raining but I do not believe it''. Conclusions. A cognitivist approach presents an alienated account of thought in normal, non-pathological cases and is no help in accounting for thought insertion.