Abstract
The "Chinese Room" controversy between Searle and Churchland and Churchland over whether computers can think is subjected to Derridean "deconstruction." There is a hidden complicity underlying the debate which upholds traditional subject/object metaphysics, while deferring to future empirical science an account of the problematic semantic relation between brain syntax and the perceptible world. I show that an empirical solution along the lines hoped for is not scientifically conceivable at present. An alternative account is explored, based on the productivity of neural nets, in which the semantic relation is found to be dynamical - a spontaneous, stochastic, self-organizing process