For a number of years, I have argued that the child is a spontaneous theoretician, that is, the way in which children go about discovering how the world functions (the physical, social, and linguistic worlds) is by building theories, not by simply observing facts (Karmiloff-Smith, 1975, 1979a, 1984; Karmiloff-Smith & Inhelder, 1974175; see, also, Brown, 1986; Carey, 1985; Gelman & Brown, i985; Gillieron, 1982; Gopnik, 1982; Keil, 1988; Leslie, 1988). And children go about their spontaneous discovery task by behaving like the typical scientist. Kuhn (1977) was right in his view that only on the very rare occasions when scientists must actually choose between competing theories do they reason like philosophers or logicians! Both for the child and the adult researcher, scientific progress does not stem from the use of logical criteria on the basis of rational induction from observations. Indeed, as Lakatos (1977) has pointed out, the still widely held opinion that science progresses by repeated overthrow of theories on the basis of observational data is based on two false assumptions: (a) that there is a natural psychological boundary between the theoretical and the observational, and (b) that if a proposition satisfies the psychological criterion of being factual or observational, then it is true. Likewise Quine has argued that the data/theoryconfirmation relation is a property of the total system of beliefs in which it is embedded (Quine, 1953).