In a recent paper Eysenck asked why conditioned responses show incrementation. In several publications he has provided the outline of an answer, which he has used in proposing a theory of the development of neuroses. It is suggested here that in the way he sets out the problem and in the development of his theory of incubation he overstates the applicability of the theory, incorporates questionable assumptions, and omits thorough consideration of competing accounts. The theory focuses on classical conditioning to the near-exclusion of other processes of behaviour change, ignores the consequences and social context of neurotic behaviour, and provides an incomplete discussion of the temporal patterning of relevant events. By taking a view which is broader in terms of psychological processes, environmental events, and temporal sequences, alternative models of the development of phobias can be offered. These involve classical and operant conditioning, and observational and informational learning.