The behavioral changes that are observed during conditioning procedures are mediated by underlying psychological mechanisms which cannot be directly observed but have to be inferred from the results of experimental manipulations. Because these underlying mechanisms are not open to direct observation, many learning theorists have been content to understand learning at the level of controlling variables: that is, they were willing simply to catalog the regularities between environmental input into the organism and the subsequent behavioral output (e.g. Skinner, 1950, 1953). This approach was typical of the behavior analysts of the 1960s and early 1970s and gave rise to the intense period of research on, for instance, schedules of instrumental reinforcement during these years. Nevertheless, although analyses of learning in terms of controlling relationships are quite valid, the inquiring scientist should still be curious as to the psychological mechanisms that mediate these relationships.