ABSTRACT
In a seminal paper, Trapold (1970) asked if animals’ expectancies of different reinforcers are discriminably different. He reasoned that the answer to this question might be evident in performance on a two-choice conditional discrimination task in which the correct response (R1) following one conditional cue (S1) always produces a particular outcome (O1), while the correct response (R2) following the alternative cue (S2) always produces another outcome (O2). With experience on this procedure, the subject could correctly anticipate the outcome appropriate to each kind of trial at the moment the trial begins. In other words, S1 could trigger an expectancy of 01 (E1), whereas S2 could trigger an expectancy of O2(E2). In keeping with the tradition of Hull-Spence rg−sg theory, Trapold assumed that the stimulus properties of expectancies become associated with whatever response is reinforced in their presence. To the extent that such expectancy stimulation is salient and varies with kind of outcome expected, the differential conditioning of expectancies to the conditional cues should augment the stimulus differential between the two kinds of trials and thus function to support higher levels of performance relative to animals for whom differential expectancy stimulation is not consistently available.
