Date of this Version
Recent experiments have shown that bluejays and rhesus monkeys experienced in object-discrimination learning set (ODLS) exhibit a rapid decline in performance when a retention interval is inserted between successive trials of individual ODLS problems (Bessemer & Stollnitz, 1971; Kamil, Lougee, & Shulman, 1973). This intraproblem retention loss (IRL), or forgetting, has been interpreted as reflecting the importance of relatively transient memory traces for events of previous trials of the ODLS problem as determinants of choice behavior on the current trial of the same problem. According to this model, these memory traces function as discriminative stimuli in a conditional discrimination which controls choice behavior in the sophisticated subject. For example, if the subject remembers having responded to Object x, and having been rewarded on the previous trial(s), then he approaches Object x on the next trial; however, if he remembers nonreward, then he avoids Object x. This model is obviously similar to the "win-stay, lose-shift" hypothesis proposed by Levine (1959), but would seem to have two advantages. First, it specifies the stimuli, specific memory traces of previous trial events, necessary for hypothesis behavior to occur. Second, it explains the finding of rapid IRL if we assume that these memory traces are transient, losing strength during retention intervals.