Stimulus control of delayed matching in pigeons: Directed forgetting.

Researchers worked with pigeons on a delayed matching task. Birds saw a color sample, then a delay, then two choice keys.

A cue appeared during the delay. One cue meant "remember." Another cue meant "forget." The team tested if these cues changed later accuracy.

Accuracy stayed high when the forget cue and the test shared the same response context. Accuracy fell when the contexts differed.

In plain words, the birds did well only if the forget cue and the remember cue controlled the same pecking pattern.

Rasing et al. (1992) extends this idea. They showed that a stimulus can carry a "history tag." The bird's past rate of reinforcement under that stimulus changes future responding, even when the current schedule stays the same.

Cullinan et al. (2001) gives a conceptual replication. Pigeons learned say-do correspondence when a correction procedure created strong conditional stimulus control. Like Kendrick et al. (1981), performance dropped when the conditional cue weakened.

Gentry et al. (1980) and Rapport et al. (1982) look at first like contradictions. They report that changing reinforcement rate shifts response bias but leaves discrimination accuracy untouched. The key difference is task: F et al. tested context cues inside delayed matching, while the D papers used signal-detection tasks with no context switch. Context, not payoff rate, drove the 1981 effect.

When you change rooms, staff, or materials, you may accidentally swap the learner's context. If the new context no longer signals the same response chain, performance can crash even though the learner "knows" the skill. Check that your cues, materials, and response requirements stay constant across training and testing. If you must change context, re-teach a few trials in the new setup so the old control transfers over.