Reinforcement contingencies as discriminative stimuli: II. Effects of changes in stimulus probability.

Pigeons pecked keys in a matching-to-sample task. The sample was not a light or a shape. It was the schedule that had just ended: DRL or DRO. The birds had to pick the comparison key that matched the schedule they had just worked under. The twist: the chance of seeing DRL next was changed across conditions.

When DRL became more likely, pigeons biased their picks toward the DRL side. The schedule itself acted like a color or a shape, guiding the next choice. The data showed a clean probability-based bias, even though the task stayed the same.

Lattal (1975) ran the same DRL/DRO matching task first. That paper proved the idea worked; this 1979 paper asks what happens when you mess with the odds. Reberg et al. (1979) also worked with pigeons and bias, but used signal-detection instead of matching-to-sample. Both labs saw the same pattern: more reinforcement for one choice pulls responses that way, even when the real signal does not change. Burgio et al. (1986) looks like a clash at first—they say delays hurt discriminability. No conflict: they tested memory gaps, while Crowley (1979) tested probability gaps. Different variables, different stories.

Your client may pick the red card more often not because red is easier to see, but because red has won more often lately. Before you add extra prompts or new teaching steps, check the payoff odds across choices. Balance the reinforcement rates and watch the bias fade.