Contrast effects in maintained generalization gradients.

Researchers worked with pigeons in a small lab chamber. The birds pecked a lit key for food. First the key was one color, then it switched to a new color. The team tracked how fast the birds pecked after each switch.

They wanted to see if the birds' response rate changed when the color changed. They also checked if the change matched how much food the birds got.

The birds pecked faster or slower right after the color changed. This jump had nothing to do with how much food they earned. Instead, the jump matched how fast they had pecked in that color before.

A color that once saw fast pecking stayed 'hot.' A color that once saw slow pecking stayed 'cold.' The stimulus carried a memory of past responding, not just a memory of food.

LeFrancois et al. (1993) later showed pigeons need both odor cues and the same chamber to keep a new rule. Both studies say the same thing: stimulus control is historical, not fixed.

Cook (2002) found pigeons only learned a same-different rule when the mapping stayed consistent. That fits today's result: consistent past response rates, not reward size, set the stimulus value.

Cohen (1969) showed pigeons could match new colors without extra training. That might seem to clash with the 'hot/cold' memory idea, but the tasks differ. Matching tests transfer; contrast tests test response rate history. Both can be true.

When you see a client respond faster or slower after a stimulus change, do not blame the new reward rate first. Ask what response rate that stimulus has seen before. If a flashcard once dragged out slow answers, it may keep doing so even after you raise reinforcement. Swap the card or pair it with a 'hot' partner to rewrite its history.