The effect of discrimination training on responses to a new stimulus.

Anger et al. (1972) asked how long discrimination training should last. They used pigeons that pecked a yellow key for food. Green and red keys never gave food. After either three or nine days, they added a new blue key that also never gave food.

The team counted how many times birds pecked the new blue key. They wanted to know if longer training would cut these errors.

Birds with nine days of training made far fewer pecks to the new blue key. Birds with only three days kept pecking the blue key more often.

Nine sessions were enough to build strong stimulus control. Short training left room for mistakes when a new negative stimulus appeared.

Reynolds (1966) showed that long training wipes out behavioral contrast and peak shift. Anger et al. (1972) now show it also blocks responding to a brand-new S-. Together, the two papers say the same rule: keep training past the first few days to lock in clean stimulus control.

Touchette (1971) ran 64 sessions and still saw steep generalization gradients. That seems to clash with K's claim that nine days are enough. The difference is the test: E kept testing the same old stimuli, while K added a fresh blue key. Long training guards against new errors, not against old ones.

Glynn (1970) found that more training per reversal sped up later reversals. K et al. add another payoff: extra days also protect against mistakes when new stimuli pop up.

When you teach a child to sort red blocks as S+ and green blocks as S-, plan for at least nine teaching sessions before you probe with a new color. Short training lets the child try to play with the new item; longer training teaches them to leave it alone. This matters in classrooms where new materials appear without warning.