Procedural antecedents of behavioral contrast: a re-examination of errorless learning.

Kodera et al. (1976) worked with pigeons to test if errorless training kills behavioral contrast. They slowed or sped how fast the S-delta stimulus appeared. Some birds saw the new stimulus many times before the test. Others saw it only a few times.

The team tracked peck rates in both the reinforced and non-reinforced components. They wanted to know if fewer errors truly meant less contrast.

Contrast still showed up, even when birds made almost zero mistakes. The size of the contrast hinged on two things: how quickly the S-delta was introduced and how much prior exposure the bird had. Error count did not predict contrast size.

In plain words, you can run a nearly perfect errorless program and still see a later jump in responding when conditions change.

Striefel et al. (1974) already showed that contrast can appear without any drop in response rate. Kodera et al. (1976) now adds that contrast can also appear without errors. Together, the two papers break the old belief that contrast needs either suppression or mistakes.

Terrace (1974) found that 'active non-responding' early in extinction forecasts contrast size. Kodera et al. (1976) agree: procedure, not errors, drives the effect. The newer study extends the 1974 idea by showing that gradual stimulus introduction can create the same predictive non-responding pattern.

Blue et al. (1971) proved that a contingency shift, not just less reinforcement, is required for contrast. Kodera et al. (1976) build on this by demonstrating that the way you roll out the S-delta acts like a procedural shift, even when the target behavior stays flawless.

If you thin reinforcement or add new stimuli slowly, watch for behavioral contrast even when your learner makes no mistakes. A sudden jump in mand or play rates after you fade in S-delta trials may reflect procedure, not skill mastery. Track response rates across components and adjust stimulus-introduction speed if you see unwanted spikes.