On the nature of non-responding in discrimination learning with and without errors.

Pigeons pecked a key for food on a two-part schedule. In one part, every peck paid off. In the other part, no pecks were paid. The bird could still peck, sit still, or turn away.

The team filmed each second of the extinction part. They counted when the bird stayed near the key but did not peck. They called this 'active non-responding.'

Active non-responding shot up in the first extinction minutes, then fell. Birds that showed the biggest early dip later showed the biggest bounce-back in the paid part.

The pattern looked like the birds were dodging frustration. They stayed close enough to see the key but avoided the bad feeling of a peck that earns nothing.

Rodewald (1974) ran almost the same birds and boxes the same year. That paper found extra pecks aimed at the lit key during contrast. Together the two studies show contrast is two moves: more elicited pecks on rich key, plus active hold-back on lean key.

Mulvaney et al. (1974) also in 1974 saw contrast tied to 'autopecking'—tiny reflex-like pecks. The current paper says the pause side matters just as much as the peck side. The papers do not clash; they split the same wave into crest and trough.

Hursh et al. (1974) showed that light offset, not lost food, kept some birds pecking. The present study adds that staying still can also be reinforced by avoiding stimulus change. Both warn: during extinction, watch what the bird is escaping, not just what it is losing.

When you run discrimination or extinction probes, count both responses and non-responses. A sudden pause that looks like 'not attending' may be active avoidance. If it peaks early and fades, expect a later burst in the reinforced component. Use that early pause as a cheap predictor of contrast size, and consider inserting brief reinforcer primes to keep the learner engaged without building escape.