Generalization of auditory intensity following discrimination training.

Thomas et al. (1960) trained pigeons to peck when they heard a tone at one loudness. They also taught the birds not to peck at a much softer or louder tone. After this two-stone drill they tested tones in between.

The test tones covered a 40-decibel range. Birds had never heard these middle volumes before. The team recorded how fast each bird pecked at every new sound.

Pecking rate made a smooth, bow-shaped curve. The highest point sat right on the trained loudness. Response dropped on both sides, making a clear generalization gradient.

The curve showed that discrimination training sharpened control by the exact trained sound. Nearby volumes still got some pecks, but fewer as they moved away.

THOMAS et al. (1963) added a third tone and saw the peak shift away from the unreinforced sound. This builds on the 1960 curve by showing how extra training moves the peak.

Reynolds (1966) later showed that longer training wipes out the peak shift. Early curves look like the 1960 shape, but keep drilling and the shift fades.

Harrison et al. (1975) swapped food for shock-avoidance and got the same steep curves. The pattern holds even when the reinforcer is escaping shock rather than earning grain.

Touchette (1971) ran 64 sessions and the steep gradient stayed put. More trials do not always flatten control; sometimes they lock it in.

The curve tells you that learners will respond most at the exact cue you reinforce. If you want narrow control, train clear extremes first. Watch for early peak-shift errors near the boundary; they fade with practice. Use extra sessions when you need lasting precision, not when you want flexible transfer.