Response patterning during stimulus generalization in the rat.

The team trained rats to press a lever for food on a variable-interval schedule. A tone acted as the cue for reinforcement.

Later they tested the rats with new tones that were higher or lower than the training tone. During these tests food was removed so the rats could not earn anything.

As the test tone moved farther from the training tone the rats paused longer between presses. The drop in total responses came from these longer pauses, not from faster pressing near the cue.

In short, the generalization gradient was built from longer quiet times, not from speed once the rat started pressing.

Stretch et al. (1966) looked at the same rat set-up one year earlier. They also found that the shape of the gradient depends on what happens after the first press, matching the pause idea.

Crowley (1979) later showed that these pauses are really small, already-learned response chunks being mixed together. The 1967 data fit that view: longer pauses are simply the rat choosing its "slow" chunk more often.

Kuch et al. (1976) proved you can train a rat to use a specific pause length by paying only for that pause. This supports the 1967 claim that pause length, not overall rate, drives the gradient.

When you see a client respond less to a new stimulus, ask if they are pausing longer instead of just "not doing the skill." Build fluency by reinforcing quick starts after the cue. You can also probe generalization by timing the first response instead of counting total responses. Shaping the pause may give you a cleaner gradient and faster discrimination.