Auditory frequency generalization in the goldfish (Carassius auratus).

Fay (1970) trained goldfish to press a bar when they heard one tone.

Then the fish heard many tones, higher and lower. The team counted bar presses to see how close a new tone had to be before the fish treated it like the training tone.

They tested tones from 100 Hz to 1600 Hz.

For tones between 100 Hz and 800 Hz, the fish mostly pressed only near the training frequency. The generalization gradient looked like a steep hill.

At 1600 Hz the hill flattened; the fish pressed almost as much to distant tones as to the trained one.

Schwartz et al. (1971) got the same steep auditory gradient in monkeys, but they measured how fast the animals moved instead of bar presses. The shape matches even though the yardstick changed.

AZRIN et al. (1963) built similar curves in blind, severely intellectually disabled children using loudness instead of pitch. Together the three papers show the gradient phenomenon holds across species and stimulus dimensions.

RISLEY (1964) used a two-response method with humans and still saw clean gradients. That design similarity lets you trust the goldfish data even though the subjects swam instead of talked.

You now know that sharp stimulus control can be demonstrated with almost any organism and any auditory cue. When you shape listening skills with clients, start with low-frequency targets if you want tight discrimination. Expect flatter control if you later move to higher pitches.