On the form of stimulus generalization curves for visual intensity.

The team tested how people spread their responses when lights got brighter or dimmer. They used many light levels, not just a few.

Each person worked alone in a quiet lab. They pressed a key when they saw the light they had learned meant "yes."

When the test covered a wide span of brightness, the response curve made a clear peak. The highest responding sat right at the trained light level.

If the range was too narrow, the peak was hidden. You need enough steps to see the true shape.

Derenne (2010) used faces instead of lights and still saw a peak shift. This shows the same rule works with real-world pictures.

Flapper et al. (2013) swapped people for rats and flash rate for brightness. The peaked curve showed up again, proving the effect crosses species and senses.

Fields et al. (2002) looked at sorting new pictures into groups. They also found that giving many examples during training helps later spread of responses, lining up with the "test wide" message.

Next time you probe stimulus control, slide the stimulus along a long ruler. Whether you work on matching coins, telling voices apart, or any discrimination task, sample at least five steps above and below the trained value. You will catch true peaks, avoid flat curves, and write better treatment goals.