Response rate as a function of amount of reinforcement for a signalled concurrent response.

The team worked with pigeons in a two-key chamber. One key gave food only when a tone sounded first. The other key gave food with no signal.

They made the signaled food longer and longer. They watched how fast the birds pecked the unsignaled key.

As the signaled reward grew, pecks on the quiet key dropped. The birds acted as if the unsignaled side lost value.

The result shows that rich, signaled reinforcement can suppress behavior on a plain concurrent option.

Harrison et al. (1975) saw the opposite: signaling rewards raised rates on the other key. The gap comes from timing. In 1969 the signal came before every reward; in 1975 it only marked when rewards were possible.

Wetherington (1979) repeated the drop with people. More signaled rewards again cut responding on the plain side. The pattern holds across species.

Krägeloh et al. (2003) added a change-over delay. The delay softened the drop and sharpened choice. The old finding is still true, but you can tune it.

When you add signals or extra reinforcement to one task, watch the other tasks. The plain option may lose steam even though nothing there changed. If you see a slide in unprompted work, check whether another part of the session just got richer or flashier. Balance signaled and unsignaled reinforcement to keep all responses strong.