Contingency discriminability and peak shift in concurrent schedules.

Krägeloh et al. (2006) watched pigeons choose between two keys on concurrent VI VI schedules.

They lined up five components. The two end components gave very rich or very lean reinforcer rates. The middle component gave equal rates.

Each component had its own colored light. Birds could see the color before the reinforcers arrived.

The birds did not simply pick the middle, equal-pay component.

Instead, they showed peak shift. They favored the equal component that sat next to the richest extreme.

The color cue, not the pay, now controlled choice.

Jones et al. (1998) already showed that choice fades as time since the last switch grows. U et al. add a spatial twist: extreme ratios next to equal ones can pull preference sideways.

McLean et al. (2018) later flipped ratios daily and saw choice sensitivity drop across days. Together the two papers warn that both ratio extremes and fast changes can warp normal matching.

Lancioni et al. (2000) found that unequal ratios sharpen drug-appropriate responding. U et al. show the same unequal spread can also create peak shift, linking reinforcer dispersion to tighter stimulus control in two very different tasks.

When you set up concurrent teaching arrays, keep extreme reinforcer rates away from the target you want chosen. If high-rate drills sit next to the goal activity, learners may pick the wrong one by peak shift. Slide the richest rewards one slot farther over, or insert a buffer schedule, to keep stimulus control honest.