Control of responding by the elements of a compound discriminative stimulus and by the elements as individual discriminative stimuli.

Researchers looked at how fast two parts of a compound cue grab control. One cue was a light, the other was a noise. Pigeons pecked a key when the cues were on.

They trained the cues together and also alone. Then they tested which cue the birds followed.

The light won control faster than the noise in every setup. When both cues were on, the birds still tracked the light more.

Even when the noise was trained alone, it never caught up to the light.

Galbicka et al. (1981) later showed the same thing happens in concurrent schedules. Without clear cues tied to each choice, matching falls apart.

Reiss et al. (1993) added that more elements in a sample make self-monitoring worse. This lines up with the 1969 finding that parts of a compound do not share control equally.

Tanno et al. (2012) echoed the idea with rats: the slice of the stimulus closest to the reinforcer drives the choice.

When you run conditional-discrimination or matching programs, expect one part of the cue to grab control first. If that part is not the one you want, add extra trials for the weaker cue. For example, pair the weaker cue with a stronger reinforcer or present it alone first, then blend it back into the compound. This keeps the learner from only watching the "bright" part of your instruction.