Substitutability between conditioned and primary reinforcers in discrimination acquisition.

The team worked with pigeons in a lab. They wanted to know if a light or sound could work like food.

First they paired one stimulus with grain. Another stimulus was shown alone, never with food.

Then both stimuli were used as "good job" signals while the birds learned a color discrimination.

Only the stimulus that had been paired with food sped up learning. The un-paired cue did nothing.

In plain words, a conditioned reinforcer acts like food only if it has a food history.

Davis et al. (1994) ran the same birds again and showed timing matters too: the cue must come right after the correct peck. Together the two papers say pairing creates the power, and immediacy lets it work.

Bonfonte et al. (2020) moved the idea to children and tokens. New tokens lost to high-preference snacks, backing the rule that fresh conditioned reinforcers start off weaker.

Clark et al. (1970) looks like a clash at first: their food-paired cue hurt matching accuracy. The difference is where the cue was placed. In 1970 it followed wrong responses, so it acted like a punisher, not a reinforcer. Same mechanism, opposite placement.

Before you swap edibles for praise, stickers, or points, check their history. If the new reinforcer has never been paired with strong backup, it will probably fail. Spend time delivering the item with a known liked reinforcer first, then fade the primary item out. This simple pairing phase can save weeks of slow progress.