Partial Reinforcement

What is partial reinforcement?

Partial reinforcement, also called intermittent reinforcement, means that some occurrences of a response are reinforced. Ferster and Skinner’s experimental account distinguishes schedules based on response counts from those based on elapsed time.

In continuous reinforcement, every occurrence of the target response is reinforced. Partial reinforcement changes that arrangement so some responses are not reinforced.

Four partial reinforcement schedules

Consider illustrative laboratory arrangements in which a pigeon’s key peck can produce food:

• Fixed-ratio: Food follows every fifth peck. The requirement is always five responses.
• Variable-ratio: Food follows changing response counts—for example, 3, 7, 4, and 6 pecks in an order the bird cannot predict. Those requirements average five responses.
• Fixed-interval: After food delivery, a 30-second interval begins. The first peck after that interval produces food; earlier pecks do not.
• Variable-interval: The interval changes—for example, 20, 40, 25, and 35 seconds. In each case, the first peck after the interval produces food. The listed intervals average 30 seconds.

The contrast is what must occur before food becomes available: a response count for ratio schedules, elapsed time followed by a response for interval schedules. Even after an interval has elapsed, no food is delivered until the pigeon pecks.

An event delivered at a fixed time regardless of responding is not, merely for that reason, a fixed-interval reinforcement schedule. Likewise, randomly timed feedback need not constitute variable-interval reinforcement unless the response contingency fits.

Partial reinforcement extinction effect

The partial reinforcement extinction effect (PREE) refers to more persistent responding after reinforcement stops following intermittent training than following continuous training. APA’s conditioning and learning glossary defines this contrast. It concerns persistence during extinction, not proof that intermittent rewards are best for every purpose.

Evidence and limits

The book documents different response patterns under controlled schedules. This supports examining the actual contingency rather than assuming all intermittent rewards work alike. It does not show that unpredictability always increases human motivation, or that partial reinforcement is universally preferable for education, work, or habit formation.