Schedules of reinforcement

Why can't people stop pulling a slot machine's lever? Because the payout is on a variable-ratio schedule — the most powerful, most extinction-resistant reinforcement pattern known. Schedules of reinforcement are pure NCE gold: memorize one grid and a slot-machine image, and the questions answer themselves.

First, the big split. Continuous reinforcement rewards every correct response — great for acquiring a brand-new behavior fast, but the behavior extinguishes quickly once rewards stop. Partial (intermittent) reinforcement rewards only some responses — slower to establish, but far more resistant to extinction. This durability is the partial reinforcement effect.

Partial schedules cross two dimensions. Ratio schedules reinforce after a number of responses; interval schedules reinforce after an amount of time. And each can be fixed (predictable) or variable (unpredictable). That gives four cells. Fixed-ratio (FR): reward after a set number of responses (piecework pay every 10 units) — high rate with a brief post-reinforcement pause. Variable-ratio (VR): reward after an unpredictable number of responses (slot machines, gambling) — the highest, steadiest response rate and the most resistant to extinction. Fixed-interval (FI): reward the first response after a set time (a weekly paycheck) — produces a scalloped pattern, activity ramping up as the deadline nears. Variable-interval (VI): reward after unpredictable time intervals (checking for a text, pop quizzes) — a slow, steady rate.

Rule of thumb: ratio schedules produce higher response rates than interval schedules (rewards depend on how much you do), and variable schedules resist extinction better than fixed ones (you never know when the next reward comes).

Clinical Pearl: Variable-ratio (VR) — the slot machine — yields the highest response rate and is the most resistant to extinction; continuous reinforcement is best for acquiring a new behavior but extinguishes fastest.