Use variable-ratio reinforcement to make habits persistent

Once a behavior is established, shift to an unpredictable reward schedule to make it resistant to extinction.

Why it works

Variable-ratio (VR) schedules reward behavior after an unpredictable number of responses — the defining feature of gambling. VR produces the highest response rate and the slowest extinction of any reinforcement schedule because each non-rewarded response could be the last one before the reward. The animal/human does not know when to stop because the next response might be the winning one. This makes VR habits very durable — but also means they can be exploited by app designers and casinos.

How to do it

1. Establish the behavior on a continuous or fixed schedule first — reward every occurrence until it is reliable.
2. Once the behavior is consistent, introduce occasional surprise rewards: a spontaneous treat, an unexpected check-in, a random celebration.
3. Keep the rewards genuinely unpredictable, not just infrequent — a weekly scheduled reward is a fixed interval, not variable.
4. Use variable reinforcement only for behaviors you want to maintain, not for those you want to extinguish.

Evidence

Variable-ratio schedule effects on response rate and extinction resistance are among Skinner’s most replicated findings, demonstrated across species and behavior types. (rct)

Most of the foundational data comes from animal research (pigeons, rats). Translation to complex human behavior in naturalistic settings requires care; the social and cognitive context modulates the schedule effect.

Sources

• Ferster & Skinner (1957), "Schedules of Reinforcement" — foundational experimental data

Common mistake

Applying variable reinforcement before the behavior is established — variability during acquisition slows learning; it is a maintenance, not an acquisition, tool.