Knowing that conditioned reinforcers are learned is the concept. Building one on purpose, checking that it works, and keeping it working for months is the procedure. This lesson is about the procedure: pairing, testing, and running a token economy you can eventually fade.
Why it matters
Many learners start services with only a few reinforcers, often food or one favorite toy. Those are hard to deliver quickly, they satiate, and they rarely exist in classrooms or workplaces. A well-built conditioned reinforcer, such as praise, a click, or a token, can be delivered instantly, in any setting, without interrupting the task.
Token economies extend that logic. They bridge the delay between a response and a backup reinforcer, let you reinforce many behaviors with one currency, and give you a structured path toward natural contingencies. When a token system "stops working," the cause is usually a design flaw you can find and fix.
Key ideas
- Pair in the right order. Present the neutral stimulus immediately before, or overlapping with, an established reinforcer. The new stimulus should reliably predict the reinforcer. Delivering the reinforcer first and the stimulus afterward (backward pairing) builds little or no value.
- Make the stimulus informative. If the established reinforcer shows up just as often without the stimulus, the stimulus predicts nothing and gains little value. Keep the backup tied to the stimulus during pairing.
- Test; don't assume. After pairing, deliver the stimulus alone, contingent on a simple response, and compare responding with a baseline or control condition. If responding rises above that comparison, the stimulus is functioning as a conditioned reinforcer.
- Keep pairing. A conditioned reinforcer delivered again and again without its backup gradually loses value. Planned, regular exchanges or occasional pairings keep it effective.
- A token economy has core parts. You need defined target behaviors, the tokens themselves, a menu of backup reinforcers, and clear rules for how tokens are earned and when, where, and at what price they are exchanged.
- Start rich, then thin. Early on, let the learner exchange after one or a few tokens so each token quickly contacts the backup. Then gradually raise the number of tokens required and the time between exchanges.
- Protect the backups. Choose backups from preference assessments, rotate the menu, and keep them unavailable outside the system. Free access to a backup weakens the motivation to earn it.
- Plan the exit from day one. Pair every token with specific praise, then gradually thin token delivery, lengthen the exchange delay, shift from physical tokens to checkmarks, and move toward reinforcers that occur naturally in the setting.
Definitions
- Conditioned reinforcer: a once-neutral stimulus that functions as a reinforcer because of its history of pairing with other reinforcers.
- Stimulus-stimulus pairing: presenting a neutral stimulus just before or with an established reinforcer so the neutral stimulus acquires reinforcing value.
- Token economy: a system in which learners earn tokens for specified behaviors and later exchange them for backup reinforcers under set rules.
- Backup reinforcer: an item, activity, or privilege that tokens can be exchanged for.
- Exchange ratio: how many tokens a given backup costs (the token-exchange schedule). Together with the token-production schedule, or how much behavior earns one token, it sets how rich or lean the system is.
- Level system: a type of token economy in which learners move up (or down) through levels that carry more privileges and higher expectations as performance improves.
Examples from a session
Pairing, a token board, and a workplace system
- Pairing in a clinic: Theo, age 4, responds to very few social consequences. His BCBA clicks a clicker, then immediately blows bubbles. Bubbles appear only after the click. Two weeks later she tests: during a sorting task, the click alone follows each block placed in the bin, and his placements rise above the rate seen in a no-click baseline.
- A token board at home: Jasmine, age 7, earns stars on a five-space board for finishing homework steps. At first she traded each single star for a minute of drawing time. Over a month, her family raised the price to a full board and added backups she picked during a preference assessment, such as baking with her dad and picking the bedtime story.
- Points at work: Marcus, age 34, in a supported employment program, earns points for arriving on time and completing his task checklist. He exchanges points on Fridays for items on a menu he helped write. His job coach pairs each point with specific feedback so that the feedback can carry more of the load as points are faded.
Common mistakes / exam traps
- Long delays at the start. If a learner must earn ten tokens before the first exchange, the tokens may never acquire value. Begin with immediate or near-immediate exchanges.
- Skipped exchange periods. Tokens delivered week after week without trading are unpaired. Expect their effect to fade.
- One backup only. A single backup invites satiation. A varied menu makes tokens generalized and more durable.
- Calling a stimulus a reinforcer without a test. Smiling or orienting toward a sound is not evidence that it strengthens behavior.
- Abrupt withdrawal. Removing a token system all at once often leads to a drop in behavior. Thin it gradually.
Exam trap
Forward pairing vs. backward pairing (or mere co-occurrence). The discriminating feature is prediction: the new stimulus gains value only when it comes first and the established reinforcer reliably follows it. Reinforcer then stimulus, or a reinforcer that also shows up freely without the stimulus, builds little value, however many trials you run.
Exam trap
Weak tokens vs. devalued backups. When tokens "stop working," ask where the value broke. If tokens rarely contact backups (skipped exchanges, long early delays, thinning too fast), restore exchanges. If exchanges happen but the backup is freely available or satiated, the backup lost value: protect and vary the menu.
Quick check
For a month, sessions often run late, so the RBT skips the token exchange period while still delivering tokens. What is the most likely result?
Related topics
- Kinds of reinforcers: unconditioned, conditioned, and generalized (B.7)
- Motivating operations in action (B.16)
- Designing and evaluating preference assessments (F.4)
- Positive and negative reinforcement procedures (G.1)
- Group contingencies (G.14)
- Maintaining change after intervention: schedule thinning and natural reinforcers (G.16)