cumulative

G.4 Conditioned reinforcers and token economies

What the task asks: Describe how a neutral stimulus is turned into a reinforcer, and plan, run and fade a token system that uses conditioned reinforcers to strengthen behavior.

Behavior-Change Procedures5 min read4 practice questionsOfficial outline (opens in a new tab)

The big idea

Praise, points, money and a check mark on a chart only work if they have become reinforcers through experience. A conditioned reinforcer is a stimulus that was once neutral and now strengthens behavior because of its history with other reinforcers (see B.7). Many learners arrive without praise or tokens already working, so practitioners often need to build these reinforcers on purpose.

Conditioned reinforcers solve practical problems. They can be delivered instantly without stopping the lesson, they bridge the delay to a larger reinforcer, and when they can be exchanged for many things they resist satiation. The token economy is the best-known system built on them.

Breaking it down

Establishing a conditioned reinforcer by pairing

The basic method is pairing: present the neutral stimulus just before, or overlapping with, a stimulus that already works as a reinforcer.

  • Stimulus-stimulus pairing presents the two together without requiring any response. The therapist says “Nice!” and then hands over a chip, again and again.
  • Response-stimulus pairing delivers the pair contingent on a response. The learner places a puzzle piece, hears “Nice!” and then gets a chip.

Both are used. In one well-known comparison with praise (Dozier and colleagues, 2012), response-stimulus pairing produced conditioned reinforcement for more participants than stimulus-stimulus pairing, but not for everyone. Check the effect for each learner.

Features that make pairing work:

  • Order and timing. The neutral stimulus comes first and the established reinforcer follows within a second or so. Backward pairing (reinforcer first) is weak.
  • Contingency. The established reinforcer should rarely appear without the new stimulus, so the new stimulus predicts it.
  • Motivation. Pair with something the learner wants right now (see G.5).
  • A test. After pairing, deliver the stimulus alone after a new response and see whether that response increases. Pairing is not finished until the stimulus passes this test.
Forward "Nice!" comes first "Nice!" within 1 s Chip Strong Backward chip comes first Chip "Nice!" Weak
Order decides whether praise gains value. In forward pairing, "Nice!" comes first and predicts the chip a second later. In backward pairing, the chip has already arrived, so "Nice!" predicts nothing.

A conditioned reinforcer that is never backed up again will lose its value, so keep pairing it now and then.

Generalized conditioned reinforcers

A generalized conditioned reinforcer has been paired with many different reinforcers, so it does not depend on any single MO. Money, praise and tokens that buy many things are examples. This is why a token economy usually offers a menu rather than one prize.

Designing a token economy

A token economy delivers tokens contingent on target behaviors and lets the learner exchange them later for backup reinforcers. Plan each part:

  1. Target behaviors: define them and decide how many tokens each earns.
  2. Tokens: easy to deliver, durable, safe, hard to counterfeit, and visible enough to show progress (a five-space board, points on a card).
  3. Backup reinforcers: varied, chosen with the learner, and priced by value. Include cheap items so early exchanges come quickly.
  4. Exchange schedule: when and how often tokens are traded. Early on, exchange after a few tokens and often. Later, require more tokens per exchange and longer waits.
  5. Initial training: if tokens are new, give a few and let the learner exchange them right away, repeating until the learner works for them.
  6. Rules and records: who delivers tokens, how they are tracked and what happens to unspent tokens.
Token board one star per task: 3 of 5 earned board full exchange Backup menu: 1 minute of tablet bubbles music
A five-space token board like the one in the clinic example below. Each finished task earns a star, and a full board is exchanged for a minute of the backup the learner picks from a small menu.

Response cost inside a token system

Some systems remove tokens after problem behavior. That is response cost, a negative punishment procedure (see G.17). If you include it, keep fines small relative to earnings, never let a balance go below zero, and keep the earning side much richer than the losing side. A learner who loses more than they earn stops caring about tokens.

Fading the system

A token economy is a bridge to natural contingencies (see G.16):

  • Pair every token with specific praise so that praise gains strength.
  • Thin the token schedule and lengthen the time to exchange.
  • Shift backup reinforcers toward natural ones, such as class jobs or free time.
  • Move to self-monitoring, or to a level system in which privileges are earned for longer stretches.

Seeing it in practice

SettingTokenBackup reinforcersExchange
Early learner clinicStars on a 5-space boardChoice of tablet, bubbles or music for 1 minuteWhen the board is full
4th-grade classClass pointsLunch with the teacher, homework pass, extra recessFriday class store
Adult vocational programPaper “pay” slipsCafé vouchers, choice of next taskEnd of each shift

Sofia, age 4, has never worked for stars or tokens. Her therapist starts by handing her a star and trading it straight away for 20 seconds of her favorite song, over and over. Next, Sofia earns one star per task and trades after two, then after five. Each step waits until her responding holds steady.

Common mix-ups

  • Tokens are not automatically reinforcers. They must be established and kept backed up.
  • A varied backup menu is not just a nice extra. Variety is what makes tokens generalized and resistant to satiation.
  • Response cost is punishment. It belongs to G.17, and its side effects need monitoring (see G.18).
  • Pairing is not just being friendly. “Pairing yourself with reinforcement” works because you come to predict good things.
  • Backward order fails. Handing over the snack and then saying “good job” is a weak pairing.

Exam tip: If a token system “stops working,” check the backup side first. Are the backups still valued, too expensive, or freely available elsewhere? Tokens are only as strong as what they buy.

Quick recap

  • Conditioned reinforcers are built by pairing a neutral stimulus with an established reinforcer, neutral stimulus first.
  • Confirm the result by making the new stimulus contingent on a response.
  • A token economy needs defined targets, tokens, a varied backup menu and an exchange schedule.
  • Keep response cost small and never let balances go negative.
  • Fade tokens toward praise and naturally occurring reinforcers.

Further reading

  • Cooper, J. O., Heron, T. E., & Heward, W. L. (2020). Applied behavior analysis (3rd ed.), chapter 28.
  • Hackenberg, T. D. (2009). Token reinforcement: A review and analysis. Journal of the Experimental Analysis of Behavior, 91(2), 257-286.
  • Dozier, C. L., Iwata, B. A., Thomason-Sassi, J., Worsdell, A. S., & Wilson, D. M. (2012). A comparison of two pairing procedures to establish praise as a reinforcer. Journal of Applied Behavior Analysis, 45(4), 721-735.

Key terms

Conditioned reinforcer (Secondary reinforcer, Learned reinforcer)
A formerly neutral stimulus that functions as a reinforcer because it has been paired with unconditioned reinforcers or with other conditioned reinforcers. Its effect comes from the person's learning history, and it weakens if the pairing is not maintained.
Stimulus-stimulus pairing (Pairing)
Presenting a neutral stimulus together with, or just before, a stimulus that already has a function, so that the neutral stimulus acquires a similar function. Pairing establishes conditioned stimuli, conditioned reinforcers and conditioned punishers. The same name is also used for a specific procedure that pairs an adult's speech sounds with preferred items to increase a child's vocalizations.
Generalized conditioned reinforcer (Generalized reinforcer)
A conditioned reinforcer that has been paired with many other reinforcers, so it works across a wide range of motivating operations rather than depending on deprivation of any one thing. It resists satiation better than a reinforcer backed by a single item.
Token economy (Token system)
A behavior-change system in which tokens, which are generalized conditioned reinforcers, are delivered contingent on target behaviors and later exchanged for backup reinforcers. Its parts are the target behaviors, the tokens, the backup reinforcers and the exchange schedule; some systems also remove tokens as response cost.
Backup reinforcer
An item or activity that a token or other conditioned reinforcer can be exchanged for. Tokens get their value from the backups they buy, so the backups must stay valuable and should not be freely available elsewhere.
Response cost
A negative punishment procedure in which a specific amount of a reinforcer (tokens, points, money or minutes of an activity) is removed contingent on a response, which decreases the future frequency of that response.

Check yourself

4 questions on G.4. Choose an answer to see whether you're right and why.

  1. Question 1A therapist wants praise to work as a reinforcer for Arjun, who currently shows no interest in it. Which pairing arrangement is most likely to work?
    Show the answer

    Answer: A. Pairing works best when the neutral stimulus comes first and the established reinforcer follows immediately, so praise predicts the snack. Backward order (snack, then praise) is a weak arrangement. Praise and snacks at unrelated times, or snacks that are freely available, give praise no predictive relation to anything.

  2. Question 2A token board worked well for Kai for three months. Now he earns tokens slowly and often refuses to exchange them. His family recently began giving him unlimited tablet time after school, and tablet time is his main backup reinforcer. What should the analyst check first?
    Show the answer

    Answer: C. Tokens get their value from what they buy. Unlimited tablet time at home weakens the MO for the main backup, so the tokens lose value. Adding response cost or thinning the schedule would make a weak system weaker, and durability is unrelated to the change in behavior.

  3. Question 3In a classroom token system, students lose two points for each rule violation. By Tuesday one student, Zoe, has a negative balance and has stopped working for points. What design change is most appropriate?
    Show the answer

    Answer: C. When fines outrun earnings, tokens stop being worth working for and the system collapses. Response cost within a token economy should be small relative to what the learner earns, and balances should not go negative. Bigger fines or removing backups deepen the problem, and points on a fixed-time schedule would no longer reinforce any target behavior.

  4. Question 4After pairing a clicker sound with edibles over many trials, how can the team best confirm that the clicker is now a conditioned reinforcer?
    Show the answer

    Answer: B. A reinforcer is defined by its effect on behavior, so the test is to make the click contingent on a response and look for an increase. Orienting shows only that the learner notices the sound. Caregiver impressions are indirect, and more pairing without a test never tells you whether pairing has worked.

Adds a step to your study record. Saved in this browser only.

Type to search all 104 guides and the glossary.