cumulative

G.16 Maintaining behavior change after intervention

What the task asks: Plan how a behavior will keep occurring once intervention ends, by thinning schedules, shifting control to naturally occurring reinforcers and putting long-term supports in place.

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

The big idea

Response maintenance is the extent to which a learner keeps performing a behavior after the intervention that produced it has been partly or fully withdrawn (B.15). A skill that disappears a month after discharge, or the week a teacher changes, was taught but not maintained.

Maintenance does not happen on its own any more than generalization does (G.15). Dense, contrived reinforcement is excellent for building behavior, but it is also easy to discriminate: when it stops, the behavior can stop too. Programming maintenance means deciding from the start how control will move from the teaching contingencies to ones that will still be there later.

Breaking it down

Pick behaviors the environment will keep

The relevance-of-behavior rule says to teach behaviors that will go on being reinforced in the learner’s daily life. Asking for help, greeting coworkers and paying for items all meet naturally existing contingencies, which operate whether or not a behavior analyst is involved. A behavior that only the therapist ever reinforces is hard to maintain.

Thin the schedule

Schedule thinning gradually reduces how often reinforcement is delivered, for example from every response (continuous reinforcement) to every second response, then every fourth, then a variable ratio averaging 8. Behavior reinforced on intermittent schedules, especially variable ones, persists longer when reinforcement is missing (B.9).

CRF every response FR 2 every 2nd FR 4 every 4th VR 8 8th on average
The same 16 responses under each step of the thinning sequence above. Reinforcers drop from 16 to 8, 4 and then 2, and on the variable ratio the learner cannot tell which response will pay off.

Thin in steps and let the data decide each move. Thinning too fast produces ratio strain: long pauses, erratic responding or stopping, sometimes with problem behavior. When that happens, return to the last schedule that worked and use smaller steps.

Delay the reinforcer

Natural reinforcers often come late: paychecks, grades, a friend’s approval next week. Gradually lengthening the delay between behavior and reinforcer prepares the learner for that. Tokens, a written note or a verbal description (“You picked the game after lunch because you finished this morning’s work”) help bridge the gap. Delayed and intermittent reinforcement also make the contingency hard to discriminate, so the learner keeps responding when reinforcement is not obviously on offer (indiscriminable contingencies, G.15).

Transfer to naturally occurring reinforcers

Pair the contrived reinforcer with the natural one, for example by delivering praise and a smile with each token, then fade the contrived one. Teach the behavior to the speed and accuracy the natural audience will notice. Teach the learner to recruit reinforcement, and look for behavior traps where natural contingencies can take over.

Fade the other supports

Prompts, special materials, the therapist’s presence and the structure of the session are all part of the teaching contingency. Fade them on purpose, and involve the caregivers, teachers or supervisors who will be there after you leave.

Booster sessions and follow-up

Booster sessions are brief periods of retraining after the main intervention ends, scheduled in advance or triggered when follow-up data show a decline. Plan maintenance probes, for example at 2 weeks, 1 month and 3 months, so a decline is caught early.

05101520BaselineTrainingFollow-upBoosterFollow-up24681012141618Sessions and follow-up probesLabeled praise per 10 min
Ms. Kowalski's labeled praise, from the home example below. Follow-up probes at 2 weeks, 1 month and 3 months catch a drop at 3 months; two booster sessions bring it back, and later probes show it holding.

Self-management and fluency

Self-management, in which the learner sets goals, monitors and records their own behavior, and arranges their own consequences, lets the contingency travel with the person after staff leave. Self-monitoring with an occasional accuracy check is a practical starting point. Practice beyond the mastery criterion, to fluent levels, also tends to improve retention.

Seeing it in practice

  • School. Tanvir, 11, earns a token after each homework assignment. His BCBA moves to tokens for every second assignment, then a variable schedule, while his teacher’s written comments are paired with each token. By spring, grades and comments are the only consequences, and homework continues.
  • Home. After parent training, Ms. Kowalski uses labeled praise often. A three-month follow-up shows a drop, so the BCBA schedules two booster sessions with feedback and sets up a weekly self-recording sheet.
  • Workplace. Daniel, 26, learned with tokens to say “excuse me” when he needs to pass coworkers. Coworkers smile and step aside when he says it, so the tokens are faded and the natural consequence takes over.
  • Clinic. A teen earns points every 5 minutes for staying on task. The analyst shifts to checks at unpredictable times and delivers points at the end of the session, then the next morning.

Common mix-ups

  • Maintenance vs. generalization. Maintenance is about time: does the behavior last after intervention ends? Generalization is about new settings, people and forms.
  • Thinning is not stopping. Moving straight from continuous reinforcement to none is extinction, not thinning.
  • Ratio strain vs. extinction burst. Ratio strain comes from a schedule that got too lean too fast, and the fix is to back up. A burst follows the removal of reinforcement altogether.
  • Booster sessions are not a sign of failure. They are a planned part of many maintenance programs.
  • Maintenance vs. relapse planning. Maintenance keeps desired behavior going. Planning for the return of problem behavior is covered in H.5.

Exam tip: When a scenario says a behavior stopped after the program ended, look for the answer that shifts control to natural, intermittent or delayed consequences, not the one that restarts the original dense schedule or removes everything at once.

Quick recap

  • Response maintenance is behavior that lasts after intervention is withdrawn.
  • Teach behaviors natural contingencies will reinforce, and pair contrived reinforcers with natural ones before fading them.
  • Thin schedules gradually, and back up if ratio strain appears.
  • Lengthen delays to reinforcement so the contingency is hard to discriminate.
  • Use booster sessions, follow-up probes and self-management to keep gains going.

Further reading

  • Cooper, J. O., Heron, T. E., & Heward, W. L. (2020). Applied behavior analysis (3rd ed.), chapters 29 and 30.
  • Stokes, T. F., & Baer, D. M. (1977). An implicit technology of generalization. Journal of Applied Behavior Analysis, 10(2), 349-367.

Key terms

Response maintenance (Maintenance)
The extent to which a learner continues to perform a target behavior after some or all of the intervention that produced it has been withdrawn.
Schedule thinning
Gradually reducing how often reinforcement is delivered, by raising the response requirement, lengthening the interval or adding delays, while keeping behavior at the desired level. It makes a procedure practical to sustain and the behavior more durable.
Intermittent reinforcement (Partial reinforcement)
Any schedule in which some, but not all, occurrences of a response are reinforced. The basic intermittent schedules are fixed ratio, variable ratio, fixed interval and variable interval. Behavior maintained on intermittent schedules is more resistant to extinction than behavior maintained on continuous reinforcement.
Ratio strain
Disrupted responding that occurs when a ratio requirement is raised too quickly or becomes too large: long pauses, erratic responding and sometimes a near stop of the behavior. It is prevented by thinning ratios gradually.
Naturally existing contingency (Natural contingency)
A contingency of reinforcement or punishment that operates in a person's environment without the behavior analyst's involvement, including consequences other people already provide. Contrived contingencies, by contrast, are designed and run by the practitioner.
Booster session
A brief period of retraining delivered after the main intervention has ended, either on a planned schedule or when follow-up data show a decline, to restore and maintain the behavior change.
Self-management
A person's use of behavior-change tactics to change their own behavior, such as setting goals, self-monitoring, self-instruction and arranging their own consequences. Because the person carries the contingency with them, it supports maintenance and generalization.

Check yourself

4 questions on G.16. Choose an answer to see whether you're right and why.

  1. Question 1Asha earns a token for every homework assignment she completes. The team wants her homework completion to continue after the token program ends at the close of the school year. What is the best next step?
    Show the answer

    Answer: A. Thinning gradually to an intermittent, variable schedule while pairing tokens with natural consequences makes homework more persistent and shifts control to reinforcers that will remain. Stopping abruptly is extinction. Fixed-time delivery breaks the link between homework and reinforcement altogether.

  2. Question 2A BCBA moves a learner's sorting task from FR 2 straight to FR 12 in one day. Over the next few sessions the learner pauses for long periods, sorts erratically and starts pushing materials away. What is the best interpretation and response?
    Show the answer

    Answer: D. Long pauses, erratic responding and protest after a large jump in the ratio are signs of ratio strain. Backing up to a schedule that maintained responding and thinning more gradually solves it. An extinction burst follows the removal of reinforcement altogether; here reinforcement is still available, just too lean too quickly.

  3. Question 3Six months after a successful parent-training program, follow-up probes show that a parent's use of labeled praise has dropped by about half, though they still use it at times. Which response fits best?
    Show the answer

    Answer: D. A partial decline after treatment calls for brief, targeted retraining (booster sessions) plus continued monitoring to catch later drops. Restarting the whole program spends far more time than the data justify, since the skill is still partly there. Stopping data collection would hide any further decline.

  4. Question 4Geneviève, 15, stays on task in class with an aide who gives her points every 10 minutes. The aide will be reassigned in two months. Which plan gives Geneviève's on-task behavior the best chance of lasting?
    Show the answer

    Answer: D. Self-monitoring moves the contingency to something Geneviève carries with her, and fading the points gradually, with occasional teacher accuracy checks, keeps the change from being abrupt. A new aide keeps the contrived contingency in place instead of transferring control. Removing points all at once is extinction, and a rule alone rarely sustains behavior.

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

Type to search all 104 guides and the glossary.