Maintaining change after intervention: schedule thinning and natural reinforcers

A skill that disappears the month after services end was never really taught for the learner's life. Maintenance is what turns a good graph into a lasting change.

Why it matters

Dense, contrived reinforcement is great for teaching, but nobody hands out a token for every greeting at work or every finished homework page. If the program ends while the behavior still depends on those arrangements, the behavior is likely to fade. As a BCBA, you plan for maintenance from the first day of a program, not as an afterthought once the goal is met.

Practice vignettes for this task describe a skill that has been acquired and ask you to choose the next step, spot why responding fell apart during thinning, or pick the plan most likely to keep the behavior going after the intervention ends.

Key ideas

• Plan the exit when you plan the program. Choose target behaviors that will actually pay off in the learner's everyday environment. A skill that peers, family, or coworkers will reinforce is far easier to maintain than one they will ignore (the relevance-of-behavior rule).
• Thin the schedule gradually. Start with continuous or dense reinforcement for acquisition. Once responding is stable, move step by step to leaner schedules: larger ratios, longer intervals, longer delays to the reinforcer, or smaller amounts of contrived reinforcement. Use data-based criteria (for example, stable responding across several sessions) before each step.
• Intermittent schedules build persistence. Behavior reinforced intermittently usually lasts longer when reinforcement becomes rare than behavior reinforced every time. Variable schedules are especially useful because the learner cannot predict which response will pay off.
• Watch for ratio strain. If you jump too far, too fast, responding may slow, pause, become less accurate, or give way to problem behavior. The fix is to return to the last schedule that worked and take smaller steps.
• Shift from contrived to natural reinforcers. Pair the program's reinforcers with the consequences that naturally follow the behavior, such as a peer's reply, a finished snack, or a teacher's smile, then fade the contrived ones as the natural ones take over.
• Teach the learner to recruit reinforcement. Natural environments often reinforce desirable behavior too rarely. Teaching a learner to show finished work or ask for feedback increases how often natural reinforcement arrives.
• Thin carefully after functional communication training (FCT). When a replacement request has to be honored less often, signaled approaches (such as a multiple schedule with a clear "available" and "not right now" cue) or gradually increasing delays tend to keep problem behavior lower than abrupt, unsignaled changes. Less reinforcement for the replacement can also produce resurgence, so monitor closely.
• Probe for maintenance. Collect data after parts of the intervention are withdrawn, not just at mastery. Maintenance is a claim you test, not one you assume.

Definitions

• Response maintenance: the degree to which a learner keeps performing a target behavior after some or all of the intervention that produced it has ended.
• Schedule thinning: gradually reducing how often or how quickly reinforcement follows a behavior while keeping responding stable.
• Ratio strain: a breakdown in responding (pausing, slowing, errors, or problem behavior) when a response requirement is increased too quickly.
• Natural reinforcers: consequences that typically follow a behavior in the learner's everyday environment without being arranged by the program.
• Contrived reinforcers: consequences arranged by the program to support learning, such as tokens or edibles, that would not usually follow the behavior otherwise.
• Indiscriminable contingencies: arrangements in which the learner cannot tell which responses or occasions will produce reinforcement, so responding continues across occasions.
• Recruiting reinforcement: teaching the learner to prompt others to notice and reinforce the behavior, such as showing finished work and asking for feedback.

Examples from a session

Common mistakes / exam traps

• Thinning is not stopping. Withholding reinforcement all at once is extinction, not schedule thinning, and it can bring extinction bursts or resurgence.
• Intermittent reinforcement is for maintenance, not acquisition. Starting a brand-new skill on a lean schedule slows learning.
• Natural does not mean automatic. A natural reinforcer can be social, such as a friend's reply. It is "natural" because it usually follows the behavior in daily life.
• Train and hope. Ending a program at mastery with no thinning, no natural reinforcers, and no follow-up probes is the classic recipe for losing gains.

Quick check

Related topics

• Recognizing response maintenance (B.15)
• Simple schedules of reinforcement: FR, VR, FI, and VI (B.9)
• Programming for stimulus and response generalization (G.15)
• Positive and negative reinforcement procedures (G.1)
• Planning to prevent relapse (H.5)