Translated from the Portuguese original. Read the original (Português)
Generalization is when a skill learned in one context starts appearing in other places, with other people and materials (stimulus generalization) or in new ways of responding that were not directly taught (response generalization). Maintenance is the skill continuing to happen over time, after intensive teaching has ended. In Applied Behavior Analysis, generalization and maintenance are not a stroke of luck: they must be planned from day one of the program.
We speak of generalized behavior change when the target behavior occurs under conditions different from those of training — other settings, people, or stimuli (stimulus generalization) —, when it appears in the form of functionally equivalent responses that were not trained (response generalization), or when it persists over time after the intervention has been withdrawn or reduced (maintenance). Since Baer, Wolf & Risley (1968), this combination of durability and breadth — called generality — is one of the dimensions that define Applied Behavior Analysis.
Why this matters so much
Imagine a child who learns to ask for water by pointing at the cup, sitting at the clinic table, with the same therapist, with the same blue cup. If she does not ask for water at home, at school, with her grandmother, or with a different cup, the program is not finished yet. A skill that only exists inside the therapy room does not change anyone's life.
That is why, in evidence-based practice, the question is not just “did the person learn it?”, but: where does she use what she learned, in how many ways can she use it, and for how long does it keep going after the support is reduced.
The three kinds of outcome
These are different things, and each one calls for its own strategy. Tap the tabs:
The same behavior, in other contexts
The behavior learned in the presence of a stimulus starts occurring in the presence of similar stimuli: a different setting, a different person, a different material, a different time. It is the opposite side of discrimination — here we want the response to “cross over” contexts, instead of staying tied to a single cue.
Example: the child who learned to ask the therapist for help starts asking her father, her teacher, and her classmate for help too.
New ways of doing the same thing
Responses appear that were not directly taught, but serve the same function. This is useful variability: the person does not become dependent on a single formula.
Example: after learning to say “can you help me?”, the child starts saying “I can't open this” or showing the object to the adult — different ways of getting the same result.
The skill keeps going after the program ends
Maintenance is persistence over time, once intensive teaching and the programmed reinforcers have already been withdrawn or reduced. Without planning, performance may regress once support ends — which is why withdrawal is done gradually, not all at once.
Example: six months after discharge from the program, the child keeps asking for help spontaneously at school.
“Train and hope” is not a planIn a classic review of the literature, Stokes and Baer (1977) showed that the most frequent strategy in the studies was simply to teach and wait for generalization to happen on its own — what they called “train and hope.” Generalization and maintenance need to be programmed and measured, like any other goal.
How to program generalization
There are well-described strategies in the literature. Tap the cards to see what each one means:
How to program maintenance
Maintenance is built mainly through the way support is withdrawn. Some central practices:
- Thin the reinforcement schedule: move from continuous reinforcement (every correct response) to intermittent schedules, which make the response more resistant to extinction;
- Bring the reinforcer closer to the natural one: replace artificial reinforcers with consequences that life itself offers — attention, access to what was requested, praise from the group;
- Fade the prompts: gradually reduce physical and verbal help, so the skill does not depend on an adult's presence;
- Distribute and space out practice: review already-mastered skills from time to time, instead of treating them as “done”;
- Involve the people around the person: family, school, and caregivers keep the skill going in the settings where it actually matters;
- Keep measuring: collect follow-up data after discharge, to find out whether the gain holds up.
Connection with social validityGeneralization and maintenance are the practical test of a program's social validity: an outcome is only socially relevant if it shows up in the person's everyday life and lasts. That is why a good plan defines, from the start, in which contexts the skill should work and how that will be checked.
Test your understanding
Quick quiz
Three questions to reinforce the concept. You see the explanation right away.
Further reading
- Baer D. M., Wolf M. M. & Risley T. R. “Some current dimensions of applied behavior analysis”. Journal of Applied Behavior Analysis, 1968.
- Stokes T. F. & Baer D. M. “An implicit technology of generalization”. Journal of Applied Behavior Analysis, 1977.
- Cooper J. O., Heron T. E. & Heward W. L. Applied Behavior Analysis — chapter on generalization and maintenance of behavior change.
- Baer D. M. & Wolf M. M. “The entry into natural communities of reinforcement” (behavioral traps), 1970.
This entry is educational and does not replace assessment, diagnosis or treatment by a qualified health professional. Every person is unique — for a specific case, seek individual guidance.