Stimulus generalization, response generalization and maintenance
All three describe behavior change that reaches beyond the teaching sessions, which is why they blur together. They differ in what spreads beyond training.
The BACB's 6th edition outline covers them in four tasks: distinguishing stimulus from response generalization (B.14), identifying examples of response maintenance (B.15), and designing procedures to promote generalization (G.15) and to maintain behavior change after intervention (G.16).
One question sorts them
Ask: what was never part of training?
| What's new | What stays the same | Original example | |
|---|---|---|---|
| Stimulus generalization | The stimulus conditions: a new person, place, material or situation | The response | A child taught to say "hi" to her therapist also says "hi" to the receptionist |
| Response generalization | The response form | The function, and often the setting | A learner taught to say "hi" also starts saying "hello" and waving |
| Response maintenance | The time: after some or all of the intervention has ended | The response | A month after teaching stopped, the child still greets the therapist |
Setting or situation generalization is a common name for stimulus generalization across people and places. If an item names a new setting, that's usually the term it wants.
Stimulus generalization and discrimination
Stimulus generalization and discrimination sit at opposite ends of one continuum.
- Generalization: the response occurs in the presence of untrained stimuli that share features with the training stimulus. Control is loose.
- Discrimination: the response occurs in the presence of the trained stimulus and not others. Control is tight.
Too much generalization is a problem too. A toddler who calls every man "Daddy" is overgeneralizing. The fix is discrimination training: reinforce the response only in the presence of the right stimulus.
Response generalization
Response generalization is about form, not place. An untrained response appears that does the same job as the trained one.
- Example: after learning to wipe the table with a cloth, a learner also wipes it with a paper towel.
- Non-example: a learner wipes the table at home after learning at school. That's the same response in a new setting, so it's stimulus or setting generalization.
Promoting generalization (G.15)
Stokes and Baer (1977) argued that generalization should be programmed actively, not left to chance. Their review grouped generalization techniques under nine headings. The first, "Train and Hope", describes teaching and simply waiting to see whether the behavior spreads. Others include:
- Train sufficient exemplars: teach with several people, settings and materials.
- Program common stimuli: bring features of the target setting into teaching.
- Train loosely: vary non-critical features of teaching on purpose.
- Introduce to natural maintaining contingencies: teach behavior that the everyday environment will reinforce.
- Use indiscriminable contingencies: make it hard for the learner to tell when reinforcement is available.
Maintenance (G.16)
Maintenance is about time. The same behavior continues after the intervention is reduced or withdrawn. Common tactics include:
- Schedule thinning: gradually moving from continuous to intermittent reinforcement.
- Transferring to naturally occurring reinforcers: making sure the behavior contacts reinforcement that will keep happening once teaching stops.
- Teaching to fluency and planning follow-up probes so problems show early.
Plan for maintenance from the start, not after the program ends.
Common traps
- Calling a new setting "maintenance". A new place is generalization. Maintenance needs a gap in time after intervention changes.
- Calling a new setting "response generalization". If the response is the same, it isn't response generalization.
- Assuming maintenance means no reinforcement at all. Natural reinforcement often keeps a behavior going. Maintenance means the programmed intervention has been reduced or removed.
- Treating one follow-up probe as one concept. A probe in a new setting weeks later tests both generalization and maintenance. Answer the dimension the question asks about.
- Forgetting overgeneralization is still generalization. It's just unwanted.
Practise it
For each scenario, write new stimulus, new response or later time before looking at the options. If two apply, reread the question stem to see which one it asks about.
The lessons on B.14: stimulus vs response generalization, B.15: response maintenance, G.15: promoting generalization and G.16: making behavior change last cover each in depth. The free diagnostic shows where you're strong and where to focus.
FAQ
What is the difference between stimulus and response generalization? In stimulus generalization, the same response occurs under new conditions. In response generalization, a new, untrained response occurs that serves the same function.
What is the difference between generalization and maintenance? Generalization is about new stimuli or new response forms. Maintenance is about the same behavior continuing over time after the intervention is reduced or removed.
Is overgeneralization a type of stimulus generalization? Yes. It's stimulus generalization that goes too far, and discrimination training is the usual remedy.
How do you promote maintenance? Common tactics include thinning the reinforcement schedule and transferring control to reinforcers that occur naturally in the learner's environment.
Sources
- BACB, BCBA Test Content Outline (6th ed.): https://www.bacb.com/wp-content/bcba-outline-6thEd (tasks B.14, B.15, G.15 and G.16)
- Stokes, T. F., & Baer, D. M. (1977). An implicit technology of generalization. Journal of Applied Behavior Analysis, 10(2), 349–367: https://pmc.ncbi.nlm.nih.gov/articles/PMC1311194/ (generalization as something to program, and the nine categories of generalization techniques)