generalization-maintenance

name	generalization-maintenance
description	Use when programming for generalization of learned skills across people, settings, and stimuli, and for maintaining treatment gains over time using evidence-based strategies.

Generalization and Maintenance

Generalization and maintenance are not byproducts of good teaching—they must be actively programmed. A skill that only occurs in the training context with the training therapist has limited clinical value.

Definitions

• Stimulus generalization: The behavior occurs in the presence of untrained stimuli (new people, settings, materials, instructions).
• Response generalization: Untrained responses occur that are functionally equivalent to the trained response.
• Maintenance: The behavior continues over time after the intervention is reduced or withdrawn.

Stokes and Baer (1977) Technology of Generalization

The foundational framework for programming generalization in ABA.

1. Train Sufficient Exemplars

Teach the skill across enough examples that the learner abstracts the concept rather than memorizing specific instances.

• Stimuli: Use multiple materials, instructions, and settings during training. If teaching "cup," train with plastic cups, ceramic mugs, paper cups, travel mugs.
• People: Multiple trainers deliver instruction so the behavior is not under the stimulus control of one person.
• Settings: Conduct training in multiple locations rather than exclusively in the therapy room.
• Rule of thumb: Minimum 3–5 exemplars across the relevant dimension.

2. Train Loosely

Deliberately vary non-critical aspects of the training context.

• Vary your wording of instructions ("Give me the ball," "Hand me the ball," "Can I have the ball?").
• Vary seating arrangement, time of day, materials on the table.
• Vary the pace, tone of voice, and delivery style.
• Avoid inadvertently creating overly narrow stimulus control through rigid, scripted teaching.

3. Use Indiscriminable Contingencies

Make it impossible for the learner to predict exactly when reinforcement will be delivered.

• Thin the reinforcement schedule from continuous (CRF) to intermittent (VR, VI).
• Use delayed reinforcement: reinforcer is delivered minutes or hours after the behavior.
• Vary the magnitude and type of reinforcer.
• The goal: behavior persists because the learner cannot discriminate reinforced from unreinforced occasions.

4. Program Common Stimuli

Include stimuli from the generalization setting in the training setting.

• If the goal is classroom behavior, use classroom materials, a desk, and peer presence during training.
• If the goal is community skill use, practice with actual community materials (real money, real menus).
• Identify the critical stimuli in the target environment and embed them in training.

5. Mediate Generalization

Teach the learner to carry the intervention with them via self-management.

• Self-monitoring: Client tracks their own behavior.
• Self-instruction: Client uses verbal or written rules to guide behavior ("First I look, then I cross").
• Self-reinforcement: Client delivers reinforcement to themselves when they meet criteria.
• Self-management makes the learner independent of external agents.

6. Train to Generalize

Explicitly reinforce generalized responding.

• Reinforce the behavior when it occurs in novel contexts.
• Provide specific feedback: "You used your greeting with someone new—great job!"
• Create opportunities for the learner to practice in novel situations and reinforce successful generalization.

Generalization Probes

Design

• Conduct probes in untrained conditions: new people, new settings, new materials.
• Probes are conducted without teaching, prompts, or programmed reinforcement.
• Collect probe data before, during, and after training to track generalization.

Generalization Matrix

A planning tool that maps training and probe conditions across relevant dimensions.

	Therapist A	Therapist B	Parent	Teacher
Clinic	Train	Probe	—	—
Home	Probe	—	Train	—
School	—	—	—	Probe

• "Train" cells are where instruction occurs.
• "Probe" cells are where you assess generalization without training.
• Fill in the matrix during program planning to ensure systematic generalization assessment.

Response Generalization Strategies

• Teach multiple response forms that serve the same function (e.g., requesting by pointing, speaking, or using a device).
• Reinforce novel appropriate responses, not just the trained topography.
• Use general case instruction: identify the range of stimulus variations and response variations the learner will encounter and sample from that range during training.

Maintenance Strategies

Thin Reinforcement Schedules

• Gradually shift from dense, continuous reinforcement to lean, intermittent schedules.
• Match the terminal schedule to what the natural environment provides.
• Use variable schedules (VR, VI) which are more resistant to extinction than fixed schedules.

Transfer to Natural Contingencies

• Identify reinforcers that naturally exist in the target environment.
• Shape the behavior to a form that will contact those natural reinforcers.
• Example: Teaching a child to greet peers → peer responses (smiles, greetings back) become the natural reinforcer.

Teach Self-Management

• Self-monitoring with checklists or apps.
• Self-evaluation against a standard.
• Self-reinforcement for meeting goals.
• Self-management is the most portable maintenance strategy.

Booster Sessions

• Periodic retraining sessions after mastery to refresh the skill.
• Scheduled proactively (e.g., monthly for the first 3 months, quarterly thereafter).
• Triggered reactively if maintenance probes show decline.

When Generalization Fails: Troubleshooting

Checklist

1. Was generalization programmed during training, or assumed? Go back to Stokes and Baer's strategies.
2. Were sufficient exemplars used? If only one therapist in one room—add exemplars.
3. Was training too rigid? Loosen stimulus control by varying non-critical features.
4. Are the natural contingencies sufficient to maintain the behavior? If not, engineer the environment.
5. Does the learner have the prerequisite skills for the generalization context? A social skill trained 1:1 may not generalize to a group if the child lacks the prerequisite skill of tolerating proximity.
6. Is there a competing contingency in the generalization setting? Problem behavior may be more efficient in the new setting.

Community-Based Instruction

For clients whose goals involve community participation:

• Conduct training in the actual community setting (grocery store, restaurant, bus).
• Use in-vivo teaching with immediate feedback.
• Fade instructor presence and proximity over time.
• Involve natural community members (store clerks, librarians) as stimulus contexts.
• Ensure safety protocols are in place for community outings.

Key References

• Stokes, T. F., & Baer, D. M. (1977). An implicit technology of generalization. JABA, 10, 349–367.
• Cooper, J. O., Heron, T. E., & Heward, W. L. (2020). Applied Behavior Analysis (3rd ed.). Pearson.
• Horner, R. H., Sprague, J., & Wilcox, B. (1982). General case programming for community activities. In B. Wilcox & G. T. Bellamy (Eds.), Design of High School Programs for Severely Handicapped Students. Paul H. Brookes.