Why Generalization and Maintenance Matter in ABA
One of the most critical goals in applied behavior analysis is ensuring that learned skills transfer across settings, people, and time. A client who can only demonstrate a skill in the therapy room, with one therapist, using one set of materials has not truly acquired functional use of that skill. Generalization and maintenance programming must be built into every intervention plan from the start, not treated as an afterthought.
Stokes and Baer (1977) published a foundational paper outlining strategies for promoting generalization. Their work remains one of the most cited references in ABA literature, and the strategies they described continue to guide clinical practice today.
Types of Generalization
Understanding the different types of generalization is essential for both clinical practice and the BCBA exam.
Stimulus Generalization
Stimulus generalization occurs when a behavior that was reinforced in the presence of one stimulus also occurs in the presence of similar stimuli. For example, a child who learns to label a golden retriever as "dog" also labels a poodle as "dog." The behavior has generalized across different examples within the same stimulus class.
Clinically, stimulus generalization is promoted by training with multiple exemplars. If you teach a child to respond to instructions from only one therapist, the child may not follow instructions from a teacher or parent. By deliberately varying the people who deliver instructions during training, you increase the likelihood that the skill will generalize.
Response Generalization
Response generalization occurs when reinforcement of one response leads to changes in similar but untrained responses. For example, teaching a child to say "hello" as a greeting may result in the child also waving or saying "hi" without direct training on those responses.
This type of generalization is particularly important for language and social skills programming, where flexibility in responding is a hallmark of functional communication.
Maintenance
Maintenance refers to the continuation of a learned behavior after intervention has been withdrawn or reduced. A skill is only truly useful if the learner continues to use it over time. Maintenance is not the same as generalization, but the two are closely related. Skills that generalize across natural environments are more likely to contact natural reinforcement and therefore maintain over time.
Strategies for Programming Generalization
Stokes and Baer described several strategies that remain core to ABA practice.
Train and Hope
This is the default approach when no active generalization programming is included. The clinician trains a skill and simply hopes it generalizes. While this occasionally works, it is not a recommended strategy. It is included in the literature primarily as a baseline comparison for the more active strategies below.
Training Sufficient Exemplars
One of the most effective strategies involves training across enough examples that the learner abstracts the relevant features of the concept. When teaching color identification, for instance, you would use red blocks, red cars, red shirts, and red crayons rather than always presenting the same red card. Research suggests that generalization often emerges after training three to five exemplars, though this varies by learner and skill.
Training Loosely
Training loosely means varying the non-critical aspects of the teaching situation. Rather than delivering instructions with the same wording, tone, and setting every time, the clinician deliberately introduces variation. This teaches the learner to attend to the relevant discriminative stimuli rather than irrelevant features of the training context.
Programming Common Stimuli
This strategy involves incorporating stimuli from the generalization setting into the training environment. If a child will need to use a skill in a classroom, bring classroom materials, desks, or even peers into the training sessions. This creates a bridge between the two environments.
Mediating Generalization
Teaching the learner a strategy that facilitates generalization is called mediation. Self-management strategies, rule-following, and problem-solving skills all serve as mediators. A learner who can self-monitor and self-reinforce is more likely to maintain and generalize skills independently.
Using Indiscriminable Contingencies
When the learner cannot predict exactly when reinforcement will be delivered, they are more likely to continue performing the skill across settings. Thinning reinforcement schedules from continuous to intermittent before transitioning to natural environments helps create this indiscriminability.
Programming for Maintenance
Maintenance programming focuses on ensuring skills persist after formal intervention ends.
Schedule Thinning
Gradually reducing the frequency of reinforcement during the latter stages of intervention prepares the learner for the thinner schedules of reinforcement found in natural environments. Moving from a fixed ratio 1 (FR1) schedule to variable ratio schedules helps build resistance to extinction.
Transferring to Natural Reinforcers
Contrived reinforcers like tokens and edibles should eventually be replaced by or paired with naturally occurring reinforcers. A child who learns to greet peers should eventually be reinforced by the social interaction itself rather than by a token. Identifying and leveraging natural reinforcers is a key clinical skill.
Involving Caregivers and Natural Change Agents
Training parents, teachers, and other individuals in the learner's natural environment to support and reinforce target skills is one of the most powerful maintenance strategies. These individuals are present long after formal intervention ends and can provide ongoing support.
Booster Sessions
Periodic check-ins after formal intervention has ended allow clinicians to assess maintenance and provide brief retraining if needed. Building booster sessions into the treatment plan helps catch skill regression early.
BCBA Exam Considerations
For the BCBA exam, expect questions that require you to identify which generalization strategy is being described in a clinical scenario. You should be able to distinguish between stimulus generalization and response generalization, and you should know the Stokes and Baer strategies by name. Questions may also ask you to select the best strategy for a given situation or identify why generalization failed in a scenario.
Practice applying these concepts to realistic clinical vignettes. The exam tests your ability to analyze situations, not just recall definitions.