Improving the Generalization of Skills in Learners with Autism

Individuals with autism spectrum disorders have many difficulties in learning. One of the consistent characteristics of learners with autism is that they have difficulty transferring skills to new situations and environments and maintaining skills they have mastered. Such difficulties in the maintenance and generalization of skills have been noted since the disorder was first identified, and continue to be a source of clinical concern and a focus of educational programming.

Everyone who has worked with or known many individuals with autism can think of examples of their lack of generalization. For example, a student may be able to respond when someone greets him or her saying “hello,” but not when he or she is greeted with an equivalent but different greeting such as “hi,” “what’s up,” or “how is it going?” Similarly, they may be able to make a sandwich adequately, but fail to do so if the type of bread or brand of jelly is changed.

It is certainly a characteristic of learners with autism that such transfer of skills is a challenge. Such characteristics make it necessary for teachers to train with variability.

Students need to be prepared for the diversity of circumstances that they are likely to encounter in the natural environment. Teachers can use a variety of educational strategies to facilitate the generalization of skills.

As Baer (1999, p. 1) said, “No one learns a generalized lesson unless a generalized lesson is taught.” Since generalization often does not occur without skillful planning, it is imperative that such planning occurs in educational programming for learners with autism.

Instructional Strategies

The field of Applied Behavior Analysis has studied the transfer of skills in individuals with autism for many years. It is clear that generalization is enhanced by incorporating variability into instruction. ABA instruction has emphasized the importance of varying instructions, varying materials, and teaching functionally equivalent responses. Whenever possible, learners are taught with variability from the earliest stages of instruction.

Stokes and Baer (1977) outlined a number of specific strategies to improve the transfer of skills, and emphasized that generalization should not be a “train and hope” approach. They suggested that behavior analysts must plan for generalization in a systematic manner to ensure that the target behavior occurs in similar settings and that desirable responses are strengthened. They suggested a variety of instructional approaches to aid generalization. We will discuss two of these strategies in this article – training loosely and programming common stimuli. One procedure for increasing the likelihood of generalization is what Stokes and Baer refer to as training loosely.

Excessive standardization of instruction (i.e., having every aspect of the instructional situation the same every time) impedes both stimulus and response generalization. Training loosely involves varying as many noncritical dimensions of the antecedent stimuli as possible during instruction and accepting a wide range of correct responses to increase the likelihood that skills will generalize to the natural setting.

When behavior analysts train loosely, they vary antecedent stimuli in a systematic manner. Baer (1981, 1999) suggested varying such stimuli as position (therapist or student), tone, words, how stimuli are presented (e.g., from different angles), settings in which instruction occurs, clothing worn by the therapist, reinforcers offered, time of day of instruction, and other environmental characteristics such as persons present, lighting, temperature, smells, and noise. To maximize the benefits of training loosely, these variations should occur as unpredictably as possible. When a student is learning how to match pictures, the teacher may change the pictures, the area of the classroom or school in which the skill is practiced, and whether the task is done on a table or on a vertical board. Each variation does not significantly change the task itself, but all of the changes help the learner to tolerate minor changes in the instructional context. In general, this helps to prepare the learner for the wide variety of situations that he or she may encounter outside of the instructional context.

Training loosely is often a challenge for teachers, as it seems counter-intuitive and inconsistent with the emphasis on consistency in instruction. The art of ABA intervention involves understanding which instructional components must be consistently presented and which components can be varied.

Another strategy that can be incorporated into instruction is programming common stimuli. Specifically, programming common stimuli involves incorporating stimuli and typical features of the generalization (natural) environment into the instructional setting to increase the likelihood of generalization. For instance, if a learner is being taught to purchase food in a grocery store, it may be appropriate to teach the component skills necessary for this complex skill in a controlled setting such as a classroom initially (quickly moving into the community). When this is done, it is important to use actual stimuli that the learner may encounter in a store (e.g., real food found on shelves in a supermarket, a counter, real money, a cash register, a cashier). One of the benefits of programming common stimuli is that it allows for repeated practice in a controlled setting. Practitioners need to identify the critical elements and objects present in the target environment to ensure that the learner is exposed to them in training. This eases the process of transferring skills to the natural environment.

Preparing the Learner for the Next Setting

In order to prepare the learner well for other environments, it is important to know information about those environments. Specifically, it is helpful to know how assistance is given to learners (i.e., prompting) and how learners are given feedback on their performance (i.e., reinforcement). For example, if students in an included classroom are never helped to respond with physical guidance, we can ensure that the learner is responds to other types of prompts that are used in that setting (e.g., verbal prompts, gestural prompts).

In addition, it is helpful to know how reinforcement is delivered in the target setting. It is usually the case that learners transition to environments with leaner schedules of reinforcement. Teachers in the current environment can fade the use of extrinsic rewards and provide rewards that are commonly available in the target environment. If these changes are made prior to the transition, it eases the difficulty for the learner. If a student has been reinforced for participating in group instruction with edible treats, the teacher might fade this out and replace it with the kinds of rewards that will be available in the next environment, such as positive teacher attention, and nonverbal gestures of praise). Similarly, if a learner has been used to a rich and predictable schedule of reinforcement, it may be important to thin the schedule and to make it more intermittent and less predictable. It is important to have the natural contingencies of reinforcement available in that environment maintain the behavior. Toward that end, the types and schedules of reinforcement available to the learner can be altered while the student is still in the current environment. In this way, the learner is well-prepared for the kinds of rewards they will be offered.

The Individual as the Source of Change

It is possible to involve the individual him or herself in generalization training by equipping them with skills that will increase their success in other environments. One way to accomplish this is to teach the individual to recruit their own reinforcement. Many learners have been taught to recruit teacher attention. A variety of procedures have been used in this context, including teaching students to request feedback and teaching students to show teachers their work products.

This approach has several benefits. It increases the reinforcement delivered to the learner. It also serves to cue the teacher that attention is needed. Perhaps the best element of this approach is that it ensures that teacher attention is given for appropriate behaviors. This is important, as many students receive more attention when they misbehave than they do when they behave. Finally, it is a bridge to other kinds of self-management.

Self-management facilitates the transfer of skills by equipping the individuals themselves with skills in monitoring and managing their own behaviors. Self-management involves making learners aware of and responsible for their own behaviors (Cooper, Heron, & Heward, 2007). Skills taught include planning (e.g., schedules and checklists), self-recording, and self-reinforcement. Self-management allows for more independence and decreases the need for an external agent of behavioral change.


Generalization must be a priority in educational planning. It should impact how we conceptualize and define target skills, how we teach skills, and how we evaluate progress and mastery. We must program, plan, and assess for generalization throughout instruction. The ultimate indicator of successful instruction is the transfer of skills into natural environments and everyday interactions.

Cecelia M. McCarton, MD is the founder and CEO of The McCarton Foundation. Mary Jane Weiss, PhD, BCBA is the Director of Research and Training at the Douglass Developmental Disabilities Center and a Research Associate Professor at Rutgers University. She consults to The McCarton School. Ivy Feldman, PhD is Educational Director and Jackie Hickie, MA, BCBA is Associate Educational Director at the McCarton Foundation.

Have a Comment?