5.2 Generalization Tactics (Stokes & Baer), Maintenance & Self-Management
Key Takeaways
- Stimulus generalization occurs when a response is evoked by untrained stimuli sharing physical or functional properties with the trained SD.
- Response generalization refers to engaging in topographically different, untrained responses that accomplish the same functional outcome as the target behavior.
- Stokes and Baer (1977) categorized nine tactical principles for promoting generalization, emphasizing that generalization must be actively programmed rather than left to chance.
- Tactics such as Training Sufficient Exemplars, Program Common Stimuli, and Mediate Generalization directly build resistance to stimulus overselectivity and environmental novelty.
- Indiscriminable contingencies maintain behaviors over time by making schedule shifts unpredictable, ensuring responses persist in the natural environment without dense reinforcement.
5.2 Generalization Tactics (Stokes & Baer) and Maintenance
A behavior change intervention is clinically successful only if the target behavior occurs in non-training environments, with different instructors or caregivers, across novel materials, and persists long after formal training has ended. In Applied Behavior Analysis, this overarching outcome is divided into generalization and maintenance. Rather than assuming that behavior change will naturally spread outside the clinical setting, behavior analysts must program for generalization from the very onset of intervention design. For the Qualified Behavior Analyst (QBA) candidate, mastering the distinction between stimulus and response generalization, alongside the seminal framework of Stokes and Baer (1977), is essential for constructing robust, durable treatment plans.
Core Concepts: Generalization and Maintenance
Stimulus Generalization vs. Response Generalization
Understanding the boundary between stimulus generalization and response generalization is a frequent area of evaluation on credentialing exams:
- Stimulus Generalization: Occurs when a trained target response is evoked by antecedent stimuli that were not present during direct instruction. These untrained stimuli typically share physical, visual, auditory, or functional properties with the original discriminative stimulus ($S^D$).
- Clinical Example: A child is taught to say "Hello" when their therapist enters the clinic room. Without explicit training, the child also says "Hello" when their grandmother, teacher, or peer enters the room. The behavior (saying "Hello") remains constant, but the controlling antecedent stimuli expand.
- Response Generalization: Occurs when a learner engages in topographically different, untrained responses that produce the same functional outcome or reinforcement as the trained target behavior.
- Clinical Example: A client is directly taught to greet peers by waving their hand. Later, in the presence of a peer ($S^D$), the client spontaneously says "Hi there" or nods their head. The stimulus remains constant, but the topography of the response varies while achieving the same social function.
Maintenance Over Time
Maintenance refers to the persistence of a trained behavior over time after formal instruction, prompt systems, or high-density reinforcement schedules have been reduced or removed completely. While generalization addresses expansion across stimuli and responses, maintenance addresses temporal durability. Behavior analysts evaluate maintenance by conducting maintenance probes weeks or months following intervention termination under naturalistic contingency conditions.
Stokes & Baer (1977) Nine Generalization Tactics
In their landmark paper, An Implicit Technology of Generalization, Donald Stokes and Donald Baer (1977) surveyed the literature and categorized nine distinct tactics for promoting generalization. They emphasized that historically, clinicians relied on "Train and Hope"—an ineffective passive approach—and urged the field to adopt active programming strategies.
| Tactic | Description | Clinical Implementation Example |
|---|---|---|
| 1. Train and Hope | Teaching a behavior without active generalization programming and merely testing if it generalizes. | Not recommended. Training greeting in a clinic and hoping it occurs at school without intervention. |
| 2. Sequential Modification | Systematically training the target behavior in every non-generalized setting, person, or context sequentially. | Training toothbrushing in the clinic, then at home, then at school after failures are documented in each location. |
| 3. Introduce to Natural Maintaining Contingencies | Selecting target behaviors that will naturally contact reinforcement in the learner's natural environment. | Teaching a child to request preferred snacks vocally so family members naturally reinforce the mand at home. |
| 4. Train Sufficient Exemplars | Teaching multiple variations of stimulus conditions and response topographies until generalization occurs spontaneously. | Training a student to identify "chair" using dining chairs, beanbags, office chairs, and stools until all chairs are identified. |
| 5. Train Loosely | Intentionally varying non-essential antecedent stimuli, instructions, locations, and instructors during teaching trials. | Varying tone of voice, seating arrangements, background noise, and wording ("Point to car", "Show me car", "Where is car?"). |
| 6. Program Common Stimuli | Incorporating salient physical or social features of the natural target environment directly into the training setting. | Setting up a mock grocery store inside a clinic using real price tags, shopping carts, and register sounds before community outings. |
| 7. Mediate Generalization | Teaching the learner a specific response (e.g., self-prompt, rule, visual card) that acts as a portable mediator across settings. | Teaching an adult to use a written grocery checklist that they carry from home to the supermarket. |
| 8. Train to 'Generalize' | Directly reinforcing instances of novel, varied, or generalized responding when they occur. | Using a Lag-1 reinforcement schedule where praise and tokens are delivered only if the learner provides a greeting different from the previous trial. |
| 9. Indiscriminable Contingencies | Thinning reinforcement to intermittent schedules so the learner cannot predict when reinforcement will be delivered. | Transitioning from continuous FR1 reinforcement to a Variable-Interval (VI) or Variable-Ratio (VR) schedule. |
In-Depth Analysis of Key Generalization Tactics
Programming Common Stimuli and Training Loosely
- Program Common Stimuli: This tactic reduces the novelty of the natural setting. By analyzing the natural generalization environment (e.g., a general education classroom) and introducing its key elements into the intervention environment (e.g., identical desk arrangements, noise levels, work folder formats), the clinician ensures that the stimulus control established during training transfers seamlessly.
- Train Loosely: Traditional instructional methods emphasize rigid standardization of antecedent conditions. However, rigid training can result in stimulus overselectivity—where the behavior occurs only under hyper-specific conditions (e.g., only when the instructor speaks in a specific tone while sitting in a specific chair). Training loosely intentionally breaks down irrelevant stimulus control by altering non-critical aspects of the presentation.
Training Sufficient Exemplars and General Case Analysis
When implementing Train Sufficient Exemplars, behavior analysts often utilize General Case Analysis (Horner & Albin, 1986). This systematic method involves:
- Defining the instructional universe (all contexts where the behavior should occur).
- Identifying the range of stimulus variations and response requirements across those contexts.
- Selecting a representative subset of teaching exemplars that encompasses the full range of variability.
- Teaching to criterion across those selected exemplars until untaught exemplars are correctly responded to.
Indiscriminable Contingencies and Maintenance Mechanics
To maintain behavior long-term, the learner must not be able to discriminate between periods when reinforcement is available and when it is absent. If reinforcement is delivered on a predictable Fixed-Ratio 1 (FR1) schedule, the behavior rapidly undergoes extinction when reinforcement stops. By thinning schedules of reinforcement to Variable-Ratio (VR) or Variable-Interval (VI) schedules—or utilizing delayed reinforcement procedures—the contingency becomes indiscriminable. The learner continues responding at high, steady rates because any given response might produce reinforcement.
Programming for Generalization from Intervention Onset
A central tenet of QBA practice is that generalization planning must occur prior to intervention kickoff, integrated directly into the initial behavior acquisition plan.
Step-by-Step Clinical Workflow for QBAs
- Conduct a Functional Assessment of the Natural Environment: Identify the natural contingencies, typical SDs, available reinforcers, and social expectations of the settings where the skill must function.
- Apply the Relevance-of-Behavior Rule: Ensure that every target behavior selected for instruction will contact naturally occurring reinforcement once prompts are faded. If a behavior is not reinforced naturally, it will inevitably extinguish.
- Select Active Tactics: Combine multiple Stokes & Baer tactics into the baseline protocol (e.g., train sufficient stimulus exemplars + train loosely + program common stimuli).
- Plan Schedule Thinning Early: Systematically fade artificial reinforcers and transition from continuous to intermittent, indiscriminable schedules prior to discharging the intervention.
- Conduct Regular Probe Sessions: Schedule systematic generalization and maintenance probes in natural settings without prompts to monitor skill retention and expansion.
Self-Management Procedures
Self-management teaches the learner to observe, evaluate, and consequate their own behavior so skills maintain when therapists are absent—a core Skill Acquisition competency on the QBA exam and a pivotal repertoire for independence.
| Component | Definition | Example |
|---|---|---|
| Self-monitoring | Learner records occurrences of their own target behavior | Student tallies on-task intervals on a card |
| Self-evaluation | Learner compares performance to a criterion | Checks whether homework checklist is 100% complete |
| Self-reinforcement | Learner delivers a reinforcer contingent on meeting the criterion | Accesses preferred break after accurate self-check |
| Self-instruction | Learner uses private or overt verbal rules to guide responding | "First look at the menu, then order, then pay" |
Implementation steps: (1) define the target operationally; (2) teach accurate self-monitoring to match adult IOA; (3) add evaluation against a clear criterion; (4) arrange self-delivered or adult-confirmed reinforcers; (5) fade adult oversight while probing maintenance across settings.
Clinical vignette: A teen practices self-monitoring of vocal stereotypy during homework. After achieving ≥90% agreement with a therapist for three sessions, the teen self-evaluates against a "quiet voice for 10 minutes" goal and self-delivers a token. Stereotypy decreases and homework completion maintains at home—where therapist-delivered DRA is unavailable.
Self-management is not "tell them to try harder." It is a taught contingency package that transfers stimulus control from the therapist to the learner's own recording and rule-governed behavior.
A client who was taught to request assistance by saying 'Help please' in the clinic spontaneously begins raising their hand and saying 'Can you assist me?' when experiencing difficulty in a classroom setting. Which behavioral phenomenon does this demonstrate?
To prepare a learner for community vocational outings, a QBA arranges the clinic therapy space to match the exact physical layout, cash register sounds, and shelf structures of the local retail store. According to Stokes and Baer (1977), which generalization tactic is being utilized?
A behavior analyst transitions a client's reinforcement schedule from a continuous Fixed Ratio 1 (FR1) schedule to an unpredictable Variable Ratio 5 (VR5) schedule before fading therapy sessions. What is the primary clinical rationale for this schedule thinning tactic?