Study Notes on Generalization Technology in Behavior Analysis

Discrimination has traditionally been regarded as an active process supported by a technology with established procedures.
Generalization is viewed as a passive outcome of inadequate practice of discrimination.
This paper argues for treating generalization as deserving of an active conceptualization and associated technology.
The review categorizes studies aimed at assessing or programming generalization into nine general headings.

Historically seen as a passive phenomenon akin to a natural outcome from behavior-change processes.
Generalization emerged from varying stimuli, leading to the idea that it occurs when teaching is not controlled tightly.
The predominance of discrimination as an active process led to a lack of attention to program strategies for generalization.
Notable works regarding generalization include Skinner (1953) and Keller & Schoenfeld (1950).

Generalization is crucial for therapeutic behavioral change to be effective over time, across different persons and settings, and it often requires the behavior change to extend across related behaviors.
It is acknowledged that achieving generalization is not automatic and requires proactive programming to facilitate it.

The literature on generalization indicates a need for refined specific principles and techniques beyond traditional concepts.
A review of 270 applied behavior analysis studies revealed that 120 contribute directly to a technology of generalization.

This is the most common method for examining generalization in applied behavior analysis.
After behavior change, generalization is noted but not actively pursued, with researchers hoping that some generalization will occur.
Example:
- Kifer et al. (1974) documented generalization in negotiation behaviors among parent-child pairs in home settings after training in a classroom.
Conclusion: While many studies report generalization, they lack comprehensive analysis and often do not extend findings to broader contexts.

A more systematic approach is taken if initial generalization is absent or deficient.
Research involves initiating procedures for desired changes across responses, subjects, settings, or experimenters.
Example:
- Meichenbaum et al. (1968) applied interventions across both afternoon and morning settings for behavior change.
This category formalizes therapeutic practices for generalization and emphasizes working in multiple relevant contexts.

Focus on transferring control from the teacher to stable natural contingencies that will maintain learned behaviors.
Example:
- Ayllon and Azrin (1968) discussed introducing children to peer interactions in preschool settings to promote natural reinforcement.
This technique leverages natural environments to sustain learned behaviors without continued therapeutic intervention.

To achieve generalization, teachers should provide various examples of the same behavior.
Example:
- Stokes et al. (1974) demonstrated generalization across training with different experimenters to facilitate greeting behaviors in children.
Diverse training environments and personal exemplars positively influence generalization outcomes.

Engaging in less restrictive teaching environments leads to broader real-world application of learned skills.
Example:
- Schroeder and Baer (1972) found that allowing a greater range of stimuli in training resulted in better generalization of vocal imitations.
The method suggests that variability in instructional contexts enhances transferability of learned responses.

Reinforcement arrangements that are unpredictable (i.e., indiscriminable) foster generalization because learners cannot readily determine the conditions of reinforcement.
Example:
- Schwarz and Hawkins (1970) used delayed reinforcement with a child’s behaviors to improve generalization in different classrooms.
Such unpredictability in reinforcement settings leads to broader behavior generalization.

Ensures the presence of stimuli that are common across training and generalization contexts to support learning transfer.
Example:
- In studies by Stokes and Baer (1976), peers acted as common stimuli to facilitate learning.
The effective use of familiar and functional reinforcers can enhance the likelihood of generalization where it is applied.

Techniques may involve using intermediary responses (commonly language) to bridge learning and application across contexts.
Example:
- Risley and Hart (1968) highlighted how verbal reports helped children modify their play behavior when contingencies were aligned with their disclosures.
Such strategies aim to create meaningful connections between learned behaviors and their applications.

Reinforcing the act of generalization itself as a desired behavior can yield significant results.
Example:
- Goetz and Baer (1973) rewarded variations in block forms children created, promoting exploration and generalization in their play.
Encouraging learners explicitly to make connections across examples can foster an active generalization process.

The paper concludes that the literature reflects a nascent technology for generalization programming, which is crucial for behavior analysis.
Categories outlined provide a framework for further research and practice enhancement.
Emphasis on systematic programming for generalization will advance therapeutic methods and outcomes, necessary for effective behavior change in diverse contexts.