Equivalence class

Classes are demonstrated through relations

Suppose A is a pharmacy symbol, B is the printed word “pharmacy,” and C is its spoken name. Teaching A-to-B and B-to-C relations may be followed by accurate responding on untrained relations among these stimuli. If the required tests are met, A, B, and C may be described as members of an equivalence class.

Shared topic or physical resemblance is not enough. Three blue cards form a visual category, but an equivalence-class claim depends on demonstrated relational properties.

Reflexivity, symmetry, and transitivity matter

The conventional tests are:

• Reflexivity: matching a stimulus to itself, such as A to A, without that identity relation being directly taught.
• Symmetry: reversing a trained relation, such as showing B to A after A to B was taught.
• Transitivity: combining trained relations, such as A to C after A to B and B to C were taught.

A combined symmetry-and-transitivity test may assess C to A. Sidman and Tailby set out this distinction between conditional-discrimination training and equivalence testing.

Matching-to-sample is the procedure

Matching-to-sample presents a sample and comparison choices. The sample changes which comparison is correct. It is often used to teach baseline relations and test emergent ones.

The procedure itself is not the outcome. A learner can master every directly taught A-to-B trial while showing little symmetry or transitivity. Sidman’s introductory tutorial reviews the experimental arrangement and its conceptual history.

Test design protects the interpretation

An equivalence probe should define the sample, comparisons, response window, reinforcement arrangement, prompt history, and mastery rule. Comparison positions need balancing. Baseline relations should be checked during testing so class performance is not inferred after taught relations have deteriorated.

Testing can itself create learning. Record probe order, repetition, feedback, and any retraining. Report results for each relation and each proposed class; one combined percentage can hide a failed member or relation.

A fictional three-member class

Imani learns two baseline relations for pharmacy navigation: symbol-to-printed-word and printed-word-to-spoken-name. She responds correctly on 12 of 12 final baseline trials.

Without feedback, the team presents 12 predeclared probes covering reversed and combined relations. Imani responds correctly on 10 of 12. That aggregate describes probe accuracy. It does not establish the full equivalence class until performance is reviewed by relation, class member, and the planned criterion. If both errors occurred in C-to-A probes, a required property may remain unconfirmed despite 83.3% overall accuracy.

Emergent performance can support efficient teaching

Equivalence-based instruction aims to produce additional useful relations after teaching a smaller set. Applications have included reading, vocabulary, academic concepts, and symbol systems. Efficiency should be measured directly: training time, number of taught relations, probe results, maintenance, and meaningful use.

A successful laboratory relation does not guarantee daily-life fluency. Generalization may require new people, fonts, devices, settings, and response forms. Preserve the person’s AAC and ordinary access while testing.

Mechanism claims require care

Researchers have proposed different accounts of equivalence and derived responding. The review by Lionello-DeNolf shows that even symmetry has a complex research history across species and procedures.

Clinicians can describe the observed relational pattern without claiming that one mechanism has been proven. Names, common reinforcement, verbal histories, contextual control, and procedure details may all matter.

Clinical use starts with relevance

Select class members that support a goal chosen by or meaningful to the person. Confirm sensory access, literacy demands, language, motor requirements, and valid ways to respond. Eye contact and speech are never required when another reliable response form works.

The BACB BCBA Test Content Outline, 6th edition includes equivalence concepts in examination content. It does not prescribe class size, training structure, or a clinical plan.

Class expansion needs fresh evidence

Adding a fourth stimulus, D, requires more than placing it beside established class members. Teach or test the planned link, then verify the new emergent relations while the older baseline remains stable. Report whether D joins every intended class member and whether errors occur in one direction.

Function transfer is another claim. If a safety symbol acquires a new response after class training, define that response and test it separately. Equivalence can support a prediction of transfer, while the observed transfer still needs its own denominator, context, and safeguards.