EDF5684 Chap.9 Stimulus Equivalence
Stimulus Equivalence
Stimulus equivalence is demonstrated when trained conditional discriminations support untrained relational responding with reflexive, symmetrical and transitive properties.
This chapter maps trained and tested arrows before writing procedures, prevents direct teaching from contaminating an emergence claim, and connects trial-level administration and data collection to the derived-relations section of the conditional discrimination artefact.
What this chapter covers
- 01
Match-to-sample teaching and baseline relations
- 02
Trained relations versus untrained tests
- 03
Reflexivity through identity matching
- 04
Symmetry through reversal of trained relations
- 05
Transitivity through combined trained links
- 06
Neutral tests, rotated positions and trial-level data
Worked example · free
Derive tests from an invented relation network
- +1List symbol to picture and picture to label as directly trained relations.
- +1Test each stimulus against itself for reflexivity without treating those trials as baseline training.
- +1Reverse the trained relations to test picture to symbol and label to picture symmetry.
- +1Combine the trained links to test symbol to label transitivity.
- +1Add the relevant reverse combined relation if claiming the full equivalence pattern.
- +1Rotate comparison positions and standardise instructions to reduce unintended cue control.
- +1Avoid differential corrective feedback that would teach the relation during the test.
Key terms
- Stimulus Equivalence
- Emergent relational responding that meets reflexive, symmetrical and transitive properties.
- Match-to-Sample
- A conditional-discrimination format with a sample and an array of comparison stimuli.
- Reflexivity
- Identity matching in which a stimulus is related to itself.
- Symmetry
- An untrained reverse relation derived from a trained relation.
- Transitivity
- An untrained relation derived by combining two trained relations through a shared stimulus.
Stimulus Equivalence FAQ
Why must trained and tested relations stay separate?
A relation demonstrates emergence only when it was not directly taught. If prompts, correction or differential feedback teach the test relation, correct performance can no longer show that the relation derived from the baseline training alone.
How do I plan a symmetry test?
Start from a documented trained relation such as A to B, then test B to A without directly teaching that reverse direction. Specify the sample, comparisons, position rotation, instruction, response code and criterion rather than writing only the word symmetry.
What can make an equivalence test misleading?
Fixed comparison positions, instructor gaze, inconsistent instructions, prompts, corrective feedback and collapsed data can all produce accurate-looking performance for the wrong reason. Standardise administration and preserve trial identity so the intended relation is the best explanation.
How should partial equivalence results be reported?
Report performance for each pre-specified untrained relation rather than collapsing all probes into a single success label. State the baseline relations, test directions, trial counts, position controls, feedback arrangement and mastery rule. Symmetry may emerge for some pairs while transitivity or combined tests remain below criterion, so a complete class should not be claimed automatically.
Distinguish first responses from corrections and note any procedural event that could teach a test relation. The conclusion should match the exact set of independent probes that met criterion and identify what remains untested.
Assessment move
Draw every trained relation before studying property definitions. Generate symmetry by reversing arrows and transitivity by combining compatible paths. Label each proposed test as untrained and check the entire procedure for prompts or feedback that would teach it. For the project, build a trial matrix with sample, comparisons, correct selection, position and response code.
Practise explaining why one correct trial is weaker evidence than a consistent pattern across the pre-specified relation set. Use solid arrows for trained conditional discriminations and dashed arrows for untrained tests. For each baseline relation, generate the exact symmetry probe by reversing direction.
For transitivity, trace two compatible paths through a shared middle stimulus and confirm that the proposed test connects their outer nodes. Add reflexivity probes only when identity matching is part of the declared test plan. This drawing routine makes accidental teaching and missing relations visible. Build a probe inventory before collecting results.
Include relation type, sample, comparison set, correct selection, position, trial count, feedback rule and mastery criterion. Intermix trials and rotate positions so sequence or location cannot solve the task. Keep first responses distinct from corrections, and ensure neutral test consequences do not differentially teach one relation.
Practise interpreting partial patterns: symmetry may appear without a complete equivalence class, and one correct trial is not the same as stable performance across the set. In project review, compare every network arrow with a row on the data sheet and a written procedural step. Verify that trained and tested relations are named consistently and that the test phase removes instructional prompts.
End by explaining the evidence conservatively: report which pre-specified untrained relations met criterion, then state what larger class claim those results support and what remains untested.
Working through Stimulus Equivalence in EDF5684? Sia is AskSia’s AI Education tutor — ask any EDF5684 Stimulus Equivalence question and get a clear, step-by-step explanation grounded in how EDF5684 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.
