Executive Function
Stimulus Equivalence Task
Also called: stimulus equivalence paradigm, equivalence class formation, matching-to-sample equivalence test, derived relations test, emergent relations test
Measure stimulus equivalence: participants learn by feedback which shapes go together, A with B and B with C, then show whether B goes with A, A with C and C with A without being taught, scored as the share of class-consistent choices.
8-14 minUntimed
Free to start: 10 free participants per account, then $0.99 per participant for academic research, or any yearly plan. Try it here first, with no account.
- Papers, 2021 to 2025
- 167
- Evidence grade
- D
- Registered clinical studies
- 0
What it measures
Measures whether relations that were never taught emerge from ones that were. Participants learn by feedback which of three shapes goes with a sample, A with B and B with C, for three classes of new shapes, then meet test trials without feedback: B with A (symmetry), A with C (transitivity) and C with A (equivalence). The main score is the share of class-consistent choices on those untaught trials.
Constructs
- Stimulus equivalence
- Derived relations
- Conditional discrimination
- Learning from feedback
- Symbolic behavior
Research uses
- Behavior analysis research on derived relations and equivalence-based instruction
- Language and reading research on relations between words, pictures and symbols
- Comparing training structures, criteria and test protocols within a study
How the paradigm works
The stimulus equivalence task measures whether relations that were never taught emerge from relations that were (Sidman, 1971; Sidman and Tailby, 1982). Each trial shows a sample shape above three choices from another set, one from each of three classes. By feedback, participants learn that A1 goes with B1, A2 with B2 and A3 with B3, then that B1 goes with C1 and so on, then both kinds mixed, each stage ending after a block with at most one error. The test, without feedback, mixes the trained trials with untaught ones: B with A and C with B (symmetry), A with C (transitivity) and C with A (equivalence). Choosing the shape of the sample's class on untaught trials shows that the shapes have formed classes, a central method of behavior analysis for studying derived relations, language and symbolic behavior.
Scores
- Class-Consistent Choices, Untaught Trials%
- Share of untaught test trials (B with A, C with B, A with C, C with A) answered with the shape of the sample's class.
- Higher means more derived relations; 0.33 is chance
- Symmetry%
- Class-consistent share on the B with A and C with B trials, the trained relations reversed.
- Higher means more symmetry; 0.33 is chance
- Transitivity%
- Class-consistent share on the A with C trials.
- Higher means more transitivity; 0.33 is chance
- Equivalence%
- Class-consistent share on the C with A trials, which need symmetry and transitivity together.
- Higher means more equivalence; 0.33 is chance
- Trained Relations in the Test%
- Share right on the trained A with B and B with C trials of the test, without feedback.
- Below 0.9 is flagged for review
- Equivalence Shown0/1
- 1 when the trained trials and each kind of untaught trial reach 90% class-consistent choices.
- 1 means the classes formed by a 90% criterion
- Learning Trialstrials
- Trials the three learning stages took; at least 36.
- Fewer means faster learning
- Learning Errorscount
- Wrong choices over the three learning stages.
- Lower is better
- Stages Learnedcount
- Learning stages with a block of at most one error.
- 3 means every stage was learned
- Learning Complete0/1
- 1 when all three learning stages reached their criterion.
- 0 is flagged for review
- Median Trained Trial Timems
- Median choice time on the trained trials of the test.
- Informational
- Median Untaught Trial Timems
- Median choice time on the untaught trials of the test.
- Informational
- Anticipationscount
- Presses within 100 ms of the shapes appearing, in learning and the test. They are not taken as choices.
- Quality check
- Test Choices Madecount
- Test choices in the scored run.
- Fewer than planned is flagged for review
- Planned Test Trialscount
- Test trials in the session: 18 per test block.
- Informational
Use and evidence
167 papers from 2021 to 2025 name this task in their title or abstract, down 22% compared with 2016 to 2020. That ranks it 200 of 711 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 35 |
| 2024 | 24 |
| 2023 | 46 |
| 2022 | 30 |
| 2021 | 32 |
| 2020 | 51 |
| 2019 | 36 |
| 2018 | 48 |
| 2017 | 34 |
| 2016 | 44 |
| 2015 | 48 |
| 2014 | 38 |
| 2013 | 29 |
| 2012 | 26 |
| 2011 | 33 |
| 2010 | 34 |
| 2009 | 30 |
| 2008 | 33 |
| 2007 | 14 |
| 2006 | 14 |
| 2005 | 13 |
| 2004 | 9 |
| 2003 | 7 |
| 2002 | 13 |
| 2001 | 21 |
| 2000 | 27 |
Use grade C
100 to 299 papers from 2021 to 2025.
Evidence grade D
Little published measurement evidence found.
- Papers on test-retest reliability
- 1
- Papers on reliability or validity
- 19
- Registered clinical studies using it as an outcome
- 0
Measured on ConductCognition
Pooled across the studies that feed our platform norms, by device group. A figure needs 50 or more people.
No pooled figure yet.
How the grades work
Use: A is 1,000 or more papers from 2021 to 2025, B 300 to 999, C 100 to 299, D 30 to 99 and E fewer than 30. Evidence: A means at least 25 papers report its test-retest reliability and at least 25 registered clinical studies use it as an outcome; B, at least 8 test-retest papers, and either 5 registered clinical studies or 150 reliability or validity papers; C, at least 2 test-retest papers, 30 reliability or validity papers or 3 registered clinical studies; D, little published measurement evidence found. The counts measure use. They include every paper that names the task in its title or abstract. Grades describe the published research on this task, not ConductCognition's version of it.
Source: ConductCognition test census 0.58.0, counted 13 November 2026 from OpenAlex and ClinicalTrials.gov.
Task design
The trial sequence as the test runs in ConductCognition.
- Shapes
- 9 shapes drawn by code for this task in 3 sets (A, B, C) of 3 classes, each with an outline and a color of its own. Which shape plays which role is drawn from the session's seed.
- Learning
- Three stages in a linear series: A with B, then B with C, then both mixed. Each trial shows a sample above 3 choices from the other set, one of each class, with no delay: the sample stays on screen.
- Blocks
- Blocks of 9 (18 in the mixed stage): each sample with its partner once in each place, no more than 3 answers in a row in one place. A stage ends after a block with at most one error, or after 6 blocks (standard; 3 to 10).
- Trial
- The choices stay until a click, a tap or the 1, 2 and 3 keys, with no time limit. In learning, Correct or Incorrect shows for 1,000 ms, then a 500 ms blank. Presses under 100 ms are not taken.
- Test
- 3 blocks (standard; 2 to 4) of 18, with no feedback: every sample of the trained A with B and B with C, and of the untaught B with A and C with B (symmetry), A with C (transitivity) and C with A (equivalence). 54 trials, 36 of them untaught.
- Practice
- 3 trials with gray shapes, one sample whose partner is always the same shape.
- Scores
- The share of class-consistent choices on the untaught trials (the main score) and on each kind, accuracy on the trained trials, whether all four kinds reach 90%, and the trials and errors to the learning criterion.
The shapes and their colors are this library's own. The scores describe relations learned and derived in this session; they support research and do not diagnose.
Quality checks
Each session is marked valid, needs review or invalid, with the reason, before you see its scores.
Scored rows: stimulus_equivalence_test from the scored run, a choice among three shapes with no time limit and no feedback. The chance checks use the trained trials (A with B, B with C), which have a taught answer (chance 33%); the untaught trials are the measure. Accuracy below chance invalidates and accuracy not above chance is flagged. One place on 90% of choices invalidates when accuracy is not above chance; the taught answer is in each place equally often. Answers under 200 ms on at least half the choices invalidate. A learning stage that ended without a block of at most one error, fewer than 90% right on the trained trials of the test, and a run short of its planned choices are flagged for review and withhold the percentile.
Why these rules
Accuracy significantly below chance (one-sided exact test, p < .05) cannot come from guessing; it means the response mapping was reversed or the task was not followed (Slick, Sherman and Iverson 1999 treat below-chance forced-choice performance as definite response bias). Chance-level accuracy can be a real result in impaired participants, so it is only flagged. One option on at least 90% of answered trials, when the correct options are balanced and accuracy is not above chance, means the answers did not depend on the items (the task counterpart of long-string responding in surveys; Curran 2016). An answer under 200 ms is faster than a simple visual reaction, so it cannot reflect the item. Following response-time effort screening for untimed tests (Wise and Kong 2005), the session is invalid only when at least half the answers are that fast. Derived relations mean something only once the trained ones are learned and kept: equivalence studies test after a mastery criterion and count a class as formed only while the trained relations hold in the test (Arntzen and Steingrimsdottir, 2017; Silvestre and colleagues, 2026).
Running this test
Settings you can change
| Setting | Standard | Allowed |
|---|---|---|
| Most blocks per learning stage | 6 | 3 to 10 |
| Test blocks | 3 | 2 to 4 |
Yearly plans can change settings within these limits. Free and pay-as-you-go studies use the standard settings.
Need a setting this test does not have?
Devices and languages
- Untimed test: runs on desktops, laptops, tablets and phones. Participant devices
- Languages: English, Spanish, French, German, Italian, Portuguese (Brazil), Chinese (Simplified) and Japanese.
- Available on: Every plan, including the free plan.
- Recruiting online? See the setup guides for Prolific, Sona Systems, CloudResearch Connect and Qualtrics.
Cite and describe
Participants completed the Stimulus Equivalence Task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Sidman, 1971; Sidman & Tailby, 1982), with the standard settings (most blocks per learning stage: 6; test blocks: 3). The main score was class-consistent Choices, untaught trials (share of untaught test trials (B with A, C with B, A with C, C with A) answered with the shape of the sample's class).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Sidman, M. (1971). Reading and auditory-visual equivalences. Journal of Speech and Hearing Research, 14(1), 5-13. https://doi.org/10.1044/jshr.1401.05Matching to sample, which establishes relations between stimuli and then tests relations never taught.
- Sidman, M., & Tailby, W. (1982). Conditional discrimination vs. matching to sample: An expansion of the testing paradigm. Journal of the Experimental Analysis of Behavior, 37(1), 5-22. https://doi.org/10.1901/jeab.1982.37-5The testing paradigm: three classes of three stimuli, taught relations and the untaught relations that define equivalence classes.
- Arntzen, E., & Steingrimsdottir, H. S. (2017). Electroencephalography (EEG) in the study of equivalence class formation. An explorative study. Frontiers in Human Neuroscience, 11, 58. https://doi.org/10.3389/fnhum.2017.00058A computer procedure for adults: three classes, training blocks to a mastery criterion of 14 of 15, a test without consequences and a 90% criterion for each kind of trial.
- Silvestre, M. H., Harte, C., Passarelli, D. A., & de Rose, J. C. (2026). Comparing delayed matching to sample with three variations of the training-IRAP for establishing derived relations. Journal of the Experimental Analysis of Behavior, 125(1), e70082. https://doi.org/10.1002/jeab.70082Linear series training, Correct and Wrong for 1 s after each choice, and probes for derived relations interspersed with trained trials without consequences.
Source papers
- No normative data for this task.Paradigm source, N/A
Questions
How long does the Stimulus Equivalence Task take?
About 8 to 14 minutes, as listed in the test library.
What device does a participant need?
Untimed test: runs on desktops, laptops, tablets and phones. The browser must be Chrome, Edge, Firefox or Safari.
What scores does it produce?
Class-Consistent Choices, Untaught Trials, Symmetry, Transitivity, Equivalence, Trained Relations in the Test, Equivalence Shown, Learning Trials, Learning Errors, Stages Learned, Learning Complete, Median Trained Trial Time, Median Untaught Trial Time, Anticipations, Test Choices Made and Planned Test Trials. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Stimulus Equivalence Task used in research?
167 papers from 2021 to 2025 name it in their title or abstract, down 22% compared with 2016 to 2020. That ranks it 200 of 711 test families in the ConductCognition test census (use grade C). Source: ConductCognition test census 0.58.0, counted 13 November 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Stimulus Equivalence Task have?
In the published research, 1 paper reports its test-retest reliability and 0 registered clinical studies use it as an outcome (evidence grade D in the ConductCognition test census).
Which settings can a researcher change?
Most blocks per learning stage (3 to 10) and test blocks (2 to 4). Settings can be changed on yearly plans; free and pay-as-you-go studies use the standard settings.
Which languages can participants use?
Participants can take this test in English, Spanish, French, German, Italian, Portuguese (Brazil), Chinese (Simplified) and Japanese. English is on every plan, and the other languages come with yearly plans. Have a native speaker read the machine-translated text before launch.
Is this a clinical test?
No. ConductCognition is for research use only. It runs, scores and exports the test; clinical interpretation is outside the platform.
Every test: the full test list.
Run this test in your next study
Start free with 10 free participants on any test, then use participant credits or a yearly plan.