Social Cognition
Animated Shapes Task
Also called: animated triangles task, Frith-Happé animations, Heider-Simmel animations, moving shapes task
Measure how people read intentions and mental states into moving shapes: participants class short animations of two triangles as moving on their own, reacting to actions or reacting to thoughts or feelings.
8 minUntimed
Free to start: 10 free participants per account, then $0.99 per participant for academic research, or any yearly plan. Preview any test free after you sign up.
- Papers, 2021 to 2025
- 60
- Evidence grade
- C
- Registered clinical studies
- 4
What it measures
Measures how people read intentions and mental states into moving shapes. Participants watch 12 short animations of two triangles and say whether the triangles move on their own, react to each other's actions or react to each other's thoughts or feelings. The main score is percent correct on the mental-state animations.
Constructs
- Theory of mind
- Mental state attribution
- Intention reading
- Social perception
Research uses
- Theory of mind research in adults
- Autism research
- Schizophrenia and social cognition research
- Within-study group comparison
How the paradigm works
The Animated Shapes Task measures how people read intentions and mental states into moving shapes. Heider and Simmel (1944) showed that people describe two triangles and a circle moving on a film as people with goals and feelings. Abell, Happé and Frith (2000) and Castelli, Happé, Frith and Frith (2000) made three kinds of animation of a big and a small triangle: random, in which the triangles drift or bounce like objects; goal-directed, in which one triangle reacts to what the other does, as in chasing or fighting; and mental state, in which one triangle acts on what the other thinks or feels, as in coaxing or tricking. White, Coniston, Rogers and Frith (2011) replaced coded spoken descriptions with multiple-choice answers, and Livingston, Shah, White and Happé (2021) ran that version online, where accuracy on the mental-state animations was lower in autistic adults and fell with autistic traits.
More about the paradigm
The 12 animations here, 4 of each kind, were written and are drawn by ConductScience's code; no published animation is used. Each lasts 18 seconds, about the length of the shortened animations of Zwickel and colleagues (2011), with a big red and a small blue triangle on a white stage and, in 3 animations of each kind, a house with an open door. While an animation plays, the participant chooses its kind: on their own, reacting to actions, or reacting to thoughts or feelings. The first choice counts and the animation always plays to the end, as in Livingston and colleagues (2021). Then, as a setting that is on by default, the participant chooses which of 4 sentences says what happened. Three practice animations, one of each kind, come first with feedback; the test has none. The session's seed orders the animations, with no more than 2 of one kind in a row, and the sentences of each question. The triangles are drawn on every screen frame from the time since the animation's first frame, and dropped frames are saved. About 8 minutes.
The main score is the percent of mental-state animations classed as reacting to thoughts or feelings. The test also reports the same for goal-directed and random animations, mental state minus goal-directed (the two kinds closest in movement), how often random and goal-directed animations were classed as reacting to thoughts or feelings, and the percent of right sentences for each kind. The animations are new, so accuracies published for the original set do not apply. Scores support research and do not diagnose.
Scores
- Mental State Accuracy%
- Percent of mental-state animations classed as reacting to thoughts or feelings.
- Higher is better
- Goal-Directed Accuracy%
- Percent of goal-directed animations classed as reacting to actions.
- Higher is better
- Random Accuracy%
- Percent of random animations classed as moving on their own.
- Higher is better
- Overall Accuracy%
- Percent of all animations classed as the right kind.
- Higher is better
- Mental State Minus Goal-Directed%
- Mental state accuracy minus goal-directed accuracy, the two kinds closest in movement.
- Below zero means the mental-state animations were harder
- Over-Attribution%
- Percent of random and goal-directed animations classed as reacting to thoughts or feelings.
- Informational; higher means more mental states read into movement
- Mental State Sentences%
- Percent of mental-state animations whose own sentence was chosen. Empty when the question is off.
- Higher is better
- Goal-Directed Sentences%
- Percent of goal-directed animations whose own sentence was chosen.
- Higher is better
- Random Sentences%
- Percent of random animations whose own sentence was chosen.
- Higher is better
- Overall Sentences%
- Percent of all animations whose own sentence was chosen.
- Higher is better
- Median Time to Choose the Kindms
- Median time from an animation's start to a right choice of kind, leaving out choices under 100 ms.
- Informational; there is no time limit
- Chosen During the Animation%
- Percent of kind choices made while the animation played.
- Informational
- Dropped Frames%
- Screen frames the browser dropped, as a percent of the frames the animations needed.
- Over 10% is flagged for review
- Anticipationscount
- Kind choices under 100 ms.
- Lower supports interpretation
- Animations Answereditems
- Animations with a choice of kind in the scored run.
- Fewer than planned is flagged for review
- Animations Planneditems
- Animations the run set out to show.
- Informational
Use and evidence
60 papers from 2021 to 2025 name this task in their title or abstract, up 30% compared with 2016 to 2020. That ranks it 240 of 395 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 12 |
| 2024 | 11 |
| 2023 | 16 |
| 2022 | 9 |
| 2021 | 12 |
| 2020 | 12 |
| 2019 | 6 |
| 2018 | 6 |
| 2017 | 14 |
| 2016 | 8 |
| 2015 | 4 |
| 2014 | 5 |
| 2013 | 3 |
| 2012 | 5 |
| 2011 | 5 |
| 2010 | 7 |
| 2009 | 3 |
| 2008 | 2 |
| 2007 | 1 |
| 2006 | 4 |
| 2005 | 0 |
| 2004 | 1 |
| 2003 | 1 |
| 2002 | 0 |
| 2001 | 0 |
| 2000 | 2 |
Use grade D
30 to 99 papers from 2021 to 2025.
Evidence grade C
At least 2 test-retest papers, 30 reliability or validity papers or 3 registered clinical studies.
- Papers on test-retest reliability
- 3
- Papers on reliability or validity
- 11
- Registered clinical studies using it as an outcome
- 4
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 0.14.0, counted 2 November 2026 from OpenAlex and ClinicalTrials.gov.
Task design
The trial sequence as the test runs in ConductCognition.
- Animations
- 12 animations written and drawn by our code, 4 of each kind: random, in which the triangles move like objects; goal-directed, in which one reacts to what the other does; and mental state, in which one acts on what the other thinks or feels. Each lasts 18 seconds.
- Stage
- A big red and a small blue triangle on a white stage 4 by 3 in shape, scaled to the screen. A house with an open door stands on the stage in 3 animations of each kind, so the house never gives the kind away.
- Kind
- While an animation plays, the participant chooses its kind: on their own, reacting to actions, or reacting to thoughts or feelings. The first choice counts and the animation always plays to the end. There is no time limit.
- What happened
- Then the participant chooses which of 4 sentences says what happened: the animation's own, one more of its kind and one of each other kind. On by default; it can be turned off.
- Order
- The session's seed orders the animations, with no more than 2 of one kind in a row, and the sentences of each question.
- Practice
- 3 animations, one of each kind, each followed by feedback on both answers. The test has no feedback.
- Frames
- The triangles are drawn on every screen frame from the time since the animation's first frame, so a late frame never slows the motion. Dropped frames are saved, and more than 10% is flagged.
The animations were written for this test and none of the published ones is used, so accuracies reported for the original set do not apply. Answers are multiple choice in place of coded spoken descriptions, after White and colleagues (2011).
Quality checks
Each session is marked valid, needs review or invalid, with the reason, before you see its scores.
Scored rows: animated_shapes_type from the scored run, one of three kinds for each animation (chance 1 in 3, four animations of each kind), with no time limit. Below chance invalidates and not above chance is flagged. One kind on 90% of answers invalidates when accuracy is not above chance. Answers under 200 ms on at least half the animations invalidate. Fewer animations answered than planned, or more than 10% of the animation frames dropped, is flagged and withholds the percentile. A hidden tab withholds 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. The kinds are told apart by how the triangles move, so an animation that stuttered may not have shown the movement that defines it; the 10% limit is a ConductScience choice. The animations keep their own clock, so a hidden tab skips part of one, and the norms come from complete sessions.
Running this test
Settings you can change
| Setting | Standard | Allowed |
|---|---|---|
| Ask what happened | On | On or off |
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 Animated Shapes Task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Heider F, Simmel M, 1944; Abell F, Happé F, Frith U, 2000; Castelli F, Happé F, Frith U, Frith C, 2000; White SJ, Coniston D, Rogers R, Frith U, 2011), with the standard settings (ask what happened: On). The main score was mental state accuracy (percent of mental-state animations classed as reacting to thoughts or feelings).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Heider, F., & Simmel, M. (1944). An experimental study of apparent behavior. The American Journal of Psychology, 57(2), 243-259. https://doi.org/10.2307/1416950Moving shapes described as people with goals and feelings.
- Abell, F., Happé, F., & Frith, U. (2000). Do triangles play tricks? Attribution of mental states to animated shapes in normal and abnormal development. Cognitive Development, 15(1), 1-16. https://doi.org/10.1016/S0885-2014(00)00014-9Random, goal-directed and mental-state animations of a big and a small triangle.
- Castelli, F., Happé, F., Frith, U., & Frith, C. (2000). Movement and mind: A functional imaging study of perception and interpretation of complex intentional movement patterns. NeuroImage, 12(3), 314-325. https://doi.org/10.1006/nimg.2000.0612The three kinds of animation, four of each.
- White, S. J., Coniston, D., Rogers, R., & Frith, U. (2011). Developing the Frith-Happé animations: A quick and objective test of theory of mind for adults with autism. Autism Research, 4(2), 149-154. https://doi.org/10.1002/aur.174Multiple-choice answers: no interaction, physical interaction or mental interaction.
- Livingston, L. A., Shah, P., White, S. J., & Happé, F. (2021). Further developing the Frith-Happé animations: A quicker, more objective, and web-based test of theory of mind for autistic and neurotypical adults. Autism Research, 14(9), 1905-1912. https://doi.org/10.1002/aur.2575The online version: the kind chosen while the animation plays, the first choice counted, three practice animations with feedback.
- Zwickel, J., White, S. J., Coniston, D., Senju, A., & Frith, U. (2011). Exploring the building blocks of social cognition: Spontaneous agency perception and visual perspective taking in autism. Social Cognitive and Affective Neuroscience, 6(5), 564-571. https://doi.org/10.1093/scan/nsq088Animations shortened to about 18 seconds.
Source papers
- Livingston LA, Shah P, White SJ, Happé F (2021). Further developing the Frith-Happé animations: A quicker, more objective, and web-based test of theory of mind for autistic and neurotypical adults. Autism Research, 14(9):1905-1912.Paradigm source, Adults
Questions
How long does the Animated Shapes Task take?
About 8 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?
Mental State Accuracy, Goal-Directed Accuracy, Random Accuracy, Overall Accuracy, Mental State Minus Goal-Directed, Over-Attribution, Mental State Sentences, Goal-Directed Sentences, Random Sentences, Overall Sentences, Median Time to Choose the Kind, Chosen During the Animation, Dropped Frames, Anticipations, Animations Answered and Animations Planned. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Animated Shapes Task used in research?
60 papers from 2021 to 2025 name it in their title or abstract, up 30% compared with 2016 to 2020. That ranks it 240 of 395 test families in the ConductCognition test census (use grade D). Source: ConductCognition test census 0.14.0, counted 2 November 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Animated Shapes Task have?
In the published research, 3 papers report its test-retest reliability and 4 registered clinical studies use it as an outcome (evidence grade C in the ConductCognition test census).
Which settings can a researcher change?
Ask what happened (On or off). 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.