Processing Speed
Agency Judgments Task
Also called: judgments of agency, judgment of agency, explicit sense of agency, sense of agency task, agency ratings, feeling of control, metacognition of agency
Measure how much people feel in control of what their actions do: participants press a key to light a circle and steer a box to catch falling letters while the screen adds delays or wobble, and rate their control after each trial.
7-9 minTimed: desktop or laptop with a mouse or trackpad
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
- 76
- Evidence grade
- D
- Registered clinical studies
- 1
What it measures
Measures how much people feel in control of what their actions make happen on the screen. Participants press a key to light a circle and steer a box to catch falling letters while the screen sometimes adds delays or wobble, and rate their control after each trial. The main score is how much lower the ratings are when something is added.
Constructs
- Sense of agency
- Judgments of agency
- Action monitoring
Research uses
- Agency and self research
- Human factors and interface research
- Metacognition research
How the paradigm works
Judgments of agency are people's ratings of how much they controlled an action and what it made happen. Metcalfe and Greene (2007) had people steer a box with the mouse to touch falling Xs and avoid Os and rate their control after each trial; people knew when the cursor's movements were disturbed. Metcalfe, Eich and Castel (2010) drew the cursor where the mouse had been 250 or 500 ms earlier, or added random noise to its position, and both lowered the judgments. Sato and Yasuda (2005) found that a self-produced effect at an unpredicted time lowers the sense of self-agency, and Wen, Yamashita and Asama (2015) found that the longer the delay added to keys that steer a dot, the lower people rated their control on a 9-point scale.
More about the paradigm
The task has two parts, in an order drawn from the session's seed; a setting runs one part alone. In the key part, each press of the space bar lights a circle after an added delay of 0, 200, 400 or 600 ms, 4 trials at each delay (2 to 8 as a setting), each block holding every delay once in a seeded order, after 2 practice trials with no delay. In the mouse part, the participant steers a box on a track for 10 seconds to catch falling Xs and let Os pass, 8 of each per round, while the box is drawn with nothing added, where the mouse was 250 or 500 ms earlier, or with a smooth random wobble of 6% or 12% of the track's width; 3 rounds of each (2 to 5 as a setting), each block holding every condition once in a seeded order, after 1 practice round with nothing added. The system pointer is hidden during a round, and a round starts again from the beginning if the pointer leaves the window.
After every trial the participant rates how much they felt in control, from 1 (not at all) to 9 (completely). The instructions do not mention the added delays and wobble; the end of the session explains them.
The press is timed by its key event, and the circle lights on the first screen frame within half a frame of the planned delay; the delay as shown is saved. Every pointer move is kept with its time, and each frame draws the box from the hand's place the delay before. The primary score is how much lower the ratings are with something added than with nothing added, in each part, averaged over the parts run. The ratings by delay, the drops with the box behind the mouse and with the wobble, the share of Xs caught and the correlation of the ratings with it come with it.
Scores
- Control rating droppoints
- In each part, the mean control rating with nothing added minus the mean rating with an added delay or wobble, averaged over the parts run.
- Above 0 when the added delays and wobble lowered the felt control; descriptive, no better direction
- Key part, no delaypoints
- Mean control rating of the key trials with no added delay.
- Descriptive; 1 to 9
- Key part, longest delaypoints
- Mean control rating of the key trials at the longest added delay, 600 ms.
- Descriptive; 1 to 9
- Drop with the delaypoints
- The key part's mean rating with no added delay minus that at the longest delay.
- Above 0 when the delay lowered the felt control
- Game, nothing addedpoints
- Mean control rating of the game rounds with nothing added.
- Descriptive; 1 to 9
- Drop with the box behindpoints
- The game's mean rating with nothing added minus that of the rounds with the box 250 or 500 ms behind the mouse.
- Above 0 when the lag lowered the felt control
- Drop with the wobblepoints
- The game's mean rating with nothing added minus that of the rounds with the wobble.
- Above 0 when the wobble lowered the felt control
- Xs caught, nothing added%
- Share of the Xs caught in the game rounds with nothing added.
- Higher is better play
- Xs caught, something added%
- Share of the Xs caught in the game rounds with the box behind the mouse or wobbling.
- Higher is better play
- Rating and playscore
- Pearson correlation of the control rating with the share of Xs caught over the game rounds, with at least 5 rounds.
- Above 0 when the rounds rated in control were the rounds played well
- Circle timingms
- Median of the delay at which the circle lit minus the planned delay, on the screen's frames.
- More than 20 ms either way is flagged for review
- Missed pressescount
- Key trials with no press within 10 seconds, which have no rating.
- Above max(3, 25%) of trials makes the session invalid; any is flagged
- Rounds without steeringcount
- Game rounds with fewer than 10 pointer moves.
- 3 or more is flagged for review
- Restarted roundscount
- Game rounds that started again because the pointer left the window, the tab was hidden or the window changed size.
- Informational
- Trialscount
- Key trials and game rounds in the scored run.
- Informational
Use and evidence
76 papers from 2021 to 2025 name this task in their title or abstract, up 81% compared with 2016 to 2020. That ranks it 342 of 711 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 17 |
| 2024 | 32 |
| 2023 | 4 |
| 2022 | 9 |
| 2021 | 14 |
| 2020 | 8 |
| 2019 | 9 |
| 2018 | 10 |
| 2017 | 4 |
| 2016 | 11 |
| 2015 | 12 |
| 2014 | 8 |
| 2013 | 11 |
| 2012 | 5 |
| 2011 | 8 |
| 2010 | 4 |
| 2009 | 5 |
| 2008 | 6 |
| 2007 | 6 |
| 2006 | 2 |
| 2005 | 1 |
| 2004 | 1 |
| 2003 | 0 |
| 2002 | 0 |
| 2001 | 1 |
| 2000 | 1 |
Use grade D
30 to 99 papers from 2021 to 2025.
Evidence grade D
Little published measurement evidence found.
- Papers on test-retest reliability
- 0
- Papers on reliability or validity
- 13
- Registered clinical studies using it as an outcome
- 1
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 14 November 2026 from OpenAlex and ClinicalTrials.gov.
Task design
The trial sequence as the test runs in ConductCognition.
- Parts
- A key part and a mouse part, in an order drawn from the session's seed (standard), or one of them alone.
- Key part
- Each press of the space bar lights a circle after an added delay of 0, 200, 400 or 600 ms. 4 trials at each delay (standard; 2 to 8): 16 trials, each block holding every delay once in a seeded order. A trial with no press in 10 seconds ends.
- Mouse part
- 10-second rounds in which the participant steers a box on a track with the mouse to catch falling Xs and let Os pass, 8 of each per round. The box is drawn with nothing added, where the mouse was 250 or 500 ms earlier, or with a smooth random wobble of 6% or 12% of the track's width (root mean square). 3 rounds of each (standard; 2 to 5): 15 rounds, each block holding every condition once in a seeded order. The system pointer is hidden, and a round starts again if the pointer leaves the window.
- Rating
- After every trial, how much the participant felt in control, from 1 (not at all) to 9 (completely), on buttons in one row.
- Practice
- 2 key trials with no delay and 1 game round with nothing added, each with its rating.
- After the task
- The instructions do not mention the added delays and wobble; the end of the session explains them.
- Timing
- The press is timed by its key event, and the circle lights on the first screen frame within half a frame of the planned delay; the delay as shown is saved. Every pointer move is kept with its time, and each frame draws the box from the hand's place the delay before.
- Scores
- The main score is, in each part, the mean rating with nothing added minus the mean rating with something added, averaged over the parts run. The key part's ratings with no delay and at 600 ms, the game's drops with the box behind the mouse and with the wobble, the share of Xs caught and the correlation of the ratings with it come with it.
A rating of control has no better or worse value, and this task has no published norm, so the scores describe ratings within this task, with no percentile. The scores 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: agency_judgments_trial from the scored run, one control rating from 1 to 9 after each trial. A rating has no correct answer, so accuracy is not judged against chance. A key trial with no press within 10 seconds has no rating: more than max(3, 25%) such trials invalidate and any is flagged. Ratings given under 200 ms on at least half the trials invalidate. The same rating on every trial, no comparison of trials with and without something added, 3 or more game rounds with the mouse barely moved and a circle lit more than 20 ms from the planned delay (median) are flagged.
Why these rules
There is no correct option to compare with chance, so accuracy is not judged against chance. Missed presses keep the platform's timed-test rule. 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. A person can feel in control on every trial, so the same rating throughout flags review only. A rating after a round in which the mouse barely moved says little about control of the box. The added delays are what is varied, so a display that lit the circle well off the planned delay changes what was measured. The key, the mouse and the display are timed on the page, so a hidden tab can stop the frames; a percentile is withheld after one.
No percentile. Scores and quality flags only.
Running this test
Settings you can change
| Setting | Standard | Allowed |
|---|---|---|
| Parts | Key presses and the mouse game (standard) | Key presses and the mouse game (standard), Key presses only, Mouse game only |
| Key trials per delay | 4 | 2 to 8 |
| Game rounds per condition | 3 | 2 to 5 |
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
- Timed test: participants need a desktop or laptop with a mouse or trackpad. Participant devices
- Languages: English, Spanish, French, German, Italian, Portuguese (Brazil), Chinese (Simplified) and Japanese.
- Available on: Every plan, including the free plan.
- No percentile. Scores and quality flags only. How norms work
- Recruiting online? See the setup guides for Prolific, Sona Systems, CloudResearch Connect and Qualtrics.
Cite and describe
Participants completed the Agency Judgments Task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Metcalfe and Greene, 2007; Sato and Yasuda, 2005), with the standard settings (parts: Key presses and the mouse game; key trials per delay: 4; game rounds per condition: 3). The main score was control rating drop (in each part, the mean control rating with nothing added minus the mean rating with an added delay or wobble, averaged over the parts run).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Metcalfe, J., & Greene, M. J. (2007). Metacognition of agency. Journal of Experimental Psychology: General, 136(2), 184-199. https://doi.org/10.1037/0096-3445.136.2.184The falling-letters game with a disturbed cursor, and a judgment of agency after each trial.
- Metcalfe, J., Eich, T. S., & Castel, A. D. (2010). Metacognition of agency across the lifespan. Cognition, 116(2), 267-282. https://doi.org/10.1016/j.cognition.2010.05.009The cursor drawn 250 or 500 ms behind the mouse, random noise added to it, and the scale from no control to complete control.
- Sato, A., & Yasuda, A. (2005). Illusion of sense of self-agency: Discrepancy between the predicted and actual sensory consequences of actions modulates the sense of self-agency, but not the sense of self-ownership. Cognition, 94(3), 241-255. https://doi.org/10.1016/j.cognition.2004.04.003A self-produced effect at an unpredicted time lowers the sense of self-agency.
- Wen, W., Yamashita, A., & Asama, H. (2015). The sense of agency during continuous action: Performance is more important than action-feedback association. PLOS ONE, 10(4), e0125226. https://doi.org/10.1371/journal.pone.0125226Delays added to the keys that steer a dot, and control rated from 1 (not at all) to 9 (a lot).
- Wegner, D. M., & Wheatley, T. (1999). Apparent mental causation: Sources of the experience of will. American Psychologist, 54(7), 480-492. https://doi.org/10.1037/0003-066X.54.7.480The experience of will arises when a thought comes before an act, fits it and has no rival cause.
- Moore, J. W., & Obhi, S. S. (2012). Intentional binding and the sense of agency: A review. Consciousness and Cognition, 21(1), 546-561. https://doi.org/10.1016/j.concog.2011.12.002The sense of agency beside intentional binding.
Source papers
- Metcalfe, J., Eich, T. S., & Castel, A. D. (2010). Metacognition of agency across the lifespan. Cognition, 116(2), 267-282.Paradigm source, Adults
Questions
How long does the Agency Judgments Task take?
About 7 to 9 minutes, as listed in the test library.
What device does a participant need?
Timed test: participants need a desktop or laptop with a mouse or trackpad. The browser must be Chrome, Edge, Firefox or Safari.
What scores does it produce?
Control rating drop, Key part, no delay, Key part, longest delay, Drop with the delay, Game, nothing added, Drop with the box behind, Drop with the wobble, Xs caught, nothing added, Xs caught, something added, Rating and play, Circle timing, Missed presses, Rounds without steering, Restarted rounds and Trials. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Agency Judgments Task used in research?
76 papers from 2021 to 2025 name it in their title or abstract, up 81% compared with 2016 to 2020. That ranks it 342 of 711 test families in the ConductCognition test census (use grade D). Source: ConductCognition test census 0.58.0, counted 14 November 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Agency Judgments Task have?
In the published research, 0 papers report its test-retest reliability and 1 registered clinical study uses it as an outcome (evidence grade D in the ConductCognition test census).
Does it report a percentile?
No. Scores and quality flags only.
Which settings can a researcher change?
Parts (Key presses and the mouse game (standard), Key presses only, Mouse game only), key trials per delay (2 to 8) and game rounds per condition (2 to 5). 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.