Decision Making
Probabilistic Reward Task
Also called: PRT, reward responsiveness task, signal detection reward task, Pizzagalli task
Measure how rewards shift choices: participants judge a 100 ms short or long mouth while one answer pays three times as often, scored as response bias.
13-16 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. Preview any test free after you sign up.
- Papers, 2021 to 2025
- 84
- Evidence grade
- C
- Registered clinical studies
- 38
What it measures
Measures how rewards shift choices. A cartoon face shows a short or a slightly longer mouth for 100 ms, and participants press a key for which they saw. Correct answers about one mouth win points three times as often as about the other, over blocks of 100 faces. The main score is the response bias toward the more rewarded mouth.
Constructs
- Reward responsiveness
- Reward learning
- Hedonic capacity
- Perceptual discrimination
Research uses
- Anhedonia and depression research
- Reward processing in clinical trials
- Reward learning across development and aging
- Within-study group comparison
How the paradigm works
The probabilistic reward task measures how rewards shape choices (Pizzagalli, Jahn and O'Shea, 2005). On each trial a fixation cross shows for 500 ms, then a simple face without a mouth for 500 ms, then the face with a short or a slightly longer mouth for 100 ms. The face then stays without a mouth until the participant presses one key for short and another for long. The face was drawn for this test. Its mouths are 43 and 49 CSS pixels long, about 11.4 and 13.0 mm at the CSS reference of 96 pixels per inch, so the long one is 14% longer, close to the source's 11.5 and 13 mm. The mouth switches on screen frames, and its measured duration is saved on every trial.
More about the paradigm
Each block has 100 faces, 50 short and 50 long, with no more than three of one kind in a row. One mouth, the rich one, is rewarded three times as often as the other: in each block 30 rewards are scheduled on correct answers to the rich mouth and 10 on correct answers to the lean one. A reward scheduled on a wrong answer waits for the next correct answer to the same mouth, as in the source, so every participant gets close to the same 3 to 1 ratio. A reward shows "Correct! You won 5 points" for 1.75 seconds, and other answers show nothing. Participants learn that some correct answers win points and not every one does, but not that one mouth pays more. Which mouth is rich and which key means short come from the session's seed. Six practice faces with feedback come first. The standard settings run three blocks, as the source did; a study can run two, set the points per reward and choose the payment line.
The main score is the response bias, log b: half the base-10 log of rich correct times lean incorrect, over rich incorrect times lean correct, with 0.5 added to each count. Positive values mean the answers lean toward the rich mouth. Discriminability, log d, shows how well the two mouths were told apart, and the change in bias from the first block to the last shows reward learning over the task. Trials answered less than 150 ms or more than 2,500 ms after the mouth's onset, and trials whose mouth ran more than one frame off 100 ms, are left out of these scores. A smaller bias has gone with more anhedonic symptoms (Pizzagalli, Jahn and O'Shea, 2005) and with major depression (Pizzagalli et al., 2008). Scores support research and do not diagnose.
Scores
- Response Bias (log b)score
- Half the base-10 log of rich correct times lean incorrect, over rich incorrect times lean correct, with 0.5 added to each count, over all blocks.
- Higher means the answers lean more toward the rich mouth
- Discriminability (log d)score
- Half the base-10 log of rich correct times lean correct, over rich incorrect times lean incorrect, with 0.5 added to each count, over all blocks.
- Higher means the mouths were told apart better
- Change in Biasscore
- The last block's response bias minus the first block's.
- Positive values mean the bias grew over the task
- Response Bias, Block 1score
- Response bias in block 1; empty for a block not run.
- Informational
- Response Bias, Block 2score
- Response bias in block 2; empty for a block not run.
- Informational
- Response Bias, Block 3score
- Response bias in block 3; empty for a block not run.
- Informational
- Discriminability, Block 1score
- Discriminability in block 1; empty for a block not run.
- Informational
- Discriminability, Block 2score
- Discriminability in block 2; empty for a block not run.
- Informational
- Discriminability, Block 3score
- Discriminability in block 3; empty for a block not run.
- Informational
- Rich Mouth Accuracy%
- Proportion of usable trials with the rich mouth answered correctly.
- Informational
- Lean Mouth Accuracy%
- Proportion of usable trials with the lean mouth answered correctly.
- Informational
- Rich Mouth Response Timems
- Mean correct response time from the mouth's onset, rich mouth, usable trials.
- Informational
- Lean Mouth Response Timems
- Mean correct response time from the mouth's onset, lean mouth, usable trials.
- Informational
- Accuracy%
- Proportion of all test trials answered correctly.
- Higher means more correct
- Rewards for the Rich Mouthcount
- Rewards given for correct answers to the rich mouth.
- 30 per block when every scheduled reward was given
- Rewards for the Lean Mouthcount
- Rewards given for correct answers to the lean mouth.
- 10 per block when every scheduled reward was given
- Points Wonpoints
- All rewards times the points per reward.
- Informational
- Usable Trialstrials
- Test trials answered 150 to 2,500 ms after the mouth's onset, with the mouth on screen within one frame of 100 ms.
- A block with fewer than 80 is flagged for review
- Trials with Mouth Timing Offtrials
- Test trials whose mouth ran more than one frame off 100 ms, or was not measured.
- Over 20% of trials is flagged for review
- Anticipationscount
- Answers within 100 ms of the mouth leaving the screen.
- Above the timed-test limit invalidates
Use and evidence
84 papers from 2021 to 2025 name this task in their title or abstract, up 105% compared with 2016 to 2020. That ranks it 201 of 395 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 24 |
| 2024 | 25 |
| 2023 | 18 |
| 2022 | 4 |
| 2021 | 13 |
| 2020 | 17 |
| 2019 | 10 |
| 2018 | 6 |
| 2017 | 6 |
| 2016 | 2 |
| 2015 | 7 |
| 2014 | 7 |
| 2013 | 5 |
| 2012 | 3 |
| 2011 | 4 |
| 2010 | 1 |
| 2009 | 2 |
| 2008 | 4 |
| 2007 | 1 |
| 2006 | 0 |
| 2005 | 0 |
| 2004 | 0 |
| 2003 | 0 |
| 2002 | 0 |
| 2001 | 0 |
| 2000 | 0 |
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
- 2
- Papers on reliability or validity
- 16
- Registered clinical studies using it as an outcome
- 38
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 31 October 2026 from OpenAlex and ClinicalTrials.gov.
Task design
The trial sequence as the test runs in ConductCognition.
- Trial
- A fixation cross for 500 ms, a face without a mouth for 500 ms, the face with a short or long mouth for 100 ms, then the face without a mouth until the answer. No time limit.
- Face
- A simple drawing made for this test, with mouths of 43 and 49 pixels: about 11.4 and 13.0 mm at the CSS reference of 96 pixels per inch.
- Mouth timing
- The face and the mouth switch on screen frames, and the mouth's measured duration is saved on every trial. A trial whose mouth ran more than one frame off is left out of the scores.
- Blocks
- 3 blocks (standard; 2 to 3) of 100 faces, 50 short and 50 long, with no more than three of one kind in a row.
- Rewards
- In each block, 30 rewards on correct answers to the rich mouth and 10 on correct answers to the lean one. A reward scheduled on a wrong answer waits for the next correct answer to the same mouth.
- Feedback
- "Correct! You won 5 points" for 1,750 ms after a reward (5 points standard; 1 to 100), and nothing after other answers.
- Counterbalancing
- Which mouth is rich and which of F and J means short come from the session's seed.
- Practice
- 6 faces with feedback and no points.
- Response times
- Trials answered from 150 to 2,500 ms after the mouth's onset are scored.
Pizzagalli and colleagues (2005) used a cartoon face with mouths of 11.5 and 13 mm and paid 5 cents per reward. This test draws its own face and pays points, which a study can tie to money with the payment line. With two blocks the task is shorter, and the change in bias compares block 2 with block 1. 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: probabilistic_reward_test from the scored run, F or J for a short or long mouth (chance 50%, half the mouths are short), correctness from the key for the mouth. The face stays until the answer, so no response is missed. Below chance invalidates and not above chance is flagged. One key on 90% of answers invalidates when accuracy is not above chance. Responses under 100 ms follow the timed-test rule. Completing fewer than half the planned trials invalidates, and any other incomplete run is flagged. A block with fewer than 80 usable trials, a ratio of rich to lean rewards under 2.5, or the mouth more than one frame off on over 20% of trials is flagged for review.
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). Missed responses and responses under 100 ms keep the platform's timed-test rule: more than max(3, 25%) missed responses or more than max(2, 20%) responses under 100 ms invalidate, and any is flagged. A response under 100 ms cannot be a reaction to the stimulus. The bias is a difference between answers to the two mouths, so a block needs most of its trials, and it builds from the 3 to 1 ratio of rewards, so a session given a smaller ratio tested something weaker. The mouth's 100 ms sets how hard the two are to tell apart, so a display that could not hold it changes the measure.
No percentile. Scores and quality flags only.
Running this test
Settings you can change
| Setting | Standard | Allowed |
|---|---|---|
| Blocks | 3 | 2 to 3 |
| Points per reward | 5 | 1 to 100 |
| Payment line | Hypothetical points (standard) | Hypothetical points (standard), The study pays for points |
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 Probabilistic Reward Task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Pizzagalli DA, Jahn AL, O'Shea JP, 2005; Pizzagalli DA et al., 2008), with the standard settings (blocks: 3; points per reward: 5; payment line: Hypothetical points). The main score was response bias (log b; half the base-10 log of rich correct times lean incorrect, over rich incorrect times lean correct, with 0.5 added to each count, over all blocks).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Pizzagalli, D. A., Jahn, A. L., & O'Shea, J. P. (2005). Toward an objective characterization of an anhedonic phenotype: A signal-detection approach. Biological Psychiatry, 57(4), 319-327. https://doi.org/10.1016/j.biopsych.2004.11.026
- Pizzagalli, D. A., Iosifescu, D., Hallett, L. A., Ratner, K. G., & Fava, M. (2008). Reduced hedonic capacity in major depressive disorder: Evidence from a probabilistic reward task. Journal of Psychiatric Research, 43(1), 76-87. https://doi.org/10.1016/j.jpsychires.2008.03.001The reward schedule and the formulas for bias and discriminability.
- Hautus, M. J. (1995). Corrections for extreme proportions and their biasing effects on estimated values of d'. Behavior Research Methods, Instruments, & Computers, 27(1), 46-51. https://doi.org/10.3758/BF03203619Adding 0.5 to each count.
Source papers
- Pizzagalli, Jahn and O'Shea (2005): the task, its bias and its link with anhedonia. Pizzagalli et al. (2008): reduced reward responsiveness in major depression.Paradigm source, Adults
Questions
How long does the Probabilistic Reward Task take?
About 13 to 16 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?
Response Bias (log b), Discriminability (log d), Change in Bias, Response Bias, Block 1, Response Bias, Block 2, Response Bias, Block 3, Discriminability, Block 1, Discriminability, Block 2, Discriminability, Block 3, Rich Mouth Accuracy, Lean Mouth Accuracy, Rich Mouth Response Time, Lean Mouth Response Time, Accuracy, Rewards for the Rich Mouth, Rewards for the Lean Mouth, Points Won, Usable Trials, Trials with Mouth Timing Off and Anticipations. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Probabilistic Reward Task used in research?
84 papers from 2021 to 2025 name it in their title or abstract, up 105% compared with 2016 to 2020. That ranks it 201 of 395 test families in the ConductCognition test census (use grade D). Source: ConductCognition test census 0.14.0, counted 31 October 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Probabilistic Reward Task have?
In the published research, 2 papers report its test-retest reliability and 38 registered clinical studies use it as an outcome (evidence grade C in the ConductCognition test census).
Does it report a percentile?
No. Scores and quality flags only.
Which settings can a researcher change?
Blocks (2 to 3), points per reward (1 to 100) and payment line (Hypothetical points (standard), The study pays for points). 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.