Reward and Decision Making
Loss-Prevention Learning Task
Also called: avoidance learning task, costly avoidance, instrumental avoidance learning
Measure how people learn to prevent point losses and adapt when the rules change. Separate useful protection from protection that costs more than the expected loss.
5-8 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
- 40
- Evidence grade
- C
- Registered clinical studies
- 3
What it measures
Measures learning when an action prevents a point loss. Participants learn which shapes often lead to losses, pay a smaller cost to prevent a loss or leave the points unprotected, and adapt when the shapes' probabilities switch.
Constructs
- Instrumental avoidance learning
- Costly protection
- Learning after a change in loss probability
Research uses
- Research on avoidance learning
- Studies of costly protective choices and behavioral flexibility
How the paradigm works
The Loss-Prevention Learning Task measures learning when an action prevents a loss at a smaller cost. A shape appears, and the participant chooses to protect their points or leave them unprotected. Protection always costs 3 point units and prevents a possible 10-unit loss. One shape predicts a loss more often than the other; feedback allows the participant to learn which is which.
More about the paradigm
The standard session has 120 choices in six blocks of 20. Each block presents both shapes ten times. Unprotected losses are scheduled on eight of ten appearances for one shape and two of ten for the other. Their roles switch once, after 60 choices, without a special announcement. A saved seed sets the initial cue roles, each block order, loss schedule and the protection button side. Four practice choices use a separate shape and balance. Protection never reveals whether the unchosen outcome would have lost points.
The task adapts the costly-avoidance method of Norbury and colleagues using original geometric cues and a fixed reversal. It omits perceptual generalization probes and uses points with no unpleasant images, sounds or stimulation. Choices have no deadline. The block count and point multiplier are bounded settings.
Scores
- Lower-Loss Choices%
- Proportion choosing protection for the 80% cue and no protection for the 20% cue, reconstructed from the seed.
- Higher means more choices of the action with lower expected point loss
- Protection at High Loss Probability%
- Protection rate when an unprotected choice has an 80% loss probability.
- Informational
- Protection at Low Loss Probability%
- Protection rate when an unprotected choice has a 20% loss probability.
- Informational
- Protection Difference%
- Protection at high loss probability minus protection at low loss probability.
- Positive means greater protection for the more loss-prone cue
- Acquisition Choices%
- All choices before the switch: proportion choosing the action with lower expected loss.
- Informational
- Reversal Choices%
- All choices after the switch: proportion choosing the action with lower expected loss.
- Informational
- Late Acquisition Choices%
- The last 20 choices before the switch: proportion choosing the action with lower expected loss.
- Informational
- Early Reversal Choices%
- The first 20 choices after the switch: proportion choosing the action with lower expected loss.
- Informational
- Late Reversal Choices%
- The last 20 choices of the task: proportion choosing the action with lower expected loss.
- Informational
- Protection Rate%
- Proportion of all readable choices using protection.
- Informational
- Points Spent or Lostpoints
- Protection costs plus losses from unprotected choices.
- Informational
- Points Spent on Protectionpoints
- Number of protected choices times 3 times the point multiplier.
- Informational
- Unprotected Point Lossespoints
- Loss outcomes after unprotected choices times 10 times the point multiplier.
- Informational
- Final Pointspoints
- Starting balance minus protection costs and unprotected losses, reconstructed from the seed.
- Informational
- Median Choice Timems
- Median readable choice time at or above 100 ms.
- Informational
- Anticipationscount
- Presses rejected under 100 ms, plus any saved responses under 100 ms.
- Quality check
- Choices Answeredcount
- Unique readable choices in the selected run.
- Informational
- Planned Choicescount
- 40 times the blocks-per-phase setting.
- Informational
- Complete Run0/1
- 1 when all planned choices are present with no unreadable rows; otherwise 0.
- Quality check
- Unreadable Choicescount
- Rows with duplicate or invalid sequence numbers, responses or settings.
- Quality check
Use and evidence
40 papers from 2021 to 2025 name this task in their title or abstract, up 48% compared with 2016 to 2020. That ranks it 392 of 582 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 15 |
| 2024 | 4 |
| 2023 | 8 |
| 2022 | 7 |
| 2021 | 6 |
| 2020 | 3 |
| 2019 | 4 |
| 2018 | 10 |
| 2017 | 6 |
| 2016 | 4 |
| 2015 | 5 |
| 2014 | 5 |
| 2013 | 2 |
| 2012 | 4 |
| 2011 | 4 |
| 2010 | 11 |
| 2009 | 2 |
| 2008 | 10 |
| 2007 | 6 |
| 2006 | 7 |
| 2005 | 6 |
| 2004 | 5 |
| 2003 | 9 |
| 2002 | 5 |
| 2001 | 7 |
| 2000 | 8 |
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
- 0
- Papers on reliability or validity
- 5
- Registered clinical studies using it as an outcome
- 3
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.
Note on the counts: Points-only costly avoidance with a fixed probability reversal; no aversive media or generalization probes..
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 0.39.0, counted 8 November 2026 from OpenAlex and ClinicalTrials.gov.
Task design
The trial sequence as the test runs in ConductCognition.
- Choices
- 120 choices by default in six blocks of 20; each block shows each of two shapes ten times. Two to five blocks per phase give 80 to 200 choices.
- Loss probabilities
- 80% and 20%: exactly eight and two scheduled unprotected losses per ten appearances of each cue in a block. Cue roles switch once at the midpoint, after 60 choices by default.
- Points
- Protection always costs 3 points and prevents the possible 10-point loss. A multiplier from 1 to 10 scales all amounts together. Starting points cover a loss on every choice: 1,200 by default.
- Feedback
- Only the chosen outcome is shown. Protected choices show the protection cost, never the unchosen outcome. Outcomes stay visible for 1,000 ms, followed by a 250 ms gap.
- Practice
- Four choices with a separate shape and points. A break follows every 20 test choices; the midpoint break does not announce the switch.
- Scores
- Lower expected loss choice rate is primary. Also reports protection at high and low probabilities, their difference, acquisition and reversal choices, protection costs, losses and final points. Percentiles are off.
- Devices and timing
- Untimed choices by click, tap or F/J keys on desktop, tablet or phone. Minimum touch screen 360 by 640. Estimated duration 5 to 8 minutes.
Our fixed-reversal version uses original geometric cues and points. It adapts costly avoidance without perceptual generalization probes, aversive media or a response deadline. Scores describe learning and protective choices in this task.
Quality checks
Each session is marked valid, needs review or invalid, with the reason, before you see its scores.
Scores protection choices from the most complete run. Low learning and repeated protection are outcomes of interest, so they do not invalidate. Answers under 200 ms on at least half the choices invalidate. Incomplete runs, unreadable rows and lower-loss choices on no more than half of trials are flagged for review.
Why these rules
Costly avoidance can persist even when its expected cost exceeds the unprotected loss. Discarding repeated protection would remove the behavior being measured. Each cue must still be seen before choosing.
No percentile. Scores and quality flags only.
Running this test
Settings you can change
| Setting | Standard | Allowed |
|---|---|---|
| Blocks before and after the switch | 3 | 2 to 5 |
| Point multiplier | 1 | 1 to 10 |
| 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
- 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.
- 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 Loss-Prevention Learning Task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Norbury, Robbins and Seymour, 2018), with the standard settings (blocks before and after the switch: 3; point multiplier: 1; payment line: Hypothetical points). The main score was lower-loss choices (proportion choosing protection for the 80% cue and no protection for the 20% cue, reconstructed from the seed).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Norbury, A., Robbins, T. W., & Seymour, B. (2018). Value generalization in human avoidance learning. eLife, 7, e34779. https://doi.org/10.7554/eLife.34779Full methods read, including the online monetary-loss version. Costly avoidance is adapted here to points and a fixed reversal.
Source papers
Questions
How long does the Loss-Prevention Learning Task take?
About 5 to 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?
Lower-Loss Choices, Protection at High Loss Probability, Protection at Low Loss Probability, Protection Difference, Acquisition Choices, Reversal Choices, Late Acquisition Choices, Early Reversal Choices, Late Reversal Choices, Protection Rate, Points Spent or Lost, Points Spent on Protection, Unprotected Point Losses, Final Points, Median Choice Time, Anticipations, Choices Answered, Planned Choices, Complete Run and Unreadable Choices. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Loss-Prevention Learning Task used in research?
40 papers from 2021 to 2025 name it in their title or abstract, up 48% compared with 2016 to 2020. That ranks it 392 of 582 test families in the ConductCognition test census (use grade D). Source: ConductCognition test census 0.39.0, counted 8 November 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Loss-Prevention Learning Task have?
In the published research, 0 papers report its test-retest reliability and 3 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 before and after the switch (2 to 5), point multiplier (1 to 10) 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.