Reversal Learning
Overview
Reversal learning assesses cognitive flexibility by training an animal on a stimulus-response contingency and then reversing the reward contingencies. For example, if the animal has learned that lever A is reinforced and lever B is not (discrimination phase), the contingencies are switched so that lever B is now reinforced and lever A is extinguished. The number of errors and trials to reach criterion after reversal measures the ability to suppress a previously learned response and acquire a new one.
Reversal learning depends critically on orbitofrontal cortex (OFC) and serotonergic neurotransmission. OFC lesions impair reversal but spare initial discrimination, making this paradigm a selective assay for cognitive flexibility. Serial reversals can be used to assess learning-to-learn effects and are impaired in models of autism, schizophrenia, and addiction.
ConductMaze automates both the discrimination and reversal phases, applying criterion-based advancement rules to switch contingencies automatically. The software tracks perseverative errors (continued responding on the previously correct option), regressive errors, and learning curves across multiple serial reversals.
Trial Flow
Discrimination Phase
Train stimulus-response contingency to criterion
Criterion Check
Has subject reached discrimination criterion?
Contingency Reversal
Reward contingencies switched between options
Trial Presentation
Subject chooses between options under new contingency
Outcome Delivery
Reinforcer for correct choice, no reward for incorrect
Reversal Criterion
Has subject reached reversal criterion?
Phase Complete
Log errors, advance to next reversal or end
Parameters
| Parameter | Type | Default | Description |
|---|---|---|---|
| Criterion | integer | 8 | Consecutive correct trials to meet criterion (e.g., 8/10 correct) |
| Criterion Window | integer | 10 | Sliding window for criterion calculation |
| Number of Reversals | integer | 1 | Total reversal phases (1 = single reversal, >1 = serial reversal) |
| Max Trials per Phase | integer | 200 | Maximum trials before forced phase advancement |
| ITI Duration | seconds | 5 | Inter-trial interval between choice trials |
| Correction Procedure | boolean | false | Repeat incorrect trials until correct (reduces side bias) |
| Stimulus Type | enum | Spatial | Discrimination dimension (spatial, visual, auditory) |
Metrics
| Metric | Unit | Description |
|---|---|---|
| Trials to Criterion | count | Total trials to reach criterion after reversal |
| Perseverative Errors | count | Errors on the previously correct option (first errors after reversal) |
| Regressive Errors | count | Errors occurring after initial shift away from old rule |
| Total Errors | count | All incorrect choices during reversal phase |
| Choice Latency | seconds | Mean time from trial start to choice response |
| Win-Stay Probability | proportion | Probability of repeating a rewarded choice |
| Lose-Shift Probability | proportion | Probability of switching after an unrewarded choice |
Sample Data
| Phase | Trials_to_Criterion | Perseverative_Errors | Regressive_Errors | Total_Errors | Mean_Latency_s |
|---|
Representative data for illustration purposes. Actual values will vary by species, strain, and experimental conditions.
Applications
- 1Cognitive flexibility — selective OFC-dependent measure dissociated from initial learning
- 2Psychiatric modeling — reversal impairments in autism, schizophrenia, and OCD models
- 3Addiction research — drug-induced perseveration as a model of compulsive behavior
- 4Serotonin pharmacology — 5-HT depletion selectively impairs reversal learning
- 5Developmental neuroscience — tracking maturation of flexible behavior in juveniles
Related Protocols
Compatible Products
Ready to Automate Your Behavioral Protocols?
Contact us for a demo and pricing information.
