The critical signal is the mismatch between expected and received feedback. When a formerly rewarded choice stops producing reward, that discrepancy indicates that the existing prediction no longer fits the current contingency. The participant must use the error signal to revise the expected outcome and guide the next choice, making reversal performance an observable index of error-based updating.
Success requires more than noticing that feedback changed. The participant must disengage from a response that was previously reinforced, avoid repeating it as a habit, and select the alternative under the new contingency. Performance therefore reflects the ability to adapt behavior when learned relationships no longer remain reliable, providing a behavioral measure of flexible decision-making.
By comparing behavioral adaptation during changed contingencies with activity in brain systems involved in decision-making and reward, researchers can examine the neural basis of updating. This design links observable choices and feedback-driven learning to activity in relevant systems, rather than treating performance as an isolated behavioral outcome. It therefore supports investigation of neural mechanisms underlying flexible behavior.
First, an organism learns a stimulus-response or stimulus-outcome relationship under an established contingency. Researchers then alter that relationship so the formerly correct choice is no longer rewarded. Subsequent choices and responses to feedback reveal whether the participant updates expectations, suppresses the prior response, and adopts the alternative. This sequence tests adaptation after the contingency changes.
These trials provide information about how effectively an organism adapts after feedback changes. Performance can be interpreted in terms of cognitive flexibility, reinforcement learning, and error-based updating, including the ability to revise a prediction and change the selected response. The resulting behavioral pattern can then be related to activity in brain systems involved in decision-making and reward.
They support comparisons across development, disease-related impairments, and treatment effects because each context may alter adaptation to changed contingencies. Researchers can use performance differences to investigate whether flexible updating and reward-guided decision-making are affected, and whether a treatment is associated with changes in these learning-related behaviors. This makes the approach relevant to both basic and applied neuroscience.