Test-Retest Reliability of Two Computationally-Characterised Affective Bias Tasks

Area of Science:

• Psychology
• Neuroscience
• Clinical Psychology

Background:

• Affective biases, common in depression and anxiety, involve preferential processing of negative stimuli.
• Understanding and modifying these biases is crucial for improving mental health interventions.
• Reliable measurement of affective bias over time is essential for evaluating intervention effectiveness in clinical trials.

Purpose of the Study:

• To examine the test-retest reliability of two behavioral affective bias tasks: the Ambiguous Midpoint and Go-Nogo tasks.
• To assess the reliability of both standard summary statistics and computational model parameters.
• To inform the development of more effective randomized controlled trials for mental health interventions.

Main Methods:

• An online study involving 58 general population participants.
• Participants completed the Ambiguous Midpoint and Go-Nogo tasks twice, with a minimum 14-day interval.
• Reliability was analyzed using Pearson's correlations and intra-class correlations for summary statistics and computational model parameters.

Main Results:

• Standard summary statistics for affective bias tasks showed moderate reliability (0.18–0.49).
• Computational model parameters were often less reliable than summary statistics.
• A generative modeling approach incorporating session covariance improved stability estimates.

Conclusions:

• Measures derived from these affective bias tasks exhibit moderate test-retest reliability.
• Further enhancements in task reliability are needed to maximize the informativeness of randomized controlled trials.
• Improved reliability will enhance the ability to detect changes and draw inferences from intervention studies.