When Testing Becomes Learning—Underscoring the Relevance of Habituation to Improve Internal Validity of Common Neurocognitive Tests

Abstract

Testing neurocognitive function is receiving growing attention in psychological and physical health research. To counteract the costs, reduced accessibility, and complexity of brain imaging (e.g., CT scans and fMRI) or function tests, neurocognitive performance tests (e.g., the Stroop test, the Trail Making Test, or the Choice Reaction Task) are commonly implemented. Although reliability is considered paramount when interpreting intervention effects, a detailed quantification of systematic and random errors is scarce. By recruiting 68 healthy participants from different age groups (7–64 years), we quantified population-specific measurement errors in the aforementioned neurocognitive tasks. The goal was to raise awareness about the impact of learning effects on reliability assessments and their interpretation. By performing five testing sessions with two trials per day, we observed significant learning effects from repeated testing. Trial-to-trial improvements of up to 50% were measured, accompanied by a random measurement error reduction from day to day. These learning effects were task and population specific, highlighting the need for caution when transferring reliability coefficients from other studies. The quantification of systematic and random measurement errors underscores the importance of conducting sufficient habituation sessions in neurocognitive tasks, as test protocols lack validity if they do not ensure reliability. Therefore, sufficient habituation sessions (i.e., until no meaningful learning effects can be observed) may be warranted when testing is repeated within short timeframes.