Abstract
Age-related differences in cognitive test performance must be characterised for each assessment format before scores can inform individual interpretation. This cross-sectional study examined age-related differences on the Trail Making Test in Virtual Reality (TMT-VR), an immersive adaptation of the Trail Making Test, across the adult lifespan. Specifically, it examined whether completion time and wrong-target errors increased with age, whether the age association differed between Parts A and B, whether it was linear or accelerated at older ages, whether it held within each of three adult developmental periods, namely early (19–39 years), middle (40–59 years) and late (60–90 years) adulthood, and which of the exported and derived scores carried the age signal. The sample comprised 230 healthy adults aged 19–90 years (131 women). Part-level mixed-effects models with participant random intercepts tested the linear, quadratic and part-specific age terms; age-band and age-balancing robustness analyses followed, together with a comprehensive, multiplicity-controlled examination of all 26 TMT-VR scores and a direct comparison of their age associations. Older age was strongly associated with longer completion time and moderately with more wrong-target errors on both parts. The associations were linear across the adult range on the modelled scales, with no evidence of acceleration at older ages, and the completion-time slope per decade was of similar size within each developmental period. The completion-time Age × Part interaction was not clearly supported; the absolute Part B minus Part A difference grew with age while the B/A ratio did not. This is compatible with proportional slowing across parts, without establishing equivalent slopes or an absence of executive ageing. Error-adjusted time, which combines duration with correctness, was among the most age-related scores at every level, but it correlated .99 with completion time and its age association was no stronger, making it an economical summary rather than additional evidence in healthy adults. Sex and education were unrelated to the outcomes beyond age; mean selection distance varied too narrowly across participants for an age association to be detectable. Completion time, error-adjusted time and wrong-target errors are therefore the TMT-VR scores on which age referencing should build.