Web-Based Assessment of Visuospatial Processing Speed Across the Lifespan

Francesca C. Fortenbaugh, PhD

VA Boston Healthcare System
Web-Based Assessment of Visuospatial Processing Speed Across the Lifespan

Scientific Abstract

Background: Declines in visuospatial processing speed in older adults are well documented, yet fine-grained lifespan changes are not well characterized. As many psychiatric and neurological disorders impact processing speed, there is a need for measures that can assess performance across the lifespan.

Methods: We developed a novel visuospatial processing speed task for web-based platforms adapted from the Useful Field of View paradigm. This dual task required both discriminating a tumbling E’s orientation at fixation and localizing a blue diamond among a peripheral ring of distractors (conjunction search) on each trial using a 50-trial adaptive bestPEST procedure to determine stimulus duration thresholds. Data was collected from 4,718 volunteers between 12-62 years old (55% male) on TestMyBrain.org.

Results: Binned average thresholds as a function of age were calculated using a 3-year sliding window. The average thresholds were well modeled by a segmented linear function with peak performance occurring at 22 years (95%CI: 19-28) and increasing by approximately 9ms/year afterward. This is reflected in the observed doubling of threshold durations across 20 vs. 60-year- old participants (412ms vs. 817ms). As this task assumes fixed slope and lapse rate parameters for the Weibull function used in the bestPEST algorithm, a simulation study was also completed to assess how deviations in these parameters from an individual’s best-fitting parameters impact the stability of threshold results. Over 7 million simulations were completed varying combinations of threshold, slope, and lapse rate parameters to determine the probability of a correct response when the bestPEST was run using the fixed slope and lapse rate. Results show strong correlations between true and estimated thresholds for all slope/lapse rate combinations (r = 0.84-0.99). Comparisons across pairs of thresholds show that ordinal rankings for threshold differences >20% were preserved over 90% of the time.

Conclusions: These simulation results validate that the slowing observed in our online data reflects age-related declines in processing speed and use of adaptive algorithms for web-based testing platforms.

Live Zoom Session – April 21st

research Areas

Authors

Francesca C. Fortenbaugh, PhD, Julia M. Brau, Joseph DeGutis, PhD, Laura Germine, PhD, Michael Esterman, PhD, & Regina McGlinchey, PhD