Eye activity correlates of workload during a visuospatial memory task

Changes in six measures of eye activity were assessed as a function of task workload in a target identification memory task. Eleven participants compl eted four 2-hr blocks of a mock anti-air-warfare task, in which they were r equired to examine and remember target classifications (friend/enemy) for s ubsequent prosecution (fire upon/allow to pass), while targets moved steadi ly toward two centrally located ship icons. Target density served as the ta sk workload variable; between one and nine targets were simultaneously pres ent on the display. For each participant, moving estimates of blink frequen cy and duration, fixation frequency and dwell time, saccadic extent, and me an pupil diameter integrated over periods of 10 to 20 s, demonstrated syste matic changes as a function of target density. Nonlinear regression analyse s found blink frequency, fixation frequency, and pupil diameter to be the m ost predictive variables relating eye activity to target density. Participa nt-specific artificial neural network models, developed through training on two or three sessions and subsequently tested on a different session from the same participant, correlated well with actual target density levels (me an R = 0.66). Results indicate that moving mean estimation and artificial n eural network techniques enable information from multiple eye measures to b e combined to produce reliable near-real-time indicators of workload in som e visuospatial tasks. Potential applications include the monitoring of visu al activity of system operators for indications of visual workload and scan ning efficiency.