A Comparison of the Aged to the Bayesian Ideal Observer Model for Cueing Tasks with Peripheral and Central Cues

Previous studies investigating automatic (peripheral cues) and voluntary (central cues) attention decline with age have produced inconsistent findings. The present study investigated automatic and voluntary attention in the aged by comparing performance in 15 younger (18–25 years) and 15 older (65–78 years) observers to the Bayesian Ideal Observer Model (BIOM). The BIOM predicts cueing effects by the optimal differential weighting of likelihoods of targets appearing at each location (cue validity). It also allows the measure of bias in the responses and accounts for the detectability of targets, a factor that influences the predicted size of the cueing effect in the BIOM (Shimozaki et al 2003, Journal of Vision 3 209–229) and thus may explain some of the inconsistencies in previous research. Observers performed a yes/no cueing task of 2D Gaussian targets (60 ms, eccentricity 8 deg) appearing at one of two locations. Pre-cues (150 ms) were 70% valid and appeared either in peripheral (2 deg squares, Experiment 1) or central (shapes symbolizing left or right, Experiment 2) locations. Detectability was controlled across observers with a staircase, and difficulty was manipulated (contrast = threshold or one-octave below threshold). Both age groups showed cueing effects whether the cue appeared in peripheral or central locations, suggesting that automatic and voluntary attention are preserved in older adults. Both groups weighted the valid cue optimally at both levels of detectability (threshold and 1-octave below). However, for centrally located cues (voluntary attention), older adults were less accurate and exhibited more bias than their younger counterparts