Abstract
We investigate the ability of highly trained subjects to conform spatial attention into arbitrary complex patterns in a search task. Subjects search a 12×12 jittered disk array for a target disk that is 50% larger in diameter than the distractor disks. The target occurs in a random one of 72 to-be-attended locations. In Experiment 1, the to-be-attended locations are defined by a pseudorandom arrangement of 8 3×3 blocks, in Experiment 2 by 18 2×2 blocks. Ten false targets are randomly placed in unattended locations to force the subject to confine attention to the to-be-attended locations. An attention instruction is a spatial diagram in which to-be-attended areas are green, to-be-ignored areas are reddish-brown. To begin each trial, an attention instruction is shown, it fades out, and is followed by a 150 msec exposure of the 12×12 search array. In each experiment, on successive trials, the instruction pattern alternates randomly with its negative.
For two subjects, the percents correct were 45%, 57% for 8-block patterns, 25%, 43% correct for 18-block patterns. Subjects conformed their distributions of attention partially but not perfectly to the block patterns. In addition to the attentional limitations, search accuracy depends on exposure duration, target-distractor difference, the number of false targets, eccentricity, and crowding. All these factors were estimated from search experiments with simple attentional distributions, and used to generate parameter free predictions of the data. The results suggest that as attention attempts to conform to more-and-more complex patterns, complexity itself becomes a limiting factor.
Air Force Office of Scientific Research, Life Science, Grant FA9950-04-1-0225