Constraints on the locus/loci of the attentional blink phenomenon

Abstract

The locus of the Attentional Blink (AB) was investigated using AB and Psychological Refractory Period (PRP) experimental paradigms. The first task in the AB experiments (Exps. 1 and 2) was to identify a target letter (T1), which was an H, O, or S, presented in an RSVP stream of distractor letters. The second task in these experiments was to identify a second target letter (T2), an X or Y. In Experiment 1, the response to T1 was speeded and in Experiment 2 it was unspeeded. T2 always appeared in one of the 8 positions in the RSVP stream immediately following T1 and the response to T2 was never speeded. Accuracy on identifying T2 was severely attenuated when T2 followed within approximately 500 ms after T1. In the PRP experiments (Exps. 3 and 4), the first task was to judge whether an auditory tone (S1) was low, medium, or high in pitch. Following the tone, at variable stimulus-onset asynchrony (SOA), a target letter (S2) was presented and the second task was to identify the target letter, which could be an H, O, or an S. In these experiments both responses were speeded. Mean response time to S2 slowed as SOA decreased.

The probability of the target letter, T1 in the AB experiments, and S2 in the PRP experiments, was manipulated so that one letter was assigned a relative frequency value of 1, another a value of 4, and another a value of 9. In the AB experiments the least frequently presented target letter produced a larger AB effect than the target letter assigned to the intermediate frequency condition, which in turn produced a larger effect than the most frequently presented target letter. These results indicate that the locus of signal probability is at, or before, the locus of the AB effect. In the PRP experiments, signal probability was additive with SOA. According to the locus of cognitive slack logic (Pashler & Johnson, 1984), additive effects indicate that the locus of probability is at, or beyond, the locus of the PRP bottleneck.

These results lead to the conclusion that a locus of the AB effect is at, or after, the PRP bottleneck. It is generally believed that the PRP bottleneck is located at a late stage of information processing, such as response selection. Thus, these results suggest that the AB effect also has its locus at a late stage of processing, possibly at, or after, the stage where responses are selected.