Abstract
In our previous studies (Seitz and Watanabe, 2003, Nature; Watanabe et al., 2001, Nature), subjects learned a motion stimulus through a reinforcement procedure in which the motion was subliminally presented as an irrelevant feature during an attentionally demanding letter task. Here we offer evidence that the bottleneck believed to be responsible for the decrease in correct report for the second target (T2) in the attentional blink (AB) effect encompasses processes critical to this perceptual learning. In our experiment, participants trained on the identification of two target digits presented within a rapid serial visual presentation of letters. When a short stimulus onset asynchrony (SOA) separated the two targets, T2 was less likely to be reported correctly, indicating an attentional blink effect. This condition is referred to as the ‘AB’ condition. When the SOA was long, T2 was reported correctly most of the time, indicating the absence of a blink effect (‘No AB’ condition). Irrelevant, unattended moving dots, with 5% motion direction coherence, were presented peripherally during the task. A particular direction was associated with targets presented in the ‘AB’ condition and another direction was paired with targets presented in the ‘no AB’ condition. Participants were tested on an identification of motion direction test before and after training in the dual target RSVP task. Results show that the motion direction associated with the no blink condition was learned, but not the direction associated with the blink condition. This suggests that the same lack of resources that are responsible for the incomplete processing of T2 affects the reinforcement process by which perceptual learning occurs.
This work was supported by grants NSF 0418182, NIH R01 EY015980-01and Human Frontier Foundation RGP18/2004 to Takeo Watanabe and NSERC finds to Pierre Jolicoeur