PhD defense by Simon Nielsen
While the richness of our visual perceptions is nearly boundless, the rate with which we can perceive information is limited. For instance when we are required to perceive two consecutive target objects following briefly after each other, the accuracy with which we can report the second target is often reduced in the first half second. This phenomenon is known as the attentional blink (Raymond, Shapiro & Arnell, 1992) and as suggests by the name is assumed to pertain to how fast attention can be reallocated. Bottleneck models suggest that the attentional blink is caused by limited capacity in processing targets, which effectively causes a perceptual bottleneck (Chun & Potter, 1995). According to bottleneck models, making the first target easier to perceive should improve processing in the bottleneck and reduce the attentional blink. However, recent studies suggest that an attentional blink may be triggered by attention capture to the first object (Folk, Leber & Egeth, 2008) and that if making the first target easier to perceive increase its saliency this may increase the attentional blink (Chua, 2005).
This thesis examines the attention capture hypothesis with focus on empirical investigations and a theoretical review. Specifically this work presents studies in which first target contrast is varied in two different attentional blink paradigms, while potential influences from bottleneck effects are controlled. Publication 1 describes findings using the two-target paradigm (Duncan, Ward & Shapiro, 1994) where two masked targets are presented in different locations. Here we find that the attentional blink increases with first target contrast, however, only when no mask follows the first target. To further examine the effect of first target contrast, we disentangle the potential influence of bottleneck effects and vary first target contrast while maintaining target difficulty constant. Again we find that first target contrast increases the attention blink. Publication describes finding using the rapid serial visual presentation paradigm (Potter & Levy, 1969), in which two targets are presented centrally in the same location embedded in a stream of distractor objects. These findings replicate those from Publication 1, and suggest that the effect is not entirely spatial, since the rapid serial visual presentation paradigm does not require a spatial shift of attention to a new location. In addition to the findings in Publication 1, Publication 2 shows that the effect of first target contrast can be cancelled by the opposing effect of second target contrast.
Thus the results presented here are consistent with an attention capture hypothesis and suggest that the first target can trigger an attentional blink, and that the size of the blink increases with first target contrast.
Supervisor: Associate Professor Tobias Andersen, Cognitive Systems, DTU Informatics