Abstract
The size of the attentional window may depend on attentional, perceptual and cognitive demands. Here, we test whether pupillometry can be used to assess such changes in attentional-window size. At three locations in the visual field, centrally and 8.2 degrees left and right, we presented streams of colored, oriented bars at a rate of 300 ms per item. In a low-load condition, observers had to respond to the appearance of a single feature at the central location: either a specific color (irrespective of orientation), or a specific orientation (irrespective of color). The task-relevant feature was blocked. In a high-load condition, observers had to respond to two conjunctions of color and orientation (e.g., to red-vertical and blue-horizontal). The peripheral streams were irrelevant to the task and embedded in circular discs (2.7 degrees in diameter) that oscillated sinusoidally from black to white at 1.2 Hz. With all observers, performance was worse and reaction times slower in the high-load as compared to the low-load condition, indicating a successful manipulation of attentional demands. We found differences in target-evoked pupil responses: responses tended to be earlier in the low-load condition, corroborating the reaction time results. Pupil responses to successfully detected targets tended to be larger in the high-load than in the low-load condition, again supporting the successful manipulation of load. Most importantly, we found the Fourier amplitude at 1.2 Hz to be significantly larger for the low-load than for the high-load condition. This indicates a narrowing of the attentional window under high load - fewer resources are allocated to the peripheral streams and oscillating discs if demands increase. Hence, the induced pupil fluctuations are reduced under high-load. Together, our results show that pupillometry can be used to map the narrowing of the attentional window under different task demands.
Acknowledgement: German-Israeli Foundation for Scientific Research and Development (grant no: I-1418-105.4/2017)