The Intensity of Internal and External Attention Assessed with Pupillometry

Abstract

Not only is visual attention shifted to objects in the external world, attention can also be directed to objects in memory. We have recently shown that pupil size indexes how strongly items are attended externally, which was reflected in more precise encoding into visual working memory. Using a retro-cuing paradigm, we here replicated this finding by showing that stronger pupil constrictions during encoding were reflective of the depth of encoding. Importantly, we extend this previous work by showing that pupil size also revealed the intensity of internal attention toward content stored in visual working memory. Specifically, pupil dilation during the prioritization of one among multiple internally stored representations predicted the precision of the prioritized item. Furthermore, the dynamics of the pupillary responses revealed that the intensity of internal and external attention independently determined the precision of internalized visual representations. Our results show that both internal and external attention are not all-or-none processes, but should rather be thought of as continuous resources that can be deployed at varying intensities. The employed pupillometric approach allows to unravel the intricate interplay between internal and external attention and their effects on visual working memory.

Keywords: External attention; Intensity; Internal attention; Pupillometry; Visual working memory.

Figure 1

Schematic overview of the procedure and the main results. Left: Procedure from top to bottom, see Methods for details. Top right: Pupil constriction was more pronounced during more precise encoding into VWM as indicated by smaller absolute hue errors. For visualization only, three groups were created for the pupil trace plots but analyses were conducted using absolute error as a continuous variable. Horizontal purple lines indicate significant effects of absolute error on pupil size (p < .05). The X-axis represents time (ms) and the Y-axis holds pupil size (arbitrary units). Middle right: Stronger dilation during prioritization leads to more precise VWM representations (same conventions as above). Bottom right: Probe conditions differed in absolute hue errors. All error shadings and bars reflect standard errors of the mean. *p < .05.

Figure 2

Relationships between visual working memory precision and A) encoding constriction amplitude (external attention), and B) prioritization dilation amplitude (internal attention). Thin lines represent linear fits per individual and the thick like represents the relationship across all data. Shaded error bars represent bootstrapped 95% confidence intervals. Y-axes are shared between A and B. To account for individual differences, data were normalized to robust z-scores by subtracting the median and dividing by the median absolute deviation for each datapoint per participant (Rousseeuw & Hubert, 2011). *p < .05, **p < .01.