Overt attentional prioritization of new objects and feature changes during real-world scene viewing

Abstract

The authors investigated the extent to which a change to an object's colour is overtly prioritized for fixation relative to the appearance of a new object during real-world scene viewing. Both types of scene change captured gaze (and attention) when introduced during a fixation, although colour changes captured attention less often than new objects. Neither of these scene changes captured attention when they occurred during a saccade, but slower and less reliable memory-based mechanisms were nevertheless able to prioritize new objects and colour changes relative to the other stable objects in the scene. These results indicate that online memory for object identity and at least some object features are functional in detecting changes to real-world scenes. Additionally, visual factors such as the salience of onsets and colour changes did not affect prioritization of these events. We discuss these results in terms of current theories of attention allocation within, and online memory representations of, real-world scenes.

Keywords:

• Attention
• Visual memory
• Oculomotor capture
• Real-world scenes
• Gaze control

Acknowledgements

This research was made possible by a grant from the Economic and Social Research Council (RES-000-22-2484) awarded to James Brockmole and John Henderson. We thank Krista Ehinger and Michael Mack for invaluable technical assistance. We also thank Carrick Williams, Geoff Cole, Robert Rauschenberger, Angus Gellatly, and an anonymous reviewer for their comments and suggestions during the course of this project.

Notes

¹Covert measures of attention capture whereby involuntary shifts of attention to dynamic singletons are measured through reaction time or response accuracy have a rich history (see Rauschenberger, 2003, and Simons, 2000, for reviews). Because our focus is on the capture of gaze, we will not review the covert capture literature in depth.

²This method resulted in postchange scenes that were not identical in the new object and colour change conditions. However, it allowed a direct comparison between the situations where a particular object appeared in a scene and when that same particular object changed its features.

³A similar conclusion follows from analysis of trial-level effects. The new object was fixated within the first four fixations after its appearance in the scene on 85% of trials in the fixation condition and on 40% of trials in the saccade condition, p<.001.

⁴To avoid issues of multicollinearity introduced by expressing the number of first looks to the onset at each of ordinal fixation position as a conditional probability, we performed the ANOVA on the raw number of times the first look occurred at each fixation position (see Brockmole & Henderson, 2005b).

⁵The choice of using four ordinal fixation positions was arbitrary and was used here to parallel results presented in our prior studies (Brockmole & Henderson, 2005a, 2005b). An ANOVA considering only the first three ordinal fixation positions revealed a reliable effect of trial type, F(1, 11) = 4.73, p=.05, demonstrating a prioritization effect specific to the transient signal. The analysis of trial-level effects also supports this conclusion; observers looked to the colour change on 54% of trials in the fixation condition and on 34% of trials in the saccade condition (p<.01).

Additional information

Notes on contributors

Michi Matsukura

M. Matsukura is now at the University of Iowa, IA, USA