Does gaze direction affect accuracy?

Abstract

The gaze direction effect is the finding that observers are typically faster at detecting or identifying a target when it appears in the direction indicated by the gaze of a centrally presented face as compared to other locations. The present research investigated whether the gaze direction effect would occur in accuracy when the target was visually degraded and accuracy was emphasized. In two experiments, the targets were easy to identify correctly, and reaction time (RT) was the dependent variable. In similar experiments, the targets were made difficult to identify and accuracy was the dependent variable. Gaze direction affected RT, but not accuracy. A theoretical mechanism for the gaze direction effect that accounts for these findings is presented.

Acknowledgements

We would like to thank Artem Violentyev for his help in running numerous pilot studies that lead to the experiments reported here and Steven Whittle and Hui (Charles) Li for their help in running Experiments 4 and 5. We would also like to thank Ayelet Landau, Allison Connell, Richard Ivry, and Art Shimamura for their helpful comments on the manuscripts, and Tom Wickens for his insightful discussions on meta-analysis.

Notes

¹During the review process, a third paper that examined the effects of gaze direction and accuracy came to our attention (Stevens, West, Al-Aidroos, Weger, & Pratt, in press). This study was conducted independently and approximately concurrently with the present study. It will be discussed in the General Discussion.

²These values were calculated from the individual condition standard deviations as if the experiment was an independent group's experiment, as suggested by Dunlap, Cortina, Vaslow, and Burke (1996).

³Given that Soto-Faraco et al.'s (2005) conclusion was based on a post hoc analysis, an independent replication of the effect seems warranted, especially in light of the present results and those of Stevens et al. (in press). To be equivalent to the present experiments, the following condition would be important: (1) Trial-by-trial feedback is given to ensure observers know the cue is noninformative; (2) it is clear there is no speed pressure whatsoever; (3) independently it is ascertained that observers know the location of the stimulus with nearly 100% accuracy.