Anxiety modulates the effects of emotion and attention on early vision

Abstract

At attended locations emotion and attention interact to benefit contrast sensitivity, a basic visual dimension. Whether there are associated costs at unattended locations is unknown. Furthermore, emotion and attention affect response time, and anxiety modulates these effects. We investigated how trait-anxiety influences the interaction of emotion and attention on contrast sensitivity. On each trial, non-predictive pre-cues (neutral or fearful faces) directed exogenous attention to four contrast-varying, tilted stimuli (Gabor patches). Attention was cued toward the target (valid), a distracter (invalid), or distributed over all locations. Observers discriminated target orientation, and completed self-report measures of anxiety. Effects of fearful expressions were mediated by trait anxiety. Only high-trait-anxious individuals showed decreased target contrast sensitivity after attention was diverted to a distracter by a fearful cue, and anxiety score correlated with degree of impairment across participants. This indicates that increasing anxiety exacerbates threat-related attentional costs to visual perception, hampering processing at non-threat-related locations.

Keywords:

• Emotion
• Attention
• Contrast sensitivity
• Fear expression
• Anxiety

Acknowledgments

This work was financed by NIH R01-EY016200 to MC and NIH R01-MH062104 to EAP. We thank Damian Stanley, Tobias Brosch and David Carmel for helpful discussions, as well as other Phelps and Carrasco Lab members for comments on earlier versions of this manuscript.

Notes

¹Data from seven observers could not be reliably fitted with a Weibull function. Deviance scores, which assess goodness of fit, exceeded χ² _.05(7)=14.1 (.05 refers to p-value; 7 refers to number of contrast levels). This was likely due to our participant population being inexperienced psychophysical observers and (1) making inconsistent responses and/or (2) having more “finger mistakes” in which the incorrect button was pressed even though the stimulus was correctly perceived. Four additional observers had contrast sensitivity > 3 SDs from the group mean after normalisation. Data from the 36 remaining observers (23 females) were included in the statistical tests. The mean deviance score across these participants and conditions was 4.43 (SD=2.76).

²Although with similar numbers of total trials others showed that emotion can improve perception (Bocanegra & Zeelenberg, 2009), we had 32 trials per data point (2 facial expressions×3 cue validities × 7 contrast levels) whereas they had 88 trials per data point (2 facial expressions × 5 spatial frequencies)—over twice the amount of data per data point.

³Both methods have been used extensively in psychophysical testing. On the one hand, the downside with using a staircase procedure is that it can be time-consuming and notoriously sensitive to inconsistent responses (a disadvantage of inexperienced observers), which could throw the contrast threshold estimate drastically off course. To get a reliable contrast threshold estimate, the results of many staircases are pooled (25 per condition in the Phelps et al. study), and if an estimate is not acceptable, the whole staircase has to be run again. Given that we tested six times the number of participants in order to investigate individual differences in anxiety, we opted to save time in data collection by not using staircase procedures. On the other hand, the downside with the method of constant stimuli is that, although all stimulus intensities are sampled, the contrast threshold may not be sampled enough in a two-hour experiment.