Illusory correlations between neutral and aversive stimuli can be induced by outcome aversiveness

Abstract

Anxiety is often characterised by an overestimation of the contingency between concern-relevant and aversive stimuli, and an amplified aversiveness of such aversive stimuli. Here we tested whether outcome aversiveness causally enhances contingency estimates. Twenty-four participants were exposed to neutral visual cues which were followed by startle sounds. The loudness and hence the aversiveness of the cue was determined by the cue colour while the likelihood of the startle sound always remained constant (50%). Results indicated an illusory correlation specifically for the cue followed by the most aversive startle sound as reflected in enhanced on-line and a posteriori covariation estimates. This bias was positively correlated with state and trait anxiety. Physiological arousal measured by pupil diameter was enhanced in response to the most aversive startle sound confirming its distinct processing. In conclusion, these results suggest that aversive outcomes may induce illusory correlations, most likely in anxious persons, and explain previous findings of illusory correlations in anxiety disorders.

Keywords:

• Covariation bias
• Illusory correlation
• Aversiveness
• Startle sound
• Anxiety
• Pupil diameter

This work is part of the dissertation of Julian Wiemer. This work is not prone to bias by any potential or actual conflicts of interest.

We thank Johanna Hamprecht and the students of Experimental Clinical Psychology of the University of Würzburg for their assistance in data acquisition.

This work is part of the dissertation of Julian Wiemer. This work is not prone to bias by any potential or actual conflicts of interest.

We thank Johanna Hamprecht and the students of Experimental Clinical Psychology of the University of Würzburg for their assistance in data acquisition.

Notes

¹ These terms are used synonymously from now on.

² The exclusion of 24% of the trials led to a better signal-to-noise-ratio, but did not change the significance of the results in comparison to no exclusion of trials.