The effects of verbal labelling on psychophysiology: Objective but not subjective emotion labelling reduces skin-conductance responses to briefly presented pictures

Abstract

Verbally labelling emotional stimuli has been shown to reliably decrease emotional responding. The present study compared the use of identical emotional labels during two types of verbal labelling: subjective labelling of one's own emotional response and objective labelling of the stimulus. We recorded skin conductance responses (SCRs) to emotional pictures presented at four brief durations preceding a backward mask. We observed that as the exposure duration increased, SCRs decreased during objective labelling of the stimuli. However, when participants subjectively labelled their own emotional state, SCRs increased as exposure duration increased. In addition, subjective labelling produced larger SCRs than objective labelling at a shorter exposure duration when the presented stimuli were biologically prepared. These results indicate that only objective labelling results in decreased emotional responding, and describe a novel interaction between bottom-up stimulus characteristics and top-down cognitive effects on physiological responses.

Keywords:

• Backward mask
• Skin conductance
• Labelling
• Emotion
• Biologically prepared

Acknowledgements

This research was supported by a grant from the Fetzer Institute (Kalamazoo, MI).

The authors would like to thank Carolyn L. Fort and Bettina Fetzer for their assistance with data management.

Notes

¹No main effects of or interactions with LEAS were observed.

²We conducted additional analyses to ensure that the SCRs reported here were not due to properties of the masks shown. We found that the results reported here remained significant when partialling out variance in SCRs due to the mask presented.

³Raters were instructed to make a forced-choice decision about the presence of material in the slide that was likely to have “elicited the same emotional response early in evolutionary history.” An example was given: snakes have likely provided a threat to humans for many generations, whereas due to the recent invention of the handgun, it is unlikely to have elicited an emotional response until very recently in history. It was clarified that for neutral stimuli, biological relevance was to be indicated if the neutral object would have been present early in evolutionary history (landscapes, trees, etc., as opposed to fire hydrants, hair dryers, etc.). In slides that presented mixed stimuli, raters were instructed that if any prominent object in the slide was biologically relevant, that rating would prevail (e.g., an angry face displayed prominently with several handguns would be biologically relevant due to the featured face). Ratings that were close to equally divided between the raters (17 of 96 images) were decided upon by the rating of an author familiar with the biological relevance literature (KM), again with a bias towards labelling mixed pictures as biologically relevant.

⁴Due to the unequal cell sizes that were a result of post hoc ratings of biological preparedness, we also conducted an analogous analysis using a hierarchical linear model (HLM), which is more resistant to missing data and uneven cell sizes. This also allowed for the inclusion of the 8 subjects for whom we did not obtain SCR data for every condition. These results show an almost identical pattern of results, with the exception of a main effect of labelling (subjective > objective, F=4.95, p<.026), the reduction of the Labelling×Exposure Duration interaction to a trend (p=.074). No other effects changed in significance or direction.

⁵This trend was even weaker in the HLM (p=.1). It was characterised by the “typical” SCR pattern during the intermediate exposure durations (33.2 and 66.4 ms; pleasant and unpleasant greater than neutral) but SCRs to neutral pictures larger than those to pleasant or unpleasant pictures during the shortest and longest exposure durations (16.6 and 132.8 ms).

⁶We also examined the distribution of responses across the pleasant, neutral and unpleasant response options. We did not observe differences in these responses by condition (p>.2).