Resting parasympathetic activity is associated with malodor‐induced change in perceived foreignness of speakers

Abstract Introduction To protect against infection, individuals have evolved context‐dependent pathogen‐avoidant strategies, including selective social behaviors aimed at avoiding foreign individuals who may possess greater risk of infection. Parasympathetic nervous system (PNS) activity is associated with social engagement and regulation of the classical immune system but has not been widely investigated in relation to changes in intergroup perception and the behavioral immune system. Method The current research investigated the relationship between parasympathetic activity and perceived foreignness of in and outgroup speakers during exposure to a pathogen‐relevant odor (butyric acid). High‐frequency heart rate variability was measured at rest and while participants rated foreignness of speakers with and without the odor present. Results Findings show that exposure to the odor was associated with higher foreignness perceptions of outgroup speakers and lower foreignness perceptions of ingroup speakers. This effect was especially evident among individuals with higher resting parasympathetic activity. Conclusion These results suggest that the PNS may play a role in changes in social perceptions during a behavioral immune response.

influences perception (Carcone et al., 2015;Fowler, Lilienfeld, & Patrick, 2009), and provides evidence that individuals can recognize that a speaker has an accent after listening to as few as four words (Hailstone et al., 2012).Therefore, we used of a range of durations to examine how the effect of time exposed to a speaker's voice affects perceptions.300 workers on Amazon's Mechanical Turk (Mturk) completed ratings of the audio recordings in order to create two sets matched in foreignness, sex, and duration to use in the main study.Each Mturk participant was asked to rate 5 recordings to limit effects of habituation.
They were asked "How foreign does this speaker sound?(1=not at all foreign, 9=very foreign)", "How comprehensible is the speaker?(1=very difficult to understand, 9=very easy to understand)", whether the speaker was male or female, and which word of four choices was spoken in the clip.The sex of the speaker and word choice questions were used to evaluate attention to the recording.Of the 300 participants, 188 were male, 112 female; the mean age was 32.5, SD 10.1; 78% were White, 9% Asian, 7% Black, 3% Bi-or Multiracial, 3% Hispanic, 3% American Indian.Participants that incorrectly identified the sex of the speaker or the word spoken in the recording were excluded from the ratings for that recording (n=31).Participants that reported a language other than English as their native language were excluded from all ratings (n=7).Additionally, any workers not originating from the United States were excluded from all ratings (n=1).Each recording was rated by an average of 34 workers (SD 2.2).After the ratings were complete, recordings were split into two sets of 18 speakers that did not differ in perceived foreignness (t34= 0.170, p=.866).This analysis was repeated with the University of Chicago sample reported in the manuscript which confirmed that there was no significant difference in perceived foreignness between speakers sets (t34= .043,p=.966).Further, all reported analyses remained unchanged when controlling for speaker set presentation order.

Assessing Model Fit
Chi-square goodness-of-fit tests using the car package in R were calculated to examine whether adding item-level random intercepts and random slopes improved model fit for the HLM examining the effect of condition (odor/control) and speaker group (ingroup/outgroup) on perceived foreignness.Goodness-of-fit tests were supplemented by power simulation analyses through the simr R package (Green & MacLeod, 2016) (1000 simulations per analysis) to decide on the model to be presented in the manuscript.Results show that adding item-level random intercepts, and random slopes, respectively, to the model did significantly improve model fit for perceived foreignness, however the random slopes model was not sufficiently powered (Table S1).Therefore, the HLM described in the manuscript implements random intercepts at the subject-and item-level, but no random slopes.

Manipulation Check for Baseline Group Differences
As a manipulation check, we conducted an HLM examining the effect of condition (odor/control) and speaker group (ingroup/outgroup) on perceived foreignness at baseline to make sure subsequent results in post-odor manipulation data were not likely to be driven by preexisting differences in the odor and control conditions.Results show no existing baseline differences in foreignness (interaction effect: β = -0.03,p =.605) between ingroup and outgroup speakers as a function of experimental condition (odor/control).

Additional Analyses
Additional HLMs were run to examine whether the principal finding that participants in the odor condition demonstrated higher foreignness ratings of outgroup speakers and lower foreignness ratings of ingroup speakers compared to control participants was moderated by the sex of the speaker or the duration of the audio recording.A HLM examining the effects of condition (odor/control), speaker group (ingroup/outgroup) and sex of speaker (male/female) on perceived foreignness was not significant (interaction effect: β = 0.09, p = .377,95% CI [-0.11, 0.28]), showing that the effect of the odor manipulation and speaker group on perceived foreignness was not influenced by the sex of the speaker.A HLM examining the effects of condition (odor/control), speaker group (ingroup/outgroup) and duration of recording (3,6,9 seconds) on perceived foreignness was not significant (interaction effect: β = 0.02, p = .276,95% CI [-0.02, 0.06]), showing that the effect of the odor manipulation and speaker group on perceived foreignness was not influenced by the length of time exposed to the speakers' voices.
We conducted additional linear models to test whether negative affect or emotional arousal may explain the difference in perceived foreignness of speakers due to the odor manipulation.Participants rated a series of unpleasant pictures from the International Affective Picture System (IAPS) (Lang, Bradley, & Cuthbert, 2008) in perceived negativity and emotional arousal during the odor manipulation, in addition to rating the accented speakers.A linear model examining the effect of condition (odor/control) on perceived negativity of unpleasant pictures was not significant (β = -0.13,p = .638,95% CI [-0.65, 0.40]), showing that our measure of negative affect was not affected by the odor manipulation.Similarly, the linear model examining the effect of condition (odor/control) on perceived emotional arousal of unpleasant pictures was not significant (β = 0.07, p = .781,95% CI [-0.46, 0.60]), demonstrating that emotional arousal was not impacted by the odor manipulation.We also examined negativity and emotional arousal ratings as statistical moderators in the HLM testing the effect of condition (odor/control) and speaker group (ingroup/outgroup) on perceived foreignness.A HLM examining the effect of condition (odor/control), speaker group (ingroup/outgroup), and negativity of unpleasant pictures was not significant (interaction effect: β = -0.04,p = .408,95% CI [-0.14, 0.06]), showing that our measure of negative affect does not explain results.Similarly, the linear model examining the effect of condition (odor/control), speaker group (ingroup/outgroup), and emotional arousal of unpleasant pictures was not significant (interaction effect: β = -0.06,p = .246,95% CI [-0.16, 0.04]), demonstrating that emotional arousal does not explain results.These analyses suggest that our findings were not due to negative affect or emotional arousal from the odor.Independent groups t-tests revealed no significant differences between conditions were detected (all ps > 0.05).

Table S2 .
Means and standard errors for individual difference measures between experimental conditionsNote.TableS2lists means and standard errors for the odor and the control conditions.