Page 121 - Emotions through the eyes of our closest living relatives- Exploring attentional and behavioral mechanisms

Self-scratch contagion observed in orangutans
We created a generalized linear mixed model (GLMM) that included the identity of the
expresser and observer as random effect and “context” (categorical; tense vs. relaxed),
“relationship quality” (categorical; high versus low relationship quality) as fixed factors
to test their effect on the occurrence of scratch contagion. Furthermore, we decided
to include “seeing the triggering scratch” (categorical; seen vs. unseen) as additional
fixed factor since auditory cues of self-scratching can already be sufficient to induce a
contagious response in humans (Swithenbank et al., 2016). We included a three‐way
interaction for context, relationship quality and seeing the triggering scratch because
we hypothesized that contagious responses triggered by unseen scratches would
not be influenced by relationship quality, simply because the observer did not have
information about the expresser. Sex of the expresser, observer, and sex combination
were considered as additional fixed factors, but due to the low sample sizes (three
males and six females), we decided to leave them out. The models used a binomial
distribution (contagion or no contagion) and a logit link function. Likelihood ratio
tests and a c2 distribution were used to compare the full model with the null model. 5 Multicollinearity between independent variables was tested and variables with a
variance inflation factor (VIF) of > 5 were rejected from the model (O’Brien, 2007). None of the factors showed high VIF values. Analyses were conducted using R version 3.6.1 (R Core Team, 2020), with the GLMM calculated using the lme4 package (Bates et al., 2015).
Results
Orangutans are susceptible to scratch contagion
We compared the scratch rates during the baseline condition with the scratch rates in each of the 30 s intervals during the contagious condition. Orangutans scratched more during the first 90 s after a triggering scratch (Figure S1; bootstrapped paired samples t test: Baseline vs. 0–30 s: p < .001; Baseline vs. 31-60 s: p < .001; Baseline vs. 61-90: p=.002). Furthermore, the scratch rates over the 90s contagious condition were higher than the 90 s baseline condition (Figure 1; bootstrapped paired samples t test: p < .001). This suggests that only those scratches happening within 90 s after another scratch can be considered contagious.
119