Higher social tolerance is associated with more complex facial behavior in macaques
- Department of Psychology, Centre for Comparative and Evolutionary Psychology, University of Portsmouth, United Kingdom
- Centre for Interdisciplinary Research on Social Interaction, Department of Psychology, Nottingham Trent University, United Kingdom
- CNRS-MNHN-Université Paris Cité, France
- School of Biological and Behavioural Sciences, Queen Mary University of London, United Kingdom
- School of Psychology, University of Lincoln, United Kingdom
Figures
Figure 1
Bootstrapped entropy ratio of facial behavior across social contexts for three species of macaques.
The entropy ratio was calculated on 100 bootstrapped samples of the data by dividing the observed entropy by the expected entropy if Action Units were used randomly for each social context. The entropy ratio ranges from 0 to 1, with higher values indicating higher uncertainty. Symbols and whiskers indicate mean and range of bootstrapped values.
Figure 2
Bipartite network of single Action Units (orange) and social context (blue) for three species of macaques.
Edges are shown for Action Units that occurred in at least 1% of observations per context. Edge thickness and transparency are weighted by specificity, which ranges from 0 (indicating an Action Unit is never observed in a context) to 1 (indicating an Action Unit is only observed in one context). Context abbreviations: agg = aggressive, aff = affiliative, sub = submissive.
Figure 3
Specificity of Action Unit combinations that were used in at least 1% of observations per species per social context.
Specificity ranges from 0 (indicating an Action Unit is never observed in a context) to 1 (indicating an Action Unit is only observed in one context). (A) Distribution of Action Unit combination specificity. Width of violin plots indicate the relative density of the data. Colored symbols indicate unique Action Unit combinations. White symbols indicate mean specificity. (B) Proportion of Action Unit combinations used with high (>0.8), moderate (0.4–0.8), or low (<0.4) specificity. Context abbreviations: agg = aggressive, aff = affiliative, sub = submissive.
Figure 4
Calculating context specificity on an imbalanced dataset.
Specificity was calculated on a simulated dataset with an imbalanced number of observations per context. The calculated specificity values deviated from the true specificity such that they were higher in the context with most observations and lower in the context with fewest observations (green circles). Randomly upsampling observations from the minority contexts (B and C) such that they have the same number of observations as the majority context (A) prior to calculating specificity minimized the bias in the calculated specificity values (purple triangles).
Tables
Table 1
Confusion matrices for random forest classifier predictions of social context from Action Unit combinations.
| Truth | |||
|---|---|---|---|
| Prediction | Affiliative | Aggressive | Submissive |
| Rhesus | |||
| Affiliative | 636 | 19 | 9 |
| Aggressive | 81 | 1205 | 17 |
| Submissive | 2 | 6 | 731 |
| Barbary | |||
| Affiliative | 2573 | 24 | 442 |
| Aggressive | 200 | 1219 | 165 |
| Submissive | 166 | 34 | 528 |
| Crested | |||
| Affiliative | 1134 | 90 | 43 |
| Aggressive | 16 | 86 | 11 |
| Submissive | 3 | 1 | 7 |
Table 2
Total number of social interactions per species and social context that were MaqFACS coded.
Note that combination of Action Units were grouped by time blocks of 500 ms. Therefore, the number of observations in the data is twice the duration of the social interaction in seconds.
| Species | Context | N interactions | N subjects | Duration (s) |
|---|---|---|---|---|
| Rhesus | Affiliative | 193 | 29 | 1197 |
| Aggressive | 413 | 32 | 2050 | |
| Submissive | 318 | 31 | 1262 | |
| Unclear | 121 | 30 | 802 | |
| Barbary | Affiliative | 683 | 43 | 4897 |
| Aggressive | 585 | 44 | 2128 | |
| Submissive | 529 | 34 | 1890 | |
| Unclear | 603 | 45 | 3500 | |
| Crested | Affiliative | 241 | 35 | 1918 |
| Aggressive | 62 | 23 | 284 | |
| Submissive | 25 | 18 | 115 | |
| Unclear | 107 | 25 | 684 |
Additional files
-
Supplementary file 1
A complete list of Action Units and Action Descriptors coded in this study.
- https://cdn.elifesciences.org/articles/87008/elife-87008-supp1-v1.docx
-
MDAR checklist
- https://cdn.elifesciences.org/articles/87008/elife-87008-mdarchecklist1-v1.pdf
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Higher social tolerance is associated with more complex facial behavior in macaques
eLife 12:RP87008.
https://doi.org/10.7554/eLife.87008.3
