A SYSTEMATIC REVIEW OF THE RELATIONSHIP BETWEEN SOCIAL LEARNING STRATEGIES AND SOCIAL DYNAMICS IN NON-HUMAN PRIMATES

Abstract

Background: the aim of this systematic review was to investigate the relationship between social dynamics and social learning strategies (or transmission biases) in non-human primate societies.
Design: a systematic literature search was conducted of publications which investigated social learning and / or social dynamics in non-human primates, both captive and wild. 1165 studies were found.
Data sources: Web of Science, Scopus, International Bibliography of the Social Sciences, Anthropology Plus, APA PsycInfo, Anthrosource, Google Scholar, OpenGrey
Eligibility criteria: all eligible studies needed to be focused on non-human primates; papers investigating humans or other animals were eliminated, but comparative studies investigating both non-human and human primates were kept. They had to be focused on social learning; studies referring to social dynamics only or broad ranges of behaviours in a social context were retained for theoretical frameworks regarding social dynamics, but were not eligible for the systematic review. Studies had to be experimental or observational (excluding theoretical studies, such as literature reviews and mathematical models). If there was a trained model or demonstrator involved in the social learning study, it had to be a hon-human primate, meaning that if it was trained by a human and then introduced into its social context to document a specific behavioural spread, the study was included; if instead, the focus was to test how a single non-human primate learns from a human model, the study was excluded. This means that the social context in which social learning occurred involved social dynamics between non-human primates only.
At this point, there was 91 eligible studies.
Data extraction and synthesis: the quality assessment was established on the capacity of each paper to answer the core question (e.g., “is there a relationship between social dynamics and social learning strategies?”). 27 studies focusing on social learning strategies did not have any reference to social dynamics and 20 studies suggested potential biases and did not report evidence supporting or not supporting a bias. Only the remaining 44 studies that clearly investigated social learning biases (whether they reported evidence for a social learning bias or not) and at the same time made references to social dynamics were included in the systematic review for the thesis. Due to heterogeneity of study designs and measurements, the findings were synthesised using narrative summary accompanied by tabs.
6
Results: I identified 44 studies that met the inclusion criteria and which reported social learning strategies, 25 in captive hon-human primates and in 19 wild non-human-primates. Overall, the 44 studies totaled 117 research efforts to investigate a bias. There was evidence for a diverse distribution of social learning strategies depending on the different social systems of non-human primates, according to predictions. Also, research efforts were biased towards particular species.
Conclusion: findings suggest that social dynamics influence the adaptive weight of different traits of individuals’ identities (e.g., kinship, rank, sex), which also influence the type of social learning strategies. In despotic species, model-biased social learning occurs more often than non-model biased social learning; in intermediate species both model-biased and non model-biased social learning occur approximately equally; tolerant species have been investigated too unfrequently and specific predictions upon their social systems could not be tested, but they displayed those biases such as “copy successful individuals” and “when naïve” that were spread evenly across all social systems, according to predictions.
Critical points: species have not been studied equally and therefore results may derive from the frequency with which one species has been studied. There is a high variability of conceptual definitions of some biases (e.g., conformity). Very few studies controlled for different biases and were able to rule out alternative explanations when reporting a bias. Similarly, there was a high variability of focus and metrics to address social dynamics. Also, one critical issue was to untangle Directed social learning from specific transmission biases: researchers might have assumed model-based bias when instead social learning was occurring as a byproduct of social dynamics and not from learning purposes.