Human social learning biases in immersive virtual environments

Abstract

AbstractSocial learning strategies describe what, when, and from whom individuals choose to learn. Evidence suggests that both humans and animals are capable of strategic social learning. However, human research generally lacks ecological and spatial realism, making it difficult to understand the importance of our use of social information in an evolutionary context. In this study, we use virtual reality to simulate three novel tasks inspired by the animal literature (Container, Route Choice and Foraging tasks) within complex, three-dimensional environments. In each experiment, combinations of demonstrators with different characteristics gave opposing solutions to the task to determine from whom participants preferentially learned. Importantly, participants were able to freely navigate the environment and attempt the task in any way they chose by using or ignoring social information. We found that participants displayed an overall bias towards learning asocially (independently) rather than socially. Asocial learning was favoured more strongly during complex tasks that spanned larger spatial scales, potentially due to the difficulties in keeping track of social information in such scenarios. When learning from others, participants displayed a bias towards learning from the majority over the minority (positive frequency-dependent social learning) and towards learning from the most successful demonstrators (payoff-based social learning), which supports the findings of previous, lab-based experiments. There was no apparent bias with respect to demonstrator dominance status, gender and body size. Our findings are the first to show a variation in the use of social learning across task and environmental complexities in humans, to carry out a comprehensive evaluation of hypothetical human learning biases, and to provide a methodological link between non-human and human social learning experiments. As demonstrated here, immersive virtual environments have great potential for research into human social evolution and we strongly encourage future research to adopt a similar approach.