Language Evolution and Computation Bibliography

Training agents to communicate with one another given taskbased supervision only has attracted considerable attention recently, due to the growing interest in developing models for human-agent interaction. Prior work on the topic focused on simple environments, where training using policy gradient was feasible despite the non-stationarity of the agents during training. In this paper, we present a more challenging environment for testing the emergence of communication from raw pixels, where training using policy gradient fails. We propose a new model and training algorithm, that utilizes the structure of a learned representation space to produce more consistent speakers at the initial phases of training, which stabilizes learning. We empirically show that our algorithm substantially improves performance compared to policy gradient. We also propose a new alignment-based metric for measuring context-independence in emerged communication and find our method increases context-independence compared to policy gradient and other competitive baselines.

It has long been claimed that Homo sapiens is the only species that has symbolic language, but only recently recognized that humans also have an unusual pattern of growth and development. Social mammals have two stages of pre-adult development: infancy and juvenility. Humans have two additional prolonged and pronounced life history stages: childhood---an interval of four years extending between infancy and the juvenile period that follows---and adolescence---a stage of about eight years that stretches from juvenility to adulthood. We begin by reviewing the primary biological and linguistic changes occurring in each of the four preadult ontogenetic stages in life history. Then we attempt to trace the evolution of childhood and juvenility in our hominid ancestors. We propose that several different forms of selection applied in infancy and childhood; and that in adolescence, elaborated vocal behaviors played a role in courtship and intrasexual competition, enhancing fitness and ultimately integrating performative and pragmatic skills with linguistic knowledge in a broad faculty of language. A theoretical consequence of our proposal is that fossil evidence of the uniquely human stages may be used, with other findings, to date the emergence of language. If important aspects of language cannot appear until sexual maturity, as we propose, then a second consequence is that the development of language requires the whole of modern human ontogeny. Our life history model thus offers new ways of investigating, and thinking about, the evolution, development, and ultimately the nature of human language.