The evolution and function of play

Abstract

The question of why some animals play has received serious scientific attention since Herbert Spencer (1872) published the first theory of the origin of play in his The Principles of Psychology. Spencer thought that play occurs when an animal builds up excess energy in its brain, which is then discharged as frivolous play behavior that imitates the animal’s functional behaviors. Groos (1898), in the first book dedicated to animal play, argued that Spencer’s imitative theory could not explain why young animals play because they lack adult functional behaviors to imitate. Spencer then argued that a more plausible version of the excess energy theory would assert that frivolous play behaviors are instead instinctive behaviors discharged when excess energy is sufficiently high. Groos, however, did not believe that any excess energy theory of frivolous play was empirically or theoretically adequate. Empirical observations suggested that young animals play even when fatigued and therefore do so while lacking excess energy. Theoretically, Groos could not accept the idea that play is merely frivolous; play must be favored by natural selection in order to evolve. Groos instead proposed an adaptive theory of play, which postulated that animals inherit partial or incomplete instincts. Play by young animals allows them to practice incomplete instinctive behaviors, which over developmental time become adult functional behaviors. Thus, in Groos’ adaptive theory of play, the function of play is practice and its adaptive benefit is the development of adult functional behaviors. Today, although the details of both theories are not seriously entertained, the broad perspectives of frivolous versus adaptive play still frame the debate. In 2010, Gordon Burghardt and Sergio Pellis proposed a series of workshops on the evolution of play, which were sponsored by the National Institute of Mathematical and Biological Synthesis (NIMBioS) and held from 2011 to 2013. The aim of these NIMBioS workshops was to develop computational and mathematical models of the evolution and mechanisms of play with the aim developing a clearer and integrative understanding of the conditions under which play could evolve. The papers in this issue represent the first systematic use of computational and mathematical models to investigate the theoretical and empirical origins of play. Pellis, Palagi, Burghard, and Mangle, in this issue, present a conceptual framework for understanding the evolution and mechanisms of play. A problem for purely adaptive approaches to the evolution of play is that play is sparsely distributed across animal phyla. Play has been identified in only five of 30 different phyla and most species that play are mammals (Burghardt, 2005). If play as practice is essential for completing or fine-tuning adult functional behaviors during development, then play should be more common across phyla. In part to address this problem, they distinguish three types of play processes: primary, secondary, and tertiary. Primary process play is the most basic and expressed during an animal’s development, it just satisfies criteria for play, and may not be adaptive. Once primary process play has evolved, it may be evolutionarily co-opted to serve adaptive functions in secondary and tertiary process play. The distinction between these different processes or levels of play integrates essential elements of the two nineteenth century views on the origin of play previously discussed. When there is resource abundance (i.e., surplus resource theory; SRT; Burghardt, 1984, 2005), primary process play may initially evolve without adaptive benefit. Once established, selection can co-opt primary process play to transform it into adaptive secondary and tertiary process play, which have adaptive benefits. This suggests that play is phylogenetically rare because initially it is typically not adaptive and requires resource abundance to evolve. Alternatively, primary process play could be adaptive as argued by Spinka, Newberry, and Bekoff (2001). They theorized that primary process play functions to build a repertoire of behavioral responses to unexpected events. Interestingly, Cully, Clune, Tarapore,