International Society for History, Philosophy, and Social Studies of Biology

twitter 2015
     facebook 2015


TUESDAY, JULY 7  /  15:30 - 17:00  /  DS-M240
Organized session / standard talks
Evolutionary modeling in biology and the behavioral sciences (2): Games and limiting assumptions

Cailin O'Connor (University of California, Irvine, United States)

Evolutionary modeling is used widely in biology and the social sciences. In recent years, this practice has spread in philosophy, and philosophy of biology in particular. Given the increasing use of evolutionary modeling in philosophy of biology, it is appropriate to ask: what are the limits of this sort of modeling? In what ways does it go wrong? In what ways can evolutionary models be misleading? At the same time, how can this be a useful methodology? How does it provide insight and explanation? This session will explore this methodology focusing on limiting assumptions of evolutionary models, including base games in evolutionary game theory.

Studying the real origins of communication and language: Models and experiments

Thom Scott-Phillips (Durham University, United Kingdom)

In proof-of-concept models, verbal theory is expressed in a formal way to establish exactly what conclusions follow (or not) from a given set of assumptions. It is widely appreciated that such models provide explanations only to the extent that the assumptions that go into them are reasonable: poor assumptions make for poor models. What is arguably less appreciated is that a similar point applies also to the explananda of such models: any lack of clarity about what is to be explained will lead to a lack of clarity about what a model can tell us. One clear illustration of this comes from the literature on the evolutionary emergence of communication. Although many models and experiments purport to study this emergence, they often fail to do so because, with few exceptions, they all build some form of communication into their initial specifications. Consequently, they in fact study how communication systems transition from one form to another, rather than how communication itself emerges in the first place. I will present research – both a recent computational model and an older experimental game – that directly addresses this problem (Blythe & Scott-Phillips, 2014; Scott-Phillips et al., 2009). I will explain what this combination of models and experiments can tell us about the origins of human communication, and in so doing I will illustrate why conceptual clarity about explananda is both important and useful, especially in evolutionary modelling.

Social behavior and evolving games

Rory Smead (Northeastern University, United States)

Models of the evolution of social behavior typically treat the external environment and the nature of the social interaction as fixed. In evolutionary game theory, most models focus is on how behavior evolves within a fixed game. However, some of the most important and significant evolutionary changes to social interaction involve a change of the game, not simply a change in behavior. Niche construction, for example, involves environmental modifications that may fundamentally alter social interactions (Laland et. al. 2000). Alternatively, signals and recognition ability can introduce previously unavailable strategies strategies and stabilize otherwise disadvantaged behavior (West Gardner 2010). Representing evolving games is necessary for a more complete view of the evolution of social behavior. Although there is relatively little work done on how to formally represent an evolving game, recent work has begun to provide some methods for doing so (e.g., Wordin Levin 2007, Akcay and Roughgarden 2011, Smead 2014). Building on previous work, I will provide a method of representing evolving games that can be largely formulated within traditional evolutionary game theory. I will then explore applications of this framework to the two cases previously mentioned niche construction and the co-evolution of social behavior and type-recognition ability.

Strategic thinking in biology

Cailin O'Connor (University of California, Irvine, United States)

Because game theory was developed to model strategic behavior, `games' are designed to be explicitly strategically interesting. As I plan to argue, however, there are cases in biology that lack strategic interest, and where, as a result, the use of game theory to model them has proved misleading. In this talk I offer two cases of this sort. The stag hunt is a model of cooperation under risk. The payoffs in the stag hunt are restricted so that the cooperative equilibrium is the payoff dominant one, while the non-cooperative equilibrium is what is called `risk dominant'. Without this restriction, the game is not necessarily interesting because it will be expected that actors always choose to cooperate. The stag hunt, restricted this way, evolves to the non-cooperative outcome the majority of the time in standard evolutionary models. Evolutionary game theorists have taken this observation as reason to investigate further mechanisms for the evolution of cooperation in the stag hunt. In real biological scenarios, however, there is absolutely no reason to think that the payoffs for cooperation are restricted as they are in the stag hunt game. This means that the motivating evolutionary puzzle for this work on the stag hunt is a spurious one. A second case relates to conflict of interest signaling in biology. Evolutionary game theorists claim that total conflict of interest signaling will not evolve. While this claim is largely correct when one considers only interactions between two actors in a single scenario, conflict of interest signaling in biology often occurs in the context of a larger strategic interaction. For example, organisms in the process of signaling to potential mates also signal to potential predators. In these two cases strategic limitations of games prevent them from appropriately modeling the biological scenarios they would seem to best represent.