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


twitter 2015
     facebook 2015

Program

TUESDAY, JULY 7  /  09:00 - 10:30  /  DS-M440
Organized session / standard talks
Costly signaling theory: Experimental and theoretical perspectives
Organizer(s):

Justin Bruner (Australian National University, Australia); Hannah Rubin (University of California, Irvine, United States)

One of the biggest puzzles in evolutionary theory pertains to how honest communication is possible when the interests of sender and receiver do not completely overlap. Such ‘partial conflict of interest’ settings are commonplace in nature, as it is often the case that an informed sender has reason to conceal information from an ignorant receiver. Yet what ensures signal reliability? In the past few decades the handicap principle has emerged as the dominant explanation. The handicap principle states that honesty is possible if there are significant costs associated with sending a signal. The popularity of the handicap principle is evident by its common use in not just biology, but a variety of related disciplines as well. However, the handicap principle has been called into question on both theoretical and empirical grounds. Modeling work has demonstrated that perfect information transfer is not guaranteed – uninformative arrangements in which little or no information is transferred from sender to receiver are real evolutionary possibilities. On the empirical side, experiments designed to measure the cost of signaling have often been unable to register the high costs stipulated by the handicap principle. For these and other reasons, many have urged that we must look beyond the handicap principle if a defensible explanation of honesty is to be attained. In this symposium we explore a variety of alternatives to the handicap principle that can ensure honesty, or at least partial honesty, in signaling interactions. Additionally, the talks in this symposium showcase both theoretical and empirical approaches to the study of animal communication, and so nicely illustrates how these two perspectives can support and mutually inform each other.


Cheaper than costly signaling in the lab

Hannah Rubin (University of California, Irvine, United States); Cailin O'Connor (University of California, Irvine, United States); Justin Bruner (Australian National University, Australia); Simon Huttegger (University of California, Irvine, United States)

An alternative ‘hybrid’ equilibrium, in which communication is partially honest, has been shown to be at least as plausible as the ‘separating’ equilibrium described by the handicap principle, in which communication is completely honest. This equilibrium has been known in economics for some time, but the evolutionary significance has only recently been considered (Zollman et. al 2013) We use an experimental approach to evaluate the evolutionary plausibility of the hybrid equilibrium. We use the methods of experimental economics because various learning models which describe the way subjects learn in the laboratory setting are equivalent to a form of the replicator dynamics with perturbations. (Hopkins 2002). We compare an experimental treatment (in which the cost associated with signaling as a low type allows for the hybrid equilibrium) with a control (where the signal cost is high enough that we would expect a separating equilibrium). We also compare the experimental treatment with a null hypothesis of no information transfer. We find that the theoretical predictions are confirmed: populations do evolve toward the hybrid equilibrium.


Cost, expenditure and vulnerability

Justin Bruner (Australian National University, Australia); David Kalkman (Australian National University, Australia); Carl Brusse (Australian National University, Australia)

Conventional wisdom holds that signals can only be reliable in conflict of interest settings if there are significant costs associated with signaling. In other words, the act of signaling must decrease some component of fitness of the sender. Yet a number of empirical studies have failed to register the existence of such costs. Moreover, empirical studies of animal communication often do not directly measure cost but merely the expenditure of time or energy associated with signaling. This vital link between expenditure and cost has only recently been explicitly discussed in the literature, and in the course of this paper we consider one natural way of connecting expenditure to cost. In particular, we assume that whether an expenditure is indeed costly is in large part receiver-dependent, and we draw upon a variety of predator-prey signaling systems which appear to have this structure. Namely, the expenditure associated with signaling only counts as a real fitness cost if the receiver (i.e. the predator) gives chase. In this case the expenditure of time and resources associated with signaling increases the vulnerability of the sender, thereby decreasing the probability of survival (conditional on the receiver pursuing them). We formalize this scenario as a signaling game and demonstrate perfect information transfer is no longer possible. Instead, a partially honest equilibrium similar to the so-called hybrid equilibrium is the most likely evolutionary outcome.


Costly signaling in finite populations

Elliott Wagner (Kansas State University, United States)

Signaling games have been used to explain the evolution of information transfer in systems as diverse as bacterial communication and human language. Almost all signaling games are grounded in the assumption that the interests of the communicators are aligned to such an extent that honest communication constitutes a Nash equilibrium of the game. Here I will demonstrate that this assumption is not necessary. Finite-population evolutionary dynamics will with high-probability lead to honest signaling in a variety of games in which honest signaling is not a Nash equilibrium. This will be demonstrated for three distinct families of games: the persuasion game (Milgrom and Roberts, 1986), the Sir Philip Sidney game (Maynard Smith, 1991), and the Spence’s signaling game (Spence, 1973). In each of these games finite-populations can evolve honest communication even when messages are costless. This result casts doubt on Zahavi’s (1975) well-known handicap principle, which claims that when the interests of sender and receiver conflict, reliable signaling is only possible if there are significant costs to producing a signal. When costs are added to a signal, so that honest communication is a Nash equilibrium of the game, finite-population evolutionary dynamics only lead to honest communication when it payoff-dominates the non-communicative pooling equilibrium in which the cheaper signal is the only signal sent. This shows that in games with costly signals it is not enough for signaling to be a Nash equilibrium. For communication to be a likely evolutionary outcome it must be both a Nash equilibrium and it must payoff-dominate pooling. This result demonstrates in a dynamic model a common-sense hypothesis that has sometimes been overlooked in the signaling literature: costly signaling should not be expected to evolve if communication leaves both parties worse of than they would be pooling.