The first part of this thesis consists of three chapters outlining a theory of the emergence of linear dominance hierarchies. The first chapter provides the necessary background and lays out the argument for the approach I have taken which is to make a mechanistic model, embed it within an evolutionary framework, and then to begin to test it using a model species. In the second chapter, I present the model of decision-making during animal contests which lies at the core of the theory. It is based on the sequential probability ratio test which is a standard decision-making mechanism. I outline the predictions of the model and in particular draw attention to the speed-accuracy trade-off faced by animals as they contest resources. In the third chapter, I extend this model to incorporate the use of prior information during decision-making. Individuals observe fights between other group-members and use the evidence they acquire from these to inform their decisions in future contests. The use of prior information in this way explains both the ‘bystander effect’, whereby individuals are more likely to lose against others they have seen win and vice versa, and the emergence of linear hierarchies at the level of the group. Armed with a model of how hierarchies form, I am then able to make novel predictions about how resource distributions affect their emergence and to suggest methods for manipulating the level of equality in groups.
The second part of this thesis is devoted to developing a model species for testing the theory outlined in the first part. I focus on the zebrafish (Danio rerio) for three reasons: first, it is such a well-established model system that tools exist to test the predictions of the model at the neural level; second, the fish are small enough to keep in the laboratory in large numbers and third, they have relatively few degrees of physical freedom, which simplifies the classification of its behavior. The first step to testing the theory is to establish that the behavior of the animals in individual contests conforms to the predictions of the assessment model. The methodology for testing between models of contest behavior has lagged behind developments in model selection, so in the first chapter of this part of the thesis I illustrate how this technique can simply and accurately discriminate between two different mechanisms of decision-making. In order to reach the widest possible audience I make use of the predictions of two existing models of contests, the war of attrition and the sequential assessment model, rather than my own version of the mutual assessment model.
The final part of this thesis is devoted to relaxing the assumption that animals have only a single currency in which they can invest in the accuracy of their decisions. In reality, the costs of accuracy come from many different sources. Some of these sources, such as predation risk, may vary in magnitude in a fashion uncorrelated with others, such as the cost of energetic expenditure: the cost of a minute of information gathering may rise in the presence of a snake or when you are hungry, but these two may not be correlated with one another. I use a simple model based on the sequential probability ratio test to show that when an animal has two currencies available, which I call time and effort, it should substitute between them as their relative costs vary. An animal with this ability gains an advantage along both axes of the traditional speed-accuracy trade-off, making more accurate decisions at a lower cost. I outline a set of testable predictions that would allow researchers to uncover the trade-off between speed, effort and accuracy in action.
The three parts of this thesis together form a coordinated attempt to advance an information-based theory of fighting and dominance in animal societies based on the way in which animals make their decisions. I suggest that of all the factors affecting contest outcomes, it is what animals learn about one another that is the most important for the emergence of social hierarchies. This is an old idea, but one whose time has now come with the advent of more powerful approaches to modeling decision-making, and I hope that it will eventually lead to a predictive theory of social structure firmly grounded in ecology and evolution. (Abstract shortened by UMI.)