A central question in behavioural and evolutionary ecology is to understand how animals make decisions between, for instance, potential mates, nesting sites, foraging patches and territories. Normative models of choice usually predict preferences between alternatives by computing their value according to some criterion and then identifying the alternative with greatest value. An important consequence of this procedure is captured in the economic concept of rationality, defined through a number of principles that are necessary for the existence of a scale of value upon which organisms base their choices. Violations of rationality are nonetheless well documented in psychological and economic studies of human choice and consumer behaviour, and have forced a reinterpretation of much of the existing data and models. Although largely unexplored in the study of animal decision-making, the systematic observation of irrationality would similarly pose serious challenges for functional approaches to behaviour. In this thesis I explore the possibility that violations of rational axioms may also be found in animal choices, using the European starling (Sturnus vulgaris) as a model species. My objectives were threefold. Firstly, I investigated the prevalence of rationality across distinct foraging paradigms, in situations involving multialternative choices, structured choice sets, choices between alternatives described by multiple attributes and risk-sensitive decisions. In a number of distinct experiments, the preferences of the starlings were consistent and stable across contexts, conforming to basic rational principles such as transitivity and regularity. A second objective was to explore possible factors underlying reported violations of rational axioms by animals. Amongst potential mechanisms, I review and examine the implications of the use of hierarchical and higher order choice rules, as well as the presence of constraints on the perception of rewards. Finally, I examine the likely effect of contextual changes on an organism's state, and consequently choice behaviour, and experimentally confirm the expectation that statedependence in foraging preferences can underlie the observation of seemingly irrational behaviour. Altogether, my results suggest that, rather than being a common phenomenon, breaches of rationality in animals might be restricted to specific sets of parameters and conditions. They also emphasize the importance of considering the potential multitude of factors underlying violations of rationality in animal choices, and suggest that students of economic rationality in animal behaviour should also view preferences as a dynamic, statedependent measure.

One basic requisite for rationality is that choices are consistent across situations. Animals commonly violate rationality premises as evidenced, for example, by context-dependent choices, and such apparent irrationalities stand as paradoxes that instigate re-examination of some assumptions in behaviour ecological modelling. The goal of the present thesis was to study the psychological mechanisms underlying apparent irrationalities in order to assess the functional implications of general processes of valuation and choice. A common thread through the different studies is the hypothesis that most animal 'irrationalities' are due to misinterpretation of what the optimum would be in natural circumstances, and hence of the maximised currency in the theoretical predictions. I believe that the trait that may have been of paramount influence in many organisms' selective history was the ability to learn about the predictability of events and their biological value, and that this is implemented in an overriding force of associative learning mechanisms. In chapters 2 and 3, I present evidence of context-dependent foraging choices in European starlings, Sturnus vulgaris, in the laboratory, and I implement a version of the Rescorla- Wagner learning model to account for both present data and apparent irrationalities reported by other authors. In chapter 4, I test the notion that context dependence may in fact be adaptive when animals face sequential choices, namely when they have to decide whether to take a prey item or to skip it in order to search for better alternatives. In chapter 5, I explore the functional implications of starlings' relative responding to incentives during an unexpected shortfall in reinforcement, and I also examine the extent to which information about the new environmental status helps them avoid energetic and time costs commonly seen in uninformed individuals. Last, in chapter 6, I present a brief summary of the main discussions considered and conclusions reached along this thesis.

From cells to societies, cooperation occurs at all levels of biological organization. In vertebrates, the most complex societies occur in cooperative breeders where some group members (helpers) forego reproduction, sacrificing their immediate direct fitness to assist in raising the offspring of others (breeders). Individuals in cooperative breeding societies can gain indirect fitness benefits from passing on shared genes when they help the offspring of close genetic relatives (kin selection), such that cooperation is expected to correlate with genetic relatedness. However, some cooperatively breeding societies include cooperation between nonrelatives. Cooperatively breeding societies range in complexity, from singular (one breeding pair) to plural (two or more breeding pairs). In the majority of singular breeding societies, helpers are relatives of breeders. Thus, kin selection is thought to underlie helping behavior in singular breeding societies. Plural breeding societies, such as in superb starlings (Lamprotornis superbus) inhabiting the East African savanna in central Kenya, involve multiple territory-sharing families raising offspring with helpers who can assist more than one family simultaneously. The superb starling’s complex and dynamic social system, mixed kin structure, relatively long lives, and stable social groups make them an ideal study species for investigating how patterns of individual decision-making have shaped and maintained cooperative societies. My dissertation research focuses on using long-term data on cooperatively breeding superb starlings to explore how temporally variable environments, such as the East African savanna, influence individual decisions across lifetimes, and subsequently how individual behavior shapes the structure and organization of the society. In Chapter 1, I apply a Bayesian approach to the animal model to estimate how genetic versus nongenetic factors influence among individual variation in the social roles: “breeder”, “helper”, and “non-breeder/non-helper”. Non-breeder/non-helper indicates that the individual maintained membership in the social group but did not breed or help during that season. I then estimated heritability and found, as predicted, overall low heritability of traits responsible for each role. This result is consistent with the findings of other studies on the heritability of social behavior, which tends to be low compared to non-social traits, primarily because the social behavior of an individual is highly influenced by interactions with other individuals. In Chapter 2, I show that superb starlings (i) are nepotistic, and (ii) switch between the social roles of “helper” and “breeder” across their lives. This role switching, which unexpectedly includes breeders going back to helping again, is linked to reciprocal helping between pairs of helpers and breeders, independent of genetic relatedness. Reciprocal helping was long thought to be irrelevant for cooperative breeders because most helping is assumed to be unidirectional, from subordinate helpers to dominant breeders, and reciprocal helping is often measured on short timescales. These long-term reciprocal helping relationships among kin and nonkin alike may be important for the persistence of this population because previous research has demonstrated that enhancing group size by immigration from outside groups, while reducing group kin structure, is necessary to prevent group extinction. Finally, the results of Chapter 3 reveal how social and ecological factors shape role switching across individual lifetimes. Overall, my dissertation highlights the remarkable flexibility of superb starling cooperative behavior and the crucial role of mutual direct fitness benefits from reciprocal helping, which may help promote the stability of cooperative group living among nonkin as well as kin group members, contributing to the resilience of this population within a harsh and unpredictable environment.