Discussions of altruism occur in three importantly different contexts. During the past four decades, evolutionary theory has been concerned with the possibility that forms of behaviour labelled as altruistic could emerge and could be maintained under natural selection. In these discussions, an agent A is said to act altruistically towards a beneficiary B when A's action promotes the expected reproductive success of B at expected reproductive cost to A. This sort of altruism, biological altruism, is quite different from the kind of behaviour important to debates about ethical and social issues. There the focus is on psychological altruism, a notion that is concerned with the intentions of the agent and that need have no connection with the spread of anyone's genes. Psychological altruists are people with other-directed desires, emotions or intentions (this is a rough preliminary characterization, to be refined below). Finally, in certain kinds of social scientific research, the important concept is that of behavioural altruism. From the outside, behavioural altruists look like psychological altruists, although their motives and preferences may be very different.

The adequacy of currently popular accounts of the genetic basis for psychological altruism, including inclusive fitness (kin selection), reciprocal altruism, sociality, and group selection, is questioned. Problems exist both with the evidence cited as supporting these accounts and with the relevance of the accounts to what is being explained. Based on the empathy-altruism hypothesis, a more plausible account is proposed: generalized parental nurturance. It is suggested that four evolutionary developments combined to provide a genetic basis for psychological altruism. First is the evolution in mammals of parental nurturance. Second is the evolution in humans (and possibly a few other species) of the ability to see others as sentient, intentional agents and, thereby, to recognize other's needs, even subtle ones. Third is the evolution in humans of tender, empathic emotions as an important component of parental nurturance. Fourth is the evolution in humans of cognitive capacities that make it possible to generalize tender, empathic feelings and, thereby, altruism beyond offspring.

In this paper, we examine different forms of altruism. We commence by analysing the ‘behavioural’ definition and, after clarifying its conditions for altruism, we argue that it is not in ‘reflective equilibrium’ with everyday linguistic usage of the term. We therefore consider a ‘psychological’ definition, which we likewise refine, and argue that it better reflects ordinary language use. Both behavioural and psychological approaches define altruism descriptively and thus fail to capture an important aspect of altruism, namely its normative component. Altruism, we argue, is a ‘thick concept’, i.e. one which embodies both positive and normative components. We discuss and compare various formulations of this normative component.

Much of economics is built on the assumption that individuals are driven by self-interest and economic development is an outcome of the free play of such individuals. On the few occasions that the existence of altruism is recognized in economics, the tendency is to build this from the axiom of individual selfishness. The aim of this paper is to break from this tradition and to treat as a primitive that individuals are endowed with the ‘cooperative spirit’, which allows them to work in their collective interest, even when that may not be in their self-interest. The paper tracks the interface between altruism and group identity. By using the basic structure of a Prisoner's Dilemma game among randomly picked individuals and building into it assumptions of general or in-group altruism, the paper demonstrates how our selfish rationality interacts with our innate sense of cooperation. The model is used to outline circumstances under which cooperation will occur and circumstances where it will break down. The paper also studies how sub-groups of a society can form cooperative blocks, whether to simply do better for themselves or exploit others.

Egoism and altruism are unequal contenders in the explanation of human behaviour. While egoism tends to be viewed as natural and unproblematic, altruism has always been treated with suspicion, and it has often been argued that apparent cases of altruistic behaviour might really just be some special form of egoism. The reason for this is that egoism fits into our usual theoretical views of human behaviour in a way that altruism does not. This is true on the biological level, where an evolutionary account seems to favour egoism, as well as on the psychological level, where an account of self-interested motivation is deeply rooted in folk psychology and in the economic model of human behaviour. While altruism has started to receive increasing support in both biological and psychological debates over the last decades, this paper focuses on yet another level, where egoism is still widely taken for granted. Philosophical egoism (Martin Hollis’ term) is the view that, on the ultimate level of intentional explanation, all action is motivated by one of the agent's desires. This view is supported by the standard notion that for a complex of behaviour to be an action, there has to be a way to account for that behaviour in terms of the agent's own pro-attitudes. Psychological altruists, it is claimed, are philosophical egoists in that they are motivated by desires that have the other's benefit rather than the agent's own for its ultimate object (other-directed desires). This paper casts doubt on this thesis, arguing that empathetic agents act on other people's pro-attitudes in very much the same way as agents usually act on their own, and that while other-directed desires do play an important role in many cases of psychologically altruistic action, they are not necessary in explanations of some of the most basic and most pervasive types of human altruistic behaviour. The paper concludes with the claim that philosophical egoism is really a cultural value rather than a conceptual feature of action.