Rachlin's argument rests on his definition of an altruistic action. Three main features characterize this definition: An altruistic act (1) always has a negative value, (2) is a subset of self-controlled actions, and (3) is meaningful only in the context of repeated interactions. All three features are highly questionable.

Altruism and cooperation are explained as learned behaviors arising from a pattern of repeated acts whose acquired value outweighs the short-term gains following single acts. But animals and young children, tempted by immediate gains, have difficulty learning behaviors of self-control. An alternative source of reinforcement, shared by animals and humans, arises from social interaction that normally accompanies cooperation and altruism in nature.

Internal mechanisms, especially those implicating the self, are crucial for the egoism-altruism debate. Self-liking is extended to close others and can be extended, through socialization and reinforcement experiences, to non-close others: Altruistic responses are directed toward others who are included in the self. The process of self-extension can account for cross-situational variability, contextual variability, and individual differences in altruistic behavior.

I argue that Rachlin's notion of self-control is imprecise and not well suited to the discussion of altruism. Rachlin's broader agenda, to improve collective welfare by identifying behavioral mechanisms that increase altruism, neglects the fact that altruism is neither necessary nor sufficient for desirable social outcomes.

We agree with Rachlin's aim to account for altruism within existing theory. However, his argument is implicitly dependent on social and cognitive constructs that are explicitly identified in other social-psychological theories. The account does not advance theory beyond available constructs (e.g., self-categorizations, motives, values, role-identities, and social structure), and Rachlin's implicit use of these strains the behaviorist account.

Rachlin provides an impressive integrative view of altruism and selfishness that helps us correct older views. He presents a highly general theory, even though he is aware of context-dependence of key notions, including altruism. The context-dependence should extend much farther than Rachlin allows it to go. We had better replace theoretical notions of altruism and selfishness by common sense.

As support for his position, Rachlin refers to the writings of Aristotle. However, Aristotle, like many social psychological theorists, would dispute the assumptions that altruism always involves self-control, and that altruism is confined to acts that have group benefits. Indeed, for Aristotle, as for equity theory and sociobiology, justice exists partly to curb the unrestrained actions of those altruists who are a social liability.

On a proximate level, altruism may well be a temporally extended pattern of behavior that often seems to be maintained without extrinsic rewards (we may find it just valuable to be an altruistic person). However, recent theory and experiments have uncovered significant and often nonobvious extrinsic rewards for “altruistic” behavioral patterns. Ultimately, these patterns may mostly lead to a net benefit.

Rachlin favors following patterns over making decisions case by case. However, his accounts of self-control and altruism do not establish the rationality of making decisions according to patterns. The best arguments for using patterns as a standard of evaluation appeal to savings in cognitive costs and compensation for irrational dispositions. What the arguments show depends on how they are elaborated and refined.

Rachlin's substantive points about the relationship between altruism and self-control are obscured by simplistic and outdated portrayals of evolutionary psychology in relation to learning theory.

It is assumed that self-control always has a higher value. What if it does not? Furthermore, although there are clearly intrinsic reinforcers, their measurement is problematic, especially for a behavioral analyst. Finally, is it more parsimonious to postulate that these behaviors are acquired rather than genetically based?

Rachlin rightly highlights behavioural reinforcement, conditional cooperation, and framing. However, genes may explain part of the variance in altruistic behaviour. Framing cannot be used to support his theory of altruism. Reinforcement of acts is not identical to reinforcement of patterns of acts. Further, many patterns of acts could be reinforced, and Rachlin's altruism is not the most likely candidate.

The new emphasis that Rachlin gives to social learning is welcome, because its role in the emergence of altruism and communication is often underestimated. However, Rachlin's account is underspecified and therefore not satisfactory. I argue that recent computational models of the evolution of language show an alternative approach and present an appealing perspective on the evolution and acquisition of a complex, altruistic behavior like syntactic language.

Some commentators have argued that all particular altruistic acts are directly caused by or reinforced by an internal emotional state. Others argue that rewards obtained by one person might reinforce another person's altruistic act. Yet others argue that all altruistic acts are reinforced by social reciprocation. There are logical and empirical problems with all of these conceptions. The best explanation of altruistic acts is that – though they are themselves not reinforced (either immediately, or delayed, or conditionally, or internally) – they are, like self-controlled acts, part of a pattern of overt behavior that is either extrinsically reinforced or intrinsically reinforcing.