Reasons without Persons is a brilliant and suggestive book. It will arouse enthusiasm in disciples and critics of Parfit alike, and more generally in all those who study the concept of rationality, including philosophers and economists. Despite the complexity of the topic, the book is accessible, including to those who have never read anything on the subject before. All one needs is logic and reason. The book is divided into 11 chapters, with each chapter short enough to be read in one sitting. Each one is, however, rich, dense and thought provoking.

It is natural to believe that agents are rationally required to coordinate with themselves in a particular way, whilst they are rationally required to coordinate with other people in a different way (2). It is not irrational, for instance, for two distinct agents to have antagonistic preferences whilst it is irrational for a single agent to have such preferences. In Reasons without Persons, however, Hedden claims that we are wrong to hold different rational standards for inter-individual and intra-individual preferences, beliefs and actions (2). This claim is not new and has been notably defended by Parfit (Reference Parfit1984), but a considerable literature has since developed to further challenge or defend this view. Hedden's book is a good illustration of the evolution of the debate over the last 30 years. References to Parfit are scarce, but his influence is implicit throughout. Parfit's psychological approach to what matters in personal identity, in particular, is not discussed but is tacitly admitted. More importantly, Hedden addresses rationality issues in a form more consistent with the discourse of modern economics. Most of Parfit's idiosyncratic language is consequently dropped.

Hedden embraces an impersonal view of agency, in which agents are reduced to a set of temporally located decision-making units, whose rationality is assessed independently of each other. Hedden calls this Time-Slice Rationality. It rests upon the combination of two basic ideas. First, the locus of rationality should be assigned not to temporally extended agents but rather to particular temporally located selves. (This principle is referred to as Synchronicity; a synchronic approach to rationality means that what you rationally ought to do at a time directly depends only on your mental states at that time (7).) Second, each of these time-sliced decision-making units should be considered as just as independent from each other as normal agents are. (This principle is referred to as Impartiality.)

Hedden's approach rejects the Principle of Conditionalization, according to which changes of your credence or preferences over time are rational only to the extent that they appropriately respond to a change of information or situation. The Principle of Conditionalization is consistent with the general intuition that it is irrational for the same person to change beliefs or preferences without reason and to act accordingly. As noticed by Hedden, Conditionalization is not an impersonal principle (31). It implicitly considers agents as physically and psychologically continuous units, whose past, present and future mental states are all functionally related. Hedden's replacement of Conditionalization by Synchronicity, together with Impartiality (the principle that the attitudes you have at other times in your life play the same role as attitudes other people have (9)) implies that the rationality of each time-slice centred unit of a person is to be assessed independently of the mental states the agent may have had at any other time. It also implies that the rationality of such a unit is assessed independently of who they are.

A further purported implication of Hedden's view is the rejection of the so-called Principle of Reflection, which commands you to defer to the rational beliefs you anticipate having in the future (56). The Principle of Reflection implies, for instance, that it is irrational for you to start smoking if you anticipate that you will later regret having started.

I am overall very sympathetic to Hedden's original work. In what follows, I shall not question the validity of Hedden's main arguments but rather some of their theoretical assumptions and implications. More specifically, I propose to investigate the purported reasons why identical rational requirements should apply both within and across agents. Hedden claims that this is a logical consequence of Synchronicity together with Impartiality. I shall argue that the plausibility of Impartiality depends on the acceptability of Impersonality. It is therefore of primary importance for Hedden to justify Impersonality, but he unfortunately fails to do so.

The importance of Impersonality for Hedden is obvious from the title Reasons without Persons. Echoing Parfit's Reasons and Persons, it clearly states a willingness to continue Parfit's reflection on rationality while abandoning any reference to personal identity. Of course, Parfit himself rejected traditional views on personal identity and their importance in relation to rationality. What really mattered, for Parfit, was the ‘Relation R’ composed of psychological continuity and connectedness between temporal selves that one could maintain with one's other temporal selves as well as with other people. A main implication of Parfit's account was that one can be relevantly related to someone else now to the same degree that one is with one's future self. Hedden's intention is to go beyond R-relatedness to full impersonality. Why impersonality? Hedden justifies his choice by stating that ‘personal identity is so messy and problematic that it should play no important role in the theory of rationality’ (15). Debates on personal identity are indeed messy and problematic, but so are debates on rationality. I suspect Hedden had another reason. It is because he wanted to embrace impersonality that he had to avoid discussing the metaphysics of identity, and not the other way around. Impersonality is not an incidental feature of his theory. It is an essential one, as it justifies the Principle of Impartiality. And, whilst Parfit's Relation R between temporal selves significantly weakens as these selves are further apart in time, it does not justify treating them as independent. Relation R comes in degrees, when impersonality does not (33). Moreover, while Parfit's argument that Relation R is what matters lessened the difference between one's temporal selves and other agents, it did not annihilate it. The problem with Hedden's view is that the absence of a theoretical or empirical justification for doing away with persons altogether undermines his claim that ‘the locus of rationality, as it were, is not the temporally extended agent, but rather the time-slice’ (7). Why should one assess individual rationality from an impersonal point of view if nothing actually supports the idea that agents are, or behave as if they were, merely an aggregation of time slices? The only possible answer is, of course, Impartiality considered as a moral (and not as a rational) principle.

This leads me to my second point, the relationship between morality and rationality. In ‘General Motivations’, the second chapter of his book, Hedden clearly explains his intention to address rationality outside of any moral considerations. Time-Slice Rationality, he points out, is consistent with many moral theories, Utilitarianism being just one of them (22). As a consequence, its two founding principles, Synchronicity and Impartiality, ought to be morally neutral. I already touched upon the importance of justifying Impartiality from a moral point of view. A similar claim applies to Synchronicity. Hedden believes it is possible to undermine diachronic principles (such as the principles of Conditionalization and Reflection) from a strictly rational point of view. There are, I agree, indeed no reasons for my preferences to be functionally related over time. Why should I only be allowed to change my beliefs and preferences according to changes of situation or information, and not, for instance, according to my own personal development? The difficulty is, of course, to be able to distinguish changes that are rationally acceptable from those that are not. Changing one's mind minute by minute, like Hedden's Fickle Frank (2), is obviously irrational. Yet changing one's political views between the age of 18 and the age of 60, like in Parfit's example of the Russian nobleman, is not. So how can one, using diachronic principles, tell apart acceptable preference changes? Which ones are rationally permissible and which are not?

Hedden's proposal is to distinguish two sorts of preferences, ‘a more fundamental preference, which is stable, and another, less fundamental preference which changes in response to information’ (45). Fundamental preferences, also called ultimate preferences, ought never to change. Hedden calls this requirement of rationality Utility Conditionalization. Utility Conditionalization is consistent with a number of preference changes, such as changes in taste-based preferences, so long as new taste-based preferences concur with ultimate preferences (e.g. preferring pleasure to pain).

The problem for Hedden is that Utility Conditionalization is ‘incompatible with the plausible claim that it is rationally permissible to be time-biased’ (50). Because of the passing of time, says Hedden, if you are biased towards the future (i.e. you prefer past to future pains or future to past pleasures), ‘your preferences over maximally specific possibilities will sometimes shift, in violation of Utility Conditionalization’ (50). He claims that ultimate preferences are actually changed by hyperbolic discounting but that Utility Conditionalization rightly rules out this time bias as irrational (51n). He concludes that exponential discounting is ‘the only sort of time-bias compatible with Utility Conditionalization’ (53) but disputes its rational legitimacy. His arguments, however, are not wholly convincing. Most economists would regard as wrong his statement that exponential discounting is undermotivated (54). His reference to Dougherty's case of predictable exploitation over time under exponential discounting (Dougherty Reference Dougherty2011) is also misleading, as Dougherty's reasoning does not and (I believe) cannot apply to ultimate preferences. More importantly, I am not convinced that his attempt to simultaneously vindicate the ‘plausible’ rationality of the bias towards the future (but not to the near) and to dispute exponential discounting turned out to be a good strategy to discredit diachronic principles. Diachronic principles could have been more readily dismissed by appealing to Impersonality, but Hedden's decision to exclude any moral and/or metaphysical considerations deprived him of this possibility.

My final criticism relates to Hedden's attempt to establish a criterion of rationality based on synchronic, rather than diachronic, principles. Hedden suggests that we should abandon the principles of Conditionalization and Reflection in favour of a synchronic principle, which he calls Uniqueness, which is meant to ‘avoid the need for diachronic principles while still respecting the datum that wildly fluctuating beliefs are (ceteris paribus) irrational’ (129). According to the Principle of Uniqueness, given a body of evidence, there is only one doxastic state that it is rational to be in (130). A similar principle, called Preference Uniqueness, states that, given a body of total evidence, there is a unique set of preferences that is rational to have (149). Rationality is not determined by the stability of an individual's preferences or beliefs over time but by the fact that, at each period, her preferences and beliefs are ‘objectively’ justified. This is, of course, controversial, since it implies that, given the same shared evidence, we should think alike. Other theories of rationality suggest that such is not the case.

I shall not discuss this point further, however, because I am more interested in Uniqueness’ moral implications. Consider, for instance, Hedden's principle of expert deference, designed to remedy the inadequacy of the principle of reflection, and to avoid all references to personal identity (165). It essentially says that, instead of adjusting to your future rational beliefs (simply because they are yours and they are posterior to your current ones), you should adjust your current beliefs to those of people you know are better informed than you. I think that this is a reasonable claim. If one accepts the principle of impersonality, one has no reason to believe that one's future beliefs are more rational than present ones. Hedden makes a similar claim for preferences with the principle of Preference Deference. This is more disputable, since preferences are more dependent on who you are than on the evidence you have. It is, however, a suggestive view and one that I wish Hedden had further developed. Together with the Principle of Uniqueness, it implies that it is rational for less informed selves to trust better informed ones (whether they belong to the same agent or not). And if it is morally acceptable for my present self to defer to the preferences and/or beliefs of my future selves, it ought to be acceptable for my present self to defer to the preferences and/or beliefs of another agent providing she is better informed than I am. ‘As far as deference principles go’, concludes Hedden, ‘any intrapersonal requirements of rationality follow from more general principles that apply equally in the interpersonal case’ (172). Does this include prudence? Unfortunately not. Prudence is strictly an intrapersonal principle. This is where diachronic and synchronic principles of rationality basically differ. If individuals did not exist as persons – that is, as temporally continuous units – this would not really matter. But they do, so it does.