ÁñÁ«ÊÓƵ

A collection of Bernard Williams's essays is a treat: civilised, sharp discussions of serious issues, spiked with asides which are deep, funny and sometimes both.

This collection includes 21 essays, mostly written since the publication of Ethics and the Limits of Philosophy (1985). They fall into three groups: six on freedom and action; six on the human sciences; and nine on ethics. It is good to have them under shared covers, if frustrating to discuss so few of their themes in a short review. (World, Mind and Ethics: Essays on the ethical philosophy of Bernard Williams, 1995, edited by J.E.J. Altham and Ross Harrison, discusses many of these themes more thoroughly and includes a reply by Williams.) The central claims of Williams's work on ethics are well known: he doubts the possibility of moral theory; he has deep misgivings about obligations, utilitarianism and "Kantian" morality and their supposed culmination in an odious "morality system" in which theory and obligation are linked to an obsession with blame. Yet he is no moral sceptic. So the interesting question is to see what he substitutes for "the morality system".

Williams begins with a thought that is partly Hegelian. As he sees it, each of us has an actual motivational set at any given time: a combination of cognitive and affective elements, such as desires, attitudes, projects and commitments. Motivational sets define who and what each of us is; they are the basis of our identity and integrity. There is no going behind them to any external source of reasons for action. There are also (at least) two unHegelian aspects to Williams's views. First, he does not locate motivational sets within any history of reason. Second, he views them not as communal but as individual, hence speaks of them as providing internal reasons for each agent. Nothing external to an agent's motivational set provides her with a reason for action; Hume was right and Utilitarians and Kantians are wrong to think that some reasons are external to agents; there is no vindication without motivation. (This disagreement is readily obscured by the fact that Kantians use the terms "external" or "alien" in a converse sense: Kantians insist that nothing external to reason can vindicate; Williams that nothing external to agents can be motivated).

Will not the normativity of ethics be threatened, and "ought" collapse into "is", if all reasons are internal to agents? Williams argues to the contrary. Actual motivational sets do not wholly determine action, which also needs deliberation. Sound deliberation can alter motivational sets in the light of "facts and reasoning" and of imaginative viewing of alternatives; but it cannot insert external prudential or ethical considerations into motivational sets, since there are no external reasons and there is no coherent ethical theory.

The normativity of ethics reflects no more than the claim that the internal critique provided by sound deliberation creates a gap between immediate implications of a motivational set and its reflectively revised claims, so allows for a distinction between "is" and "ought". Sound deliberation from a given motivational set may establish reasons for action which actual deliberation and action fail to meet.

Does internal critique alone provide enough distance? Williams rightly challenges those who think there are external reasons for action to show what they are, and how they can be reasons for action unless they motivate. Advocates of external reasons may counter-challenge, by demanding Williams shows that the elements of a motivational set, however reflectively revised, can vindicate action. "I want it" is no vindication; nor is "I still want it after thinking about it". Internalists think externalists rely on illusory sources of vindication; externalists that internalists rely on inadequate, and dangerously plastic conceptions of vindication.

For Williams one of the achievements of an internal account of reason is that it undercuts the malign practice of blaming those whose motivational sets could not have provided them with reasons to act otherwise than they did. Blame is appropriate only if an agent's motivational set, worked on by sound deliberation, could have led him to avoid the act or omission.

By contrast, externalists may blame agents for acts or omissions which they had no internal reason to avoid. Externalists are a heterogeneous lot, but I suspect that many of them would counter that the obsession with blame which Williams attributes to them is no part of externalism, or part only of some of its versions. Backward-looking questions about the appropriateness of blame for past action may be only a minor aspect of living well, and the main point of external reasons is to be forward-looking and action-guiding. Externalists will allow that motivation is opaque and often fails to live up to external reasons; but they will also suggest that a grasp of sound external reasons may sometimes form, reform, even transform motivation.

Onora O'Neill is principal, Newnham College, Cambridge.