Principles can seem as entrenched in moral experience as Kant thinks space, time, and the categories are in human experience of the world. However not all cultures have such a view. Classical Indian and Chinese philosophies treat modification of the self as central to ethics. Decisions in particular cases and underlying principles are much less discussed.

Ethics needs comparative philosophy in order not to be narrow in its concerns. A broader view can give weight to how people sometimes can change who they are, in order to lead better lives.

In the 1950's some prominent philosophers suggested a logical relation weaker than entailment between primarily descriptive statements and ethical conclusions. The paper revisits this suggestion. It examines four ways in which ethical statemnts can be supported by descriptions and evaluations. This provides a similarity bteween some kinds of reason-giving in ethics and familiar cases of logical inference, making it plausible to speak of a logic. The similarity however is limited, and the strength in ethics of descriptive reasons is never precise and always somewhat contestable.

Here are two widespread responses to Kant's categorical imperative. On one hand, one might note the absence of detailed rational derivation. On the other hand, even someone who maintains some skepticism is likely to have a sense that (nevertheless) there is something to Kant's central ideas.

The recommended solution is analysis of elements of the categorical imperative. Their appeal turns out to have different sources. One aspect of the first formulation rests on the logic of normative utterances. But others can be justified only in terms of their contributions to desirable functionings of a moral order.

Gilbert Harman has argued that it does not make sense to ascribe character traits to people. The notion of morally virtuous character becomes particularly suspect.

How plausible this is depends on how broad character traits would have to be. Views of character as entirely invariant behavioural tendencies offer a soft target. This paper explores a view that is a less easy target: character traits as specific to kinds of situation, and as involving probabilities or real possibilities. Such ascriptions are not undermined by Harman's arguments, and it remains plausible that the agent's character often is indispensable in explanation of behaviour. Character is indispensable also as processes of control that impose reliability where it really matters.

Many would consider the lengthening debate between moral realists and anti-realists to be draw-ish. Plainly new approaches are needed. Or might the issue, which most broadly concerns realism in relation to normative judgments, be broken down into parts or sectors? Physicists have been saying, in relation to a similarly longstanding debate, that light in some respects behaves like waves and in some respects like particles. Might realism be more plausible in relation to some kinds of normative judgments than others?