Adam Morton

(1945 - 2020)

I tried to tease out what Gilbert means by "normative". It isn't obvious. I conclude that assumptions about ideal agents – not just ideal in the sense of error-free but also ideal in the sense of unlimited – and assumptions about ideal placement of oneself in another person's situation, are essential to what he means. I conclude that what he says is very plausible given these assumptions, though they themselves are very problematic. Especially problematic is the idea of an unlimited simulation of a limited agent's perspective.

I give grounds for taking seriously the possibility that some of the emotions we ascribe do not exist. I build on the premise that the experience of imagining an emotion resembles that of having one. First a person imagines having an emotion. This is much like an emotion, so the person takes herself to be having the emotion that she imagines, and acts or expects a disposition to act accordingly. The view sketched here contrasts possibly impossible emotions such as disembodied passion, blind rage, and Quixotic courage with real ones such as affection, anger, and bravery. Both these real emotions and the states of imagining impossible ones are things that really happen to us.

Is there such a thing as bounded rationality? I first try to make sense of the question, and then to suggest which of the disambiguated versions might have answers. We need an account of bounded rationality that takes account of detailed contingent facts about the ways in which human beings fail to perform as we might ideally want to. But we should not think in terms of rules or norms which define good responses to an individual's limitations, but rather in terms of desiderata, situations that limited agents can hope to achieve, and corresponding virtues of achieving them. We should not take formal theories defining optimal behavior in watered-down bounded form, even though they can impose enormous computational or cognitive demands