Instead of the unified preference-map of normative decision theory, most of us operate with one or other of a number of preference-systems, according to the ‘hat’ we are currently wearing. When we come home from the office or go out to a party, we may switch easily and unnoticingly from one system to another. (A holiday-maker lazes on the beach in hot sun and hedonic mood. Then he jumps up: a swimmer is in difficulties. His hedonic calculus switches off, moral concern switches on.) But sometimes, of course, the other preference-system stays switched on. If they indicate conflicting decisions the agent will have to make some sort of meta-decision about which system shall now predominate before he can decide what to do.1