In a recent essay in this journal Amartya Sen introduced the notion of an evaluator-relative consequence-based morality. The basic idea can be described very simply. A consequence-based morality is a morality that instructs each agent to maximize some objective function defined over states of affairs. Such a morality is evaluator neutral if it assigns to every agent the same objective function. If different agents have different objective functions, then the morality is evaluator relative. For example, a morality would be evaluator relative if it assigned to Jones an objective function giving greater weight to the welfare of Jones's children than Smith's children, while it assigned to Smith an objective function giving greater weight to the welfare of Smith's children than Jones's. In more traditional language, a morality is consequence-based if it says that right acts are acts which have good consequences. A consequence-based morality is evaluator neutral if there is a universal good that all agents are required to promote; it is evaluator relative if different agents are assigned different goods. This brief description conceals some complications and possible sources of confusion, but it is not intended as a summary of or a replacement for Sen's very illuminating general discussion of agent relativity. Sen has shown that evaluator relativity must always be accompanied by some other form of relativity, but my focus will be on evaluator relativity except where I advert specifically to another form.
Regan, Donald H. "Against Evaluator Relativity: A Response to Sen." Phil. & Pub. Aff. 12, no. 2 (1983): 93-112.