IIF   26912
INSTITUTO DE INVESTIGACIONES FILOSOFICAS
Unidad Ejecutora - UE
congresos y reuniones científicas
Título:
Beyond indeterminate utilities: the case of neurotic cake-cutting
Autor/es:
ELEONORA CRESTO
Lugar:
Nueva York
Reunión:
Jornada; Conference and Memorial in Honor of Isaac Levi; 2019
Institución organizadora:
Columbia University
Resumen:
Isaac Levi famously argued that we should distinguish between the seemingly conflicting forces we experience in cases of weakness of the will, and genuine cases of value conflict. The latter require inquiry, rather than therapy. However, responsible inquiry may well fail to yield a resolution. In such cases there isn't a unique cardinal utility function (up to positive affine transformation) that can represent the preferences of an agent: as with probabilities, utilities can be genuinely indeterminate. Here I explore a different sense in which indetermination can arise. It is clear that sometimes agents undergo preference changes. Interestingly, sometimes such changes are prompted by the fact that the agent comes to believe that someone else can be credited with a particular preference structure. In certain contexts, shifts of this sort give rise to a peculiar regret after acting in agreement with older preferences. This type of neurosis, so to speak, cannot be explained (nor cured) with the aid of sets of utility functions; the solution cannot rely on therapy to bolster will power either. We find paradigmatic examples of this general phenomenon at the time of addressing fair division problems. Typically, a fair division algorithm (such as a cake-cutting algorithm) is justified by showing that it fulfills a number of desirable properties, such as Proportionality, Pareto Optimality, or No-Envy. I identify a further desirable property, which I dub ?No Conditional Regret?. I argue that virtually all common mechanisms designed to achieve fair division are prone to conditional regret. To remedy this, I propose new algorithms that allow agents to successively adjust their preference structures, until they are all happy with what they would obtain. Finally, I give necessary and sufficient conditions for the adjustment procedure to stop, in which case the No Conditional Regret property will be indeed satisfied.