Three lectures by Peter Hammond on "Utility Theory and Its Applications"

[Marcus Pivato]

Dear all, 

On April 13, 15, and 16,  Peter Hammond will deliver a series of three lectures at the Centre d'Économie de la Sorbonne, on the topic Utility Theory and Its Applications: Some Recent Developments.   The exact time of each lecture will be decided based upon the schedule constraints of people who wish to attend.  If you wish to attend these lectures, then please respond to the three surveys below to indicate the times that are convenient for you.

Here are the titles and abstracts of the three lectures.

1. Prerational Behaviour in Decision Trees with Non-terminal and Menu Consequence Nodes  (based on joint work with Agustín Troccoli Moretti)

Abstract: We extend previous work on consequentialist decision theory to allow timed consequences that accrue at intermediate, non-terminal nodes. This extension implies that each path through a decision tree is mapped to a unique intertemporal consequence stream. Based on results from Hammond (1988b, 2022) concerning consequentialist normal form invariance and prerationality, we consider actual behaviour which, unlike plans or intentions, is dynamically consistent by definition. We prove that actual behaviour is prerational and continuous on Marschak triangles if and only if it maximizes a Bayesian rational base preference relation that is represented by the expected value of a Bernoulli utility index defined on the domain of consequence streams. We also permit any intermediate consequence to include a "menu" which depends on the set of consequences that are feasible in the continuation subtree whose initial node is the relevant consequence node. Introducing menu consequences allows prerational behaviour to become consistent with a plethora of prima facie “non-consequentialist” phenomena. Prominent examples include apparent violations of the ordinaliity and independence axioms of expected utility theory. Other instances of “menu effects” include temptation and regret.

Scheduling Survey

2.  Utilitized Prerational Behavioural for an Ethical Impartial Benefactor

Abstract: As in Rawls’ (1971) original position, we consider an impartial benefactor who behaves ex ante as if facing symmetric uncertainty about which individual’s personal consequence should be decisive ex post. We apply Hammond’s (2022) definition of prerationality to base preferences over profiles of individual personal consequences that belong to possibly different personal consequence domains. Assuming that an impartial benefactor’s base preferences are continuous as hypothetical probabilities vary, we derive a modified original position in the spirit of Vickrey (1945) and Harsanyi (1953, 1955). As in the “utilitized” approach to social choice set out in our companion paper Ferguson and Hammond (2026), this requires base preferences to be represented by the ex ante expected value of a Bernoulli utility function when an equal probability is attached to each ex post event of one particular individual becoming  in society.

Scheduling Survey

3. Prerational Best Responses and Action-Dependent Beliefs in Non-cooperative Games
(based on joint work with Ganna Pogrebna and Miguel Costa-Gomes)

Abstract: Several experimental studies of strategic games have elicited not only players' strategy choices, but also their probabilistic beliefs regarding each others' strategy choices. Often between 25% and 50% of observed decisions in such games are not best responses given these elicited beliefs. Following the basic assumption of orthodox game theory, however, these experiments restrict attention to "action-independent'' beliefs that do not depend on a player's own chosen action. Our experiment, by contrast, allows players also to express "action-dependent'' beliefs which violate this basic independence postulate. Not only do elicited action-dependent and action-independent beliefs often differ. So do the players' best responses to each. Moreover, in our experiment, when the best responses do differ, action-dependent beliefs typically explain observed behaviour somewhat more frequently than action-independent beliefs do.

Scheduling Survey

Please complete all three scheduling surveys by Thursday, 19 March.  Thank you very much in advance.

Best wishes,

   Marcus