As a consumer of economic theory, I am primarily interested in social theories for their rhetorical value. While my attraction toward theoretical models developed during my studies of mainstream economic models of decision-making (rationality, sophisticated probabilistic beliefs, Bayesian updating, etc.), I continue to believe in the potential of alternative models of decision-making to illustrate economic phenomena.
Seemingly Rational Choice
I propose a model of choice (seemingly rational choice) in which choices from a given menu are maximal according to a given preference relation, but only amongst the alternatives that have been chosen themselves in some menu. Choices are not restricted to outrank any alternatives that have never been sampled by the decision-maker. These choices play a dual role of maintaining consistency among themselves with respect to a given true welfare ordering, while also defining the knowledge available to the decision-maker about their preferences. This model is observationally equivalent to rational choice, but may differ significantly from welfare-optimal choices depending on the sparsity of the choice space. I characterize the multiplicity and sub-optimality of seemingly rational choices in finite choice spaces. I then analyze some competitive models of sampling in which two senders reveal alternatives to a receiver, as well as possible forms of seemingly rational consumer choices.
A sender wishes to persuade a receiver who holds point-prediction beliefs. The sender reveals a sequence of hard evidence that evokes a belief consistent with the evidence presented, but this belief contains more specificity than the evidence itself. The evocability of beliefs are characterized under particular functional forms of belief formation inspired by the heuristics literature. When the sender is uncertain of the true state, speeches that maximize the probability of implementing their preferred state are characterized by a maximally-agreeable revelation of evidence that solve a class of set-cover problems. A dynamic version of the model with feedback to the sender is included to capture the phenomenon of dynamic ``phishing''.
Works in Progress
Sequential Choice Functions with Mauricio Ribeiro
Preference Manipulation in Pie-Splitting Problems