Consideration of limited attention in economic analysis.

Summary This thesis contributes to the consideration of limited attention in economic analysis. We defend the idea that attention allocation processes can be studied through a production process with as input the allocated attention (the amount of attentional resources invested in the decision), and as output the effective attention (the amount of information contained in the decision). In order to improve the understanding of these processes, we propose three experiments borrowing methods from psychology and cognitive sciences. In the first chapter, we manipulate the order of presentation between an incentive information and a visual stimulus, in a two-alternative forced choice paradigm. Allocated attention is controlled, and we measure effective attention using a signal detection model. We show that the last information presented has a greater weight in the decision and attribute this effect to a division of attention. The second chapter proposes an experiment in which participants allocate costly attention to reduce uncertainty in a discrimination task. Thus, we measure both allocated attention (through response time) and effective attention (through performance). This experiment allows us to study social attentional dilemmas (situations where attention is costly for the individual but beneficial for the group) and to highlight a discrepancy between social preferences - traditionally measured by monetary allocation choices - and the behaviors observed in our social attentional dilemma. The last chapter proves that it is possible to empirically implement and test the validity of a revealed preference model with random attention. We propose a new characterization and theorem of revealed preferences in the context of a more general version of the Brady and Rehbeck (2016, Econometrica) model. We develop statistical procedures - which we analyze using numerical simulations - to test the model's axioms, reveal preferences, and obtain a measure of effective attention. We test the validity of the model using a selective attention task in which participants choose a monetary payoff from a set of distractors. The behaviors observed in this experiment are consistent with the model and the experimentally induced preferences.