California Institute of Technology
Pasadena CA 91125
Cambridge MA 02138
This chapter presents an emerging "case-based" system for decision making that centers on the retrieval of similar past episodes. It can be mathematically described by the economic model of case-based decision theory accompanied by the computational background of case-based reasoning. This system stands between model-free and model-based systems on the bias-variance tradeoff and holds advantages over both when faced with novel and complex problems. The hippocampus and related medial temporal lobe structures provide a neural substrate for case-based calculations by instantiating stimulus and reward associations. This neurocomputational understanding sheds light on how such calculations can also contribute to or interfere with model-free and model-based control.
Taking longer than expected to complete a task can cause disappointment if the expectation constitutes a reference point. In a real-effort lab experiment, I show that a worker's willingness to persevere in a task is influenced by information about task completion time. To directly assess the location and impact of reference dependence, I structurally estimate labor-leisure preferences with a novel econometric approach drawing on cognitive science. Once participants exceed an expectations-based reference point, their subjective values of time rise sharply. Workers who fall behind the reference point are demoralized as measured by ratings of task satisfaction.
When choosing among alternatives, we must also choose how much time to spend evaluating them. Popular theoretical models assert that this deliberation optimally balances the costs of time expenditure and benefits of better decisions. I propose and implement a method to test the optimality of individual deliberation. The method consists of checking a condition necessary for optimality, which is consistency of underlying preferences for time versus reward when incentives changes. I measure choices and response times of human participants in motion-discrimination tasks, and find significant departures from optimality when task difficulty and monetary incentives are varied. I also assess a further implication of optimality which applies when the ease of decision making reflects the difference in value between options: as the amount of time already spent deliberating on a problem grows, the standard of confidence required to make a decision should fall. I conduct the first test of new theoretical results by Fudenberg et al. (2015) that characterize the decision rule in this setting. Model fits indicate that participants are sensitive to the information that elapsed time provides about the value of continued deliberation, especially once they have experience with the task. Thus principles of optimality help capture certain facets of deliberative behavior.
The order in which options are presented influences choice in ways that parallel primacy and recency effects in memory, but the depth of this connection remains underexplored. I present sequences of art to experimental participants who select their favorite pieces, and show that cognitive load can selectively weaken choice primacy or recency depending on its timing, analogous to past findings in memory research. Primacy is reduced by an externally-imposed distractor task in between each option, or by natural fatigue, while recency is reduced by an extra delay containing a distractor after the last option is presented. Thus effective interventions to reduce choice biases may be built upon the disruption of memory encoding and consolidation. However, the distractors affect stimulus recognition memory in the opposite way as choice, consistent with theories suggesting that value processing and memory encoding can interact competitively.
Business Analytics (Caltech BEM/Ec 150, SP 2015)
Business Analytics (Caltech BEM/Ec 150, SP 2014)
Selected Topics in Economics: Business Analytics (Caltech Ec 101, WI 2012)
Student Directed Seminar: Behavioural Economics (UBC ECON 492B, 2008W Term 2)