skip to main content

Thesis Seminar

Friday, May 6, 2016
3:00pm to 4:00pm
Add to Cal
Beckman Behavioral Biology B180
Mechanisms underlying Economic Choice
Gideon Nave, Graduate Student, Computation & Neural Systems, Caltech,

The current dissertation proposes three manners in which findings about the neuroscience of decision-making can inform traditional questions in economics that historically has been investigated using choice data alone, and without delineating the mechanism of choice.

The first chapter investigates the origins of a critical component of both economic and perceptual decision-making under uncertainty, the belief formation process. Most research has studied belief formation in economic and perceptual decisionmaking in isolation. One reason for this separate treatment may be the assumption that there are distinct psychological mechanisms that underlie belief formation ineconomic and perceptual decisions. An alternative theory is that there exists a common mechanism that governs belief formation in both domains. Here,we test this alternative theory by combining a novel computational modeling technique with two well-known experimental paradigms. I estimate a drift-diffusion model (DDM) and provide an analytical method to decode prior beliefs from DDM parameters. Subjects in our experiment exhibit strong extrapolative beliefs in both paradigms. In line with the common mechanism hypothesis, we find that a single computational model explains belief formation in both tasks, and that individual differences in belief formation are correlated across tasks. These results suggest that extrapolative beliefs in economic decision-making may stem from low-level automatic processes that also play a role in perceptual decision-making, and therefore might be difficult to suppress.

The second chapter investigates the role of the sex steroid hormone testosterone as a biological mediator that translates environmental changes into shifts in cognition, that influence decision-making. Correlational studies have linked testosterone with aggression and disorders associated with poor impulse control, but corresponding mechanisms are poorly understood and there is no evidence of causality. Building on a dual-process framework, I identify a mechanism for testosterone's behavioral effects in humans: reducing cognitive reflection. In the largest testosterone administration study to date, 243 men received either testosterone or placebo and took the Cognitive Reflection Test (CRT) that estimated their capacity to override incorrect intuitive judgments with deliberate correct responses. Testosterone administration reduced CRT scores. The effect was robust to controlling for age, mood, math skills, treatment expectancy and 14 other hormones. The effects were enhanced in subjects with high cortisol and estradiol levels. These findings suggest a unified mechanism underlying testosterone's varied behavioral effects in humans and provide novel, clear and testable predictions.

In the third chapter, I study dynamic unstructured bargaining with deadlines and one-sided private information about the amount available to share (the "pie size"). Using mechanism design theory, I show that given the players' incentives, the equilibrium incidence of bargaining failures ("strikes") should increase with the pie size, and I derive a condition under which strikes are efficient. In our setting, no equilibrium satisfies both equality and efficiency in all pie sizes. I derive two equilibria that resolve the trade-off between equality and efficiency by either favoring equality or favoring efficiency. Using a novel experimental paradigm, I confirm that strike incidence is decreasing in the pie size. Subjects reach equal splits in small pie games (in which strikes are efficient), while most payoffs are close to either the efficient or the equal equilibrium prediction when the pie is large. I employ a machine learning approach to show that bargaining process features recorded early in the game improve out of sample prediction of disagreements at the deadline. The process feature predictions are as accurate as predictions from pie sizes only, and adding process and pie data together improve predictions even more. As process data can be much richer than the series of cursor locations that we have used (for example, by including skin conductance, pupil dilation or facial expressions), better inference of outcome variables is likely feasible. Thus, if a policy maker or a mediator can access an independent measure of private information, an arbitration mechanism may allow boosting efficiency by taking this measurement into account.