Date of Award

2020

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Graduate Group

Psychology

First Advisor

Sudeep Bhatia

Abstract

Preferential choices are often products of stochastic accumulation of noisy preferences in favor of different options. In a collection of three essays, we use sequential sampling models to formally describe preference accumulation processes underlying various choice problems. The models we propose are based on the drift diffusion model (DDM), which is a benchmark sequential sampling model for studying binary choice problems, is biologically plausible, and has a clear interpretation regarding optimal decision making. Chapter 1 focuses on automatic biases in intertemporal choices. The use of DDM allows us to define the term bias precisely, and measure its direction and magnitude through quantitative model fits. Using model fits to individual-level choice and response time data, we show that automatic biases are prevalent in intertemporal choice, but that the type, magnitude, and direction of these biases vary greatly across individuals. In Chapter 2, a similar modeling approach is applied to study why decision makers often reject low-stake positive-expected-value gambles. The standard explanation for this phenomenon is utility weighting. In this chapter, we discuss alternative psychological mechanisms, such as a bias favoring rejection prior to gamble valuation. Through two preregistered experiments, we demonstrate that the pre-valuation bias provides a large contribution to model fits, predicts key response time patterns, reflects prior expectations regarding gamble desirability, and can be manipulated independently of the valuation process. Chapter 3 examines the impact of context effects on decisions; factors such as option presentation format, preference-elicitation procedures, as well as incidental affect, social belief, and cognitive capacity. We use data from two large-scale choice experiments to construct a space of context effects, based on DDM decompositions of more than fifteen different context effects. By representing a large number of context effects in terms of how they influence the parameters of the DDM, we can quantify the (dis)similarity between a pair of context effects, and interpret different effects in terms of their behavioral, mechanistic, and normative implications. Together, our findings point to the value of sequential sampling models in studying the dynamics of decision making, and in bridging the gap between formal cognitive theory and observable decision outcomes.

Embargoed

Available to all on Saturday, September 09, 2023

Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Included in

Psychology Commons

Share

COinS