Improving choice by automatically restructuring decision environments

Abstract

Many of the computational problems people face are difficult to solve under the limited time and cognitive resources available to them. Overcoming these limitations through social interactions and cognitive offloading is one of the most distinctive features of human intelligence. This dissertation explores ways of improving choice and augmenting cognition by automatically restructuring people's decision environments and social networks. This approach uses psychological models developed by researchers as engineering tools, and allows individuals to benefit from increasingly powerful artificial systems. In a series of studies, we show how this approach can lead people to better decisions and reduce harmful side effects of traditional "static" offloading. Crucially, this approach can also give individuals greater autonomy and control over how their decisions are guided and shaped. Chapter 2 introduces a novel formal framework for modeling and evaluating the effects of "nudges" based on the insights that nudges change the problem of how to make a decision without changing the decision itself. We then show how this model can be used to optimize choice environments and automatically construct optimal nudges that best improve choice. Chapter 3 shows how Bayesian and psychometric modeling can be used to develop a new model of group decision-making in settings with repeated population turnover. We then show that this model can be used to automatically restructure people's networks so that people benefit from social observation without it increasing their bias. Chapter 4 shows how restructuring environments can be extended by using modern text-to-image AI models to help people better imagine alternative futures. Crucially, we show that this approach can be used to increase support for real-world policies and proposals. Chapter 5 concludes by discussing the broader implications of this work, limitations of the studies and models discussed here, and opportunities for future work.