Investigating Sampling Approaches to Decision Making

Imagine choosing between restaurants in an unfamiliar city, deciding whether to invest in a new company, or selecting between novel medical treatments: how do we decide between options that we have never directly experienced before? A recent line of explanations for these choices are mental sampling models in which decision makers generate potential outcomes in their mind and choose the option which these mental simulations suggest they would prefer. Sampling offers a powerful and intuitive solution to a wide range of cognitive tasks, narrowing our complex and uncertain environment into concrete examples to make such decisions feasible. Existing theories do not always agree about how such sampling works, however, and the specific form of sampling can impact the resulting choice: for example, each new sample may be similar to its predecessor, meaning our imagination may not move far from the first option we consider, or certain features like high prices may come to mind more easily, leading to greater focus on these aspects. Moreover, while a great deal of work has been done on these types of decisions, currently available data do not well distinguish between different types of sampling, focussing on the choices people make rather than the information they consider beforehand. There is thus a current need for more detailed examination of the sampling process in choice. Mental sampling theories are not unique to decision making, however: work in other domains has used sampling to explain varied behaviours such as estimates of perceptual features, judgments of probability and forecasts of financial prices. This work has focused specifically on identifying the sampling process using both behavioural experiments and computational modelling. These methods may then offer new insights into the psychological systems supporting decision making, but have thus far been limited to situations with clear objective solutions (e.g., the number of dots on a screen) rather than the uncertain or subjective valuations of everyday life. This research seeks to bridge such sampling work with decision making: behavioural experiments will examine the samples produced by decision makers before choice, looking for patterns such as similarities between concurrent samples and whether the frequencies of generated items match their true probabilities. The findings from these experiments will then guide the construction of new sampling models directly incorporating these patterns into their mechanisms, which will be compared against existing standards in the field to assess their ability to capture choice behaviour. Linking these subjects will contribute to a broad sampling framework providing a common solution to the varied problems that people might encounter in their everyday life, from simple perceptual judgments like estimating the size of a crowd to complex real-world decisions with far-reaching consequences like purchasing a home or selecting medical treatments. Such a framework could offer consistent explanations for many psychological phenomena, including suboptimal choices, errors in estimates of probability, and biases in predictions of future prices, as well as potential connections between these effects not predicted by existing theories. More generally, understanding how such decisions are made holds pro-social applications in helping people towards better choices, guiding policy on education as well as the regulation of industries such as finance or gambling to discourage damaging behaviour.