Distinguishing Two Dimensions of Subjective Uncertainty

University Of California-Los Angeles, Los Angeles CA

Investigators

Abstract

The psychological meaning of uncertainty is critical in many theories across the social and behavioral sciences. Existing theories of judgment and decision making rarely differentiate types of uncertainty. This research suggest that people intuitively experience uncertainty along two distinct dimensions (1) random uncertainty which represents the degree to which an event is seen as inherently unpredictable; and (2) knowable uncertainty which represents the degree to which an event is seen as potentially predictable. For example, predicting whether a fair coin will land heads on the next toss seems to reflect purely random uncertainty, whereas determining whether a suspect is guilty or innocent appears to reflect pure knowable uncertainty. Most judgments under uncertainty such as forecasting the weather or the potential return on a stock investment require individuals to evaluate events that involve both dimensions of uncertainty. The research proposes that people make intuitive distinctions between random and knowable uncertainty, and this has important consequences for understanding not only for understanding behavior but also for policy making. This research will validate measures of random and knowable uncertainty; develop methods to study both types of uncertainty; and explore judgments and decisions that are affected by both types of uncertainty. The particular types of decision studied are probability and overconfidence; investment decisions; everyday budgeting and planning; and beliefs about social justice and social equity. The research will lead to better understanding of the complex decision we face in our lives and that policy makers face in their work. Although virtually all decision theories treat uncertainty as a unitary construct, probability theory since its inception has been bifurcated into models of chance events and models of belief in events that are or will be true or false The major schools today include Frequentists, who interpret probabilities as long run stable frequencies of events, and Bayesians, who interpret probabilities as subjective degrees of belief. In this project the investigators assert that this philosophical bifurcation of uncertainty mirrors an ambivalence that resides within most decision makers. To illustrate, compare one?s uncertainty whether a flipped coin will land heads or tails to one?s uncertainty whether a suspect is guilty or innocent. The former is aleatory in nature, and involves unknown outcomes that can differ each time one runs an experiment under similar conditions. In contrast, the latter is epistemic in nature, and arises from missing knowledge concerning an event that is, in principle, knowable.The investigators argue that: (i) people intuitively distinguish epistemic from aleatory uncertainty;(ii) these dimensions are logically independent and segregable, with most events entailing a mixture of both variants; (iii) perceptions of epistemic and aleatory uncertainty can vary systematically across uncertainty have important consequences for judgment and choice. Furthermore, the distinction between epistemic and aleatory uncertainty can help to parsimoniously explain a diverse range of existing research findings in the judgment and decision making literature. This project explores the psychological implications of reasoning under epistemic versus aleatory uncertainty. First, it develops three distinct methodologies for measuring and experimentally manipulating perceptions of uncertainty: (1) an Epistemic-Aleatory Rating Scale (EARS) reliably measures perceptions of uncertainty as two independent dimensions. Epistemic uncertainty is associated with a judge being ascribed credit (blame) for correct (incorrect) predictions whereas aleatory uncertainty is associated with a judge being seen as lucky (unlucky) for correct (incorrect) predictions. The investigators propose to refine this scale and further establish its validity. (2) Expanding and validating a lexicon of natural language expressions that distinguish epistemic thinking from aleatory thinking. (3) Developing a variety of techniques for priming the salience of epistemic vs. aleatory uncertainty, including a linguistic prime a pattern detection prime and a causal reasoning prime. Next, it explores behavioral implications of the perceived nature of uncertainty across a wide variety of domains, including: (1) judged probability and (2) investment decisions; (3) budgeting and planning; and (4) beliefs about social justice and social equity. .