Are People Probabilistically Challenged? Book Review of Daniel Kahneman, Thinking, Fast and Slow

Abstract

Daniel Kahneman’s recent book, Thinking, Fast and Slow, is a must-read for any scholar and policymaker interested in behavioral economics. Thus far, behavioral economists did predominantly experimental work that uncovered discrete manifestations of people’s bounded rationality: representativeness, availability, anchoring, overoptimism, base-rate neglect, hindsight bias, loss aversion, and other misevaluations of probability and utility. This work has developed no causal explanations for these misevaluations. Kahneman’s book takes the discipline to a different level by developing an integrated theory of bounded rationality’s causes and characteristics. This theory holds that humans use two distinct modes of reasoning, intuitive (System 1) and deliberative (System 2), while systematically allowing their fast intuitions to supersede deliberation. These intuitions grow from familiar stereotypes, resemblances and emotions that oftentimes do not align with reality and take the person astray. The dominance of these untutored, and yet powerful, intuitions explains people’s various misconceptions in the domains of probability and utility.

The book is beautifully written: its insights are rich, profound, and at the same time lucid and exceptionally well presented. The book develops clear explanations for complex economic and psychological phenomena and accompanies them with brilliantly selected examples from real life and controlled experiments.

These virtues, however, do not make the book uncontroversial. My Review critically examines Kahneman’s account of people’s probabilistic irrationality that occupies the majority of the book. This examination unfolds my profound disagreement with Kahneman’s grim assessment of ordinary people’s reasoning, widely known as the bounded-rationality theory.

Specifically, I argue that Kahneman and his collaborators have used an incomplete and unstable set of criteria for appraising people’s determinations of probability. This set of criteria is incomplete for two reasons. First, it does not account for the divide between rule-free 'beliefs' and rule-based 'acceptances,' drawn by philosophers of rationality. Acceptance is a mentally active process that includes application of decisional rules to available information. Belief, by contrast, is a person’s feeling, sensation or hunch: an intellectually passive state of mind generated by unanalyzed experiences. People participating in Kahneman’s and other behavioral economists’ experiments have formed no rule-driven acceptances. All they did was to report their beliefs because this is what the experimenters asked them to do. Individuals’ rationality, however, properly manifests itself only in acceptances that apply rules of reasoning. A person who fails to align her intuitive beliefs with the rules of probability may well be – and oftentimes is – perfectly rational in her rule-driven decisions.

Kahneman’s criteria for probabilistic rationality also fail to recognize a distinct – and perfectly rational – framework of reasoning, known as causative or Baconian. Under this framework, an event’s probability corresponds to the quantum and variety of the evidence that confirms the event’s occurrence while eliminating rival scenarios. This qualitative evidential criterion separates causative probability from the mathematical calculus of chances. Individuals who apply this criterion in making case-specific decisions are far from being irrational.

Kahneman’s criteria for probabilistic decisions are unstable because they tolerate the presence of unspecified causality and malleable reference classes in experimental settings and provide no directives on how to deal with these problems. These problems are present in all of Kahneman’s experiments that mixed statistical data with case-specific events. As a result, these experiments do not reveal anything about the rationality of participants’ probabilistic decisions. A person who ascribes probability to a causally unspecified event featuring a malleable reference class can never go wrong: her guesswork is as good as Kahneman’s or anyone else’s.

Based on this analysis, I revisit the flagship experiments substantiating Kahneman’s theory of bounded probabilistic rationality ('The Blue Cab,' 'Feminist Bank Teller,' and 'The Librarian'). Contrary to Kahneman’s assessment, I demonstrate that these experiments reveal no irrationalities in the participants’ probabilistic decisions. Moreover, I show that the rationality criteria favored by Kahneman and other behavioral economists can set up the 'irrationality maze' - a situation in which all decisions available to a person are irrational.

I argue, in conclusion, that the 'bounded probabilistic rationality' theory is unproven. Policymakers therefore ought to put on hold the behavioral economists’ recommendations urging the government to step in and fix people’s probabilistic decisions. This paternalism, 'soft' and invasive alike, is wasteful and potentially pernicious. Implementing it will suppress individuals’ creativity and heterogeneity without introducing substantial improvements in their decisions.