Rationality

From Liberpedia

Rationality in economics concerns choice among alternatives in relation to purposes, information and constraints. Different uses of the term ask different questions.[1]

Purposeful action

Ludwig von Mises uses rational to describe action directed towards an end. This does not imply that the actor's beliefs are correct, that the chosen means will succeed, or that the aim is material gain. His analysis takes the actor's ultimate ends as given; psychology investigates the processes producing particular actions.[2]

Decision rules

Herbert A. Simon distinguishes maximising an expected payoff, maximising the worst possible payoff, and choosing the best known outcome under certainty. The rules use different assumptions about information and how outcomes are evaluated.[1]

For illustration, suppose there are two equally probable states and the following utility payoffs:

Alternative State 1 State 2 Expected payoff Worst payoff
A 6 6 6 6
B 0 14 7 0

The expected-payoff rule selects B; the max-min rule selects A. The difference follows from the decision rules applied to the stated payoffs. It does not establish how people actually choose.

Information, search and calculation

Simon proposes models that incorporate the alternatives a person considers, limited calculation and incomplete knowledge. One procedure searches for an alternative meeting an aspiration level and stops when it finds one. The result can depend on search order and adjustments to that level.[1]

An acceptable alternative need not be the best available one. Nor does adding search costs to an optimisation problem show that the chooser knows those costs or can calculate the optimal search. In the appendix, Simon derives an optimal acceptance-price rule under detailed assumptions about future offers. He also describes approximations for a seller who lacks that information.[1]

These models propose decision procedures that can be investigated empirically.

Cognitive reflection and calculation

Performance on cognitive-reflection tests can combine noticing a tempting error with doing the calculations needed to correct it. Sinayev and Peters separated these responses and found that numerical ability accounted for the predictive contribution of their reflection measure in the decision tasks they examined.[3] Pennycook and Ross found a separate association with religious and moral judgments in another sample, arguing that the willingness to think analytically also matters.[4]


Inference with limited information

Gerd Gigerenzer and Daniel Goldstein’s Take The Best rule searches cues in order of their estimated predictive validity and stops when one distinguishes the alternatives. The decision uses that cue without combining the remaining evidence. Recognition can also guide a choice when recognised objects tend to have higher values on the quantity being predicted.[5]

Their simulations used 83 German cities and nine binary cues. Average accuracy was 65.8% for Take The Best, 65.8% for weighted tallying and 65.7% for multiple regression across the tested knowledge states. Take The Best looked up an average of 5.91 cue values, compared with 20 for the integration methods. These are simulated lookups, rather than measured human reaction times.[5]

Successful prediction does not by itself establish the psychological procedure producing a judgement. The authors assumed the rule’s descriptive validity for this comparison. Performance depends on the information and environment: they also noted that regressing on city ranks rather than populations performed slightly better.[5]

Expertise and feedback

Daniel Kahneman and Gary Klein argue that skilled intuition develops when an environment supplies reliable cues and a person has sufficient practice and corrective feedback. Rapid recognition can then draw on learned patterns without comparing every alternative.[6]

Experience must be assessed for the particular task. Someone can be skilled at one activity and poorly calibrated at another. Confidence alone does not establish accuracy, especially when feedback is delayed or ambiguous. The authors also distinguish outperforming novices from outperforming a statistical model: these are different standards of success.[6]

An algorithm's comparative advantage does not make its predictions certain. Its use depends on measurable outcomes, relevant past cases and conditions stable enough for learned relationships to remain useful.[6]

Framing and observed choice

Amos Tversky and Daniel Kahneman asked students to choose a programme for a hypothetical disease expected to kill 600 people. The certain option saved 200 people; the risky option had a one-third chance of saving all 600 and a two-thirds chance of saving none. Another version expressed the same outcomes as deaths.[7]

Description of the certain option Respondents Choosing that option
200 people survive 152 72%
400 people die 155 22%

The percentages compare separate respondent groups. Both descriptions specify the same final survival or death totals, but the authors found different patterns of choice. They treat perception and evaluation of the problem as part of the explanation of decision making.[7] See evolutionary psychology and economics.

Individual choice and institutional performance

Vernon Smith distinguishes constructivist rationality, involving conscious models and deliberately designed rules, from ecological rationality, concerning how behaviour and institutions emerge and function through experience. In market experiments he reviews, participants can achieve high gains from exchange with private information and without solving the equations used by the experimenter.[8]

Individual judgment and market performance are therefore different empirical questions. Smith also discusses asset-market bubbles and failures of coordination: efficiency cannot simply be inferred from the existence of a market. Rules, information, experience and behaviour need to be examined together.[8]


See also

  • Prebunking — training and the measurement of judgement about misleading information.

References

  1. ↑ 1.0 1.1 1.2 1.3 Herbert A. Simon, “A Behavioral Model of Rational Choice”, Quarterly Journal of Economics 69(1), 1955, pp. 99–118; Carnegie Mellon University archival reprint. See sections I–IV and the appendix. DOI.
  2. ↑ Ludwig von Mises, Human Action, chapter I, sections 1 and 4, “Acting Man”, Econlib edition.
  3. ↑ Aleksandr Sinayev and Ellen Peters, “Cognitive reflection vs. calculation in decision making”, Frontiers in Psychology 6, article 532, 2015, especially pp. 3–5 and 11–14; published full text.
  4. ↑ Gordon Pennycook and Robert M. Ross, “Commentary: Cognitive reflection vs. calculation in decision making”, Frontiers in Psychology 7, article 9, 2016, pp. 1–3 and Table 1; published full text.
  5. ↑ 5.0 5.1 5.2 Gerd Gigerenzer and Daniel G. Goldstein, “Reasoning the Fast and Frugal Way: Models of Bounded Rationality”, Psychological Review 103(4), 1996, pp. 650–669, especially pp. 653–660, Table 2 and note 4.
  6. ↑ 6.0 6.1 6.2 Daniel Kahneman and Gary Klein, “Conditions for Intuitive Expertise: A Failure to Disagree”, American Psychologist 64(6), 2009, pp. 515–526, especially pp. 519–524; public author-uploaded text.
  7. ↑ 7.0 7.1 Amos Tversky and Daniel Kahneman, “The Framing of Decisions and the Psychology of Choice”, Science 211, 30 January 1981, pp. 453–458, especially Problems 1–2 on p. 453 and the discussion on pp. 457–458.
  8. ↑ 8.0 8.1 Vernon L. Smith, “Constructivist and Ecological Rationality in Economics”, American Economic Review 93(3), June 2003, pp. 465–508, especially pp. 466–471, 475–476 and 480–483; university-hosted published article, a revised version of his 2002 prize lecture.