New frameworks of rationality

Abstract

New Frameworks of Rationality Nick Chater (Nick.Chater@wbs.ac.uk) Behavioural Science Group, Warwick Business School, University of Warwick Coventry CV4 7AL, UK Klaus Fiedler (klaus.fiedler@psychologie.uni-heidelberg.de) Psychologisches Institut, Universit¨at Heidelberg, Hauptstrasse 47–51 69117 Heidelberg, Germany Gerd Gigerenzer (gigerenzer@mpib-berlin.mpg.de) Center for Adaptive Behavior and Cognition, Max Planck Institute for Human Development, Lentzeallee 94 14195 Berlin, Germany Karl Christoph Klauer (klauer@psychologie.uni-freiburg.de) Institut f¨ur Psychologie, Universit¨at Freiburg, Engelberger Str. 41 79085 Freiburg, Germany Mike Oaksford (m.oaksford@bbk.ac.uk) Department of Psychological Sciences, Birkbeck College, University of London, Malet Street London, WC1E 7HX, UK Keith Stenning (k.stenning@ed.ac.uk) School of Informatics, University of Edinburgh, 10 Crichton Street Edinburgh EH8 9AB, UK approach which assumes that players can resolve this prob- lem by a process of “virtual agreement” – that is, the players figure out what they would agree to do, if they could dis- cuss or bargain. Where the answer is well-defined, the agree- ment can be reached “virtually”, i.e., without any information being exchanged. Virtual agreement requires common aims and knowledge – and can fail when players mis-estimate this common ground. Interesting, the process of reaching agree- ment can, in some cases, be modeled by conventional game theory. Keywords: rationality; bounded rationality; reasoning; prob- abilistic reasoning; deductive reasoning; decision making; heuristics Introduction The nature and extent of human rationality is an issue of on- going debate. In the last two decades, this debate has been enlivened by the development and application of new theo- retical frameworks. These include Bayesian notions of ad- justing and using uncertain beliefs in an inductive manner as well as deductive probability-based logics as normative guidelines against which to weigh human judgments and de- cisions; the notion of ecological rationality based on lean and frugal heuristics well adapted to the structure of the envi- ronment; the notion of meta-cognitive myopia according to which people are accurate and sensitive in the processing the information in a given sample of observations, but are blind and naive to the history and validity of the sampled data; and game theory. Cognitive Myopia (Klaus Fiedler) What I have come to call “meta-cognitive myopia” (MM), using a term once suggested by Robyn Dawes, is the phe- nomenon that people are pretty accurate in utilizing even large amounts of stimulus information, whereas they are naive and almost blind regarding the history and validity of the stimulus data. This uncritical reliance on the information given is the most conspicuous when the task context makes it crystal-clear that the stimulus data should not be trusted. In the introduction, MM is located within a broader frame- work of meta-cognition research, and several examples are provided to illustrate the phenomenon. The central message is laid out that MM offers an alternative account of many bi- ases in judgment and decision making, which have been tra- ditionally explained in terms of capacity constraints, limited reasoning ability, motivational forces, or severely biased en- vironmental input. The explanatory power of the MM con- struct, and its theoretical potential to predict new findings, is Virtual Agreement: A Rational Framework for Joint Action and Communication (Nick Chater) Game theory typically models interactions between agents in terms of players that are rational at the level of the individual. But when people need to coordinate their behaviour, which arises in joint action and communication, a vicious circle arises. What is rational for each player depends on what the other does; but figuring out what the other will do is no easier than figure out what one should do oneself. I will describe an

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 2 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.