Expected utilityis always used as aheuristic
Konrad Talmont-KaminskiMarie Curie-Sklodowska U., Poland
Line of argument
Problem with bounded rationalityDual systems of reasoningWhat is a heuristic?Expected utility as a heuristic
H. Simon Reason in Human Affairs, 1983
Bounded rationality
Processes satisficing not maximisingMethods context-dependentReasoning consists of heuristics
All reasoning?
Heuristics all the way up?
What about logical reasoning?What about scientific reasoning?Does not look like heuristicsBut…
Simon’s explanation of discovery of Boyle’s Law
J. EvansK. FrankishS. SlomanK. StanovichD. Kahneman& others
Dual reasoning
System 1Heuristics
Intuitive, quick, innate, automatic
System 2Real reasoningLogical, slow, learned, conscious
Kahneman’s system 1
Simple heuristicsAnchoring & adjustment, representativeness, availabilityDescriptiveFocus on errors producedNecessitated by empirical data
But Gigerenzer’s critique
Kahneman’s system 2
Logical reasoningUtility theory, etc.Descriptivebut also normativeNecessitated by desire to maintain classical norms of rationality &Need to explain ‘system 2’ reasoning
Expected utility
Paradigmatic exampleIf EU functions as heuristic, (plausibly) all reasoning doesEU does not function as Kahneman heuristicNeed to go back to Simon’s concept
Simon
Not necessarilyNot necessarilyNot necessarilyNot necessarily
Kahneman
IntuitiveInnateQuick
Automatic
Simon’s heuristics
Very broad categoryKahneman examplesBehavioural adaptationsScientific reasoningIn danger of becoming vacuousOnly useful if it supports substantive generalisations
Bill WimsattReengineering Philosophy for Limited Beings, 2007
Common traits of heuristics1. Fallible2. Efficient3. Systematically biased4. Problem-transforming5. Purpose relative6. Descended from other
heuristics
EU as heuristic?
EU formalism does not look like a heuristicBut how is EU applied in real situations?Using simplifying assumptionsNeed to consider formalism+assumption
Traits of EU+assumptions
1. Fallible2. Efficient3. Systematically biased4. Problem-transforming5. Purpose relative6. Descended from other
heuristics
?
?
Heuristics all the way up
EU functions as a heuristicMaximising formalism situated in satisficing methodologySame phenomenon in other cases?Kahneman’s ‘system 2’ not descriptiveProbably not primitively normative, either
H. MercierD. SperberWhy do Humans Reason,BBS forthcoming
Conclusions
It is heuristics all the way upKahneman’s conception of heuristics very limitedSimon’s conception much broader and more usefulNeed to go back to Simon
Thank you
Konrad Talmont-KaminskiIn a Mirror, Darkly: How the Supernatural Reflects Rationality (forthcoming)