Analyses of Bounded Rationality: Towards Economic Decision-Making
description
Transcript of Analyses of Bounded Rationality: Towards Economic Decision-Making
Analyses of Bounded Rationality: Towards Economic Decision-Making
Farley S.M. NobreEmail: [email protected] or [email protected]
Home Page: http://web.bham.ac.uk/fsn019/fsnobre.html
Ph.D. StudentThe University of Birmingham, England
Guest ResearcherThe Humboldt-University of Berlin, Germany
Seminar of Behavioural EconomicsJuly 11th 2003
Analysis and Design of Organisational Systems: Towards a Unified Theory
1. Problem Choice and Analysis
2. Solution Design:Definitions, Variables,
and Propositions
5. Evidences and Conclusions: Theory as Proposed vs. Findings
Part I
Part II
3. Data Gathering4. Data Analysis
Parts III and IV
Part V
Fig.1. Thesis Structure and Its Process of Theorizing
Contents: Part I – Problem Analysis
i. Classical Theories on Rationalityii. Bounded Rationality Theoryiii. The Genesis of Bounded Rationality Theoriesiv. Economic Decision-Making and Approximate
Reasoningv. Organisations and Conflictsvi. Conclusions
Contents: Part II – Solution Design
i. Proposal: CTP - explores bounded rationality theories
ii. Cognitive Psychology Modelsiii. A Model of Information-Processing Systemsiv. Knowledge Representation and Organisationv. Computing Perceptions for Decision-Makingvi. Conclusions
Motivations(i) Organisations subsume economic decision-making and problem
solving processes that involve trade-offs among alternatives characterised by uncertainties and incompleteness of information. Such processes lead organisational members to both intra-individual and group conflicts.
(ii) The former conflict arises in an individual mind and it can emerge from the influence of others. The latter type arises from differences between the choices made by distinct individuals in the organisation. In this case, individual participants are not in conflict but the organisation as a whole is.
(iii) The intra-individual and group conflicts that arise in organisations as exposed in (ii) are determined by cognitive limits of humans, and thus these conflicts cannot be solved by incentive and reward systems - i.e. inducements. Such cognitive limitations are synonymous of bounded rationality [March and Simon, 1993].
Part I:Problem Analysis
Part I: (i) Classical Theories on Rationality
Rationality is synonymous of: optimal choice; optimal procdures and outcomes
(intelligence); statistical decision analysis.
Rationality is defined as: A particular class of procedures for making
choices [March, 1994].
Part I: (i) Classical Theories on Rationality
The Theory of Subjective Utility (SEU): It underlies neo-classical economics; It postulates that choices are made:
a) among a given, fixed set of alternatives;b) with (subjectively) known probability distributions of
outcomes for each;c) And in such a way to maximize the expected value of
a given utility function.
[Simon, 1997a]
Part I: (ii) Bounded Rationality Theory
Bounded Rationality [Simon, 1947; and March and Simon, 1958]:
It is also concerned with rational choice; But it takes into account the cognitive limitations of the decision
maker; It is concerned with human decision-making processes; It is investigated on the basis of empirical knowledge of the
capabilities of the human mind, and thus on the basis of psychology research.
Humans have limitations of both: Knowledge and computational capcity:
For discovering alternatives; Computing their consequences under certainty or uncertainty; And making comparisons among them.
Part I: (ii) Bounded Rationality Theory
Theories of Bounded Rationality [Simon, 1997a]: Can be generated by relaxing one or more of the
assumptions of the SEU theory.
New assumptions subsume that: Alternatives are not simply given, and thus they have to be
generated by some processes; probability distributions of outcomes are unknown, and thus they
have to be estimated by some procedures; Satisfactory is used rather than optimal or maximal standards; Probability distributions are unknown and they cannot be
estimated due to the sources uncertainty - like vagueness, instead of ambiguity.
Part I: (iii) The Genesis of Bounded Rationality
• Bounded rationality emerged with the advent in cognitive psychology research (Bruner and Piaget), and thus cognitive science and artificial intelligence along the 1950’s [Newell and and Simon, 1972].
• Cognitive psychology deals with high mental processes, rather than with stimuli and responses of behaviourism.
• Cognitive psychology aims the scientific research on models of human mind and its processes like perception, attention, categorisation, concept formation, knowledge representation, memory, language, probelm solving, decision making - among others.
Part I: (iv) Economic Decision-Making and Approximate Reasoning
Bounded Rationality is: Synonymous of Economic Decision-Making.
Since it concerns the use of cognitive processes to the achievement of low solution cost, robutness, and tractability to the reality.
Agents have cognitive limitations, but they are also constrained by time and space.
Humans [Zadeh, 1965 and 1973]: Have a remarkable ability for reasoning in complex
environmnets, under uncertainties, where information is ill-defined, incomplete, or lacking in reliability.
Human reasoning is approximate rathen than exact (driving a car in a havy traffic, sharing stocks, and so on).
Part I: (iv) Economic Decision-Making and Approximate Reasoning
Eg.- Parking a car- Travelling sallesman problem
Solutions for Travelling Sallesman Problem
Numbe of Cities Accuracy of Solution
Computing Time
100,000 1.00% 2 days
100,000 0.75% 7 months
1,000,000 3.50% 3.5 hours
[Zadeh, 1994] - Source: New York Times
Part I: (v) Organisations and ConflictsOrganisations of Today: (i) The members of organisations are decision makers and problem solvers [March and Simon, 1993].(ii) Processes of decision-making and problem solving involve trade-offs among alternatives
characterised by uncertainties and incompleteness of information, and hence they lead organisational members to both intra-individual and group conflicts.
(iii) The intra-individual and group conflicts that arise in organisations as exposed in (ii) are determined by cognitive limits of humans, and thus these conflicts cannot be solved by incentive and reward systems. Such cognitive limitations are synonymous of bounded rationality.
(iv) The members of organisations have motives that differ from organisational goals. (Use of incentive and reward systems for alignment and equilibria).
(v) Organisations shape participants’ behaviour through social structure, technology, and goals, and participants shape organisations through their behaviour, motives, and cognitive skills.
(vi) The environment shapes organisations (i.e. their social structure, technology, goals, participants, and behaviour), through its sources of complexity and uncertainty, but also through information, services, goods, and so technology.
(vii) Organisations also shape the environment through the same means.
Part I: (vi) Conclusions
Bounded rationality theories complement classical theories on rationality, but they also extend them to the analysis of human decision-making behaviour as it happens in real-world (everyday) situations. New approaches of decision analysis has to be considered in order to coupe with uncertainties that do not lie with statistical and analytical tools as applied to rational choices under certainty and risk (probabilities).
Part II:Solution Design
Part II: (i) Proposal CTP [Zadeh, 2001]CTP - Computational Theory of Perceptions
• Humans have a remarkable capability to perform a wide variety of physical and mental tasks without any measuments, and so any computation of numbers:
• Parking a car; • Playing golf;• Cooking a meal;• And summarizing a story.
• Instead, humans use information which are formed from perceptions – like information of time, distance, colour, lenght, spped, possibility, likelihood, truth, and so on.
Part II: (ii) Cognitive Psychology ModelsPerceptual Symbol Systems
[Barsalou, L.W., 1999] Perceptual Symbol Systems. Behavioral and Brain Science, 22.
Figure 1: Perceptual Symbol Systems
Analogue Modal Symbols
Extraction
Reference
Perceptual States (si=1,...,M)
Perceptual Symbols (pi=1,...,M)
Part II: (ii) Cognitive Psychology ModelsAmodal Symbol Systems (Information-Processing Systems)
Perceptual States (s i=1,...,M )
Arbitrary Amodal Symbols
New Representational Structure
Transduction
Reference
(CAR = y i=1,...,N) (wheels = y1,...,4)
(doors = y5,6) (colour = y7)
…
[Barsalou, L.W., 1999] Perceptual Symbol Systems. Behavioral and Brain Science, 22.
[Newell, A. and Simon, H.A. 1972] Human Problem Solving. Prentice-Hall.
Figure 2: Amodal Symbol Systems
Part II: (iii) A Model of Information-Processing Systems
[Newell, A. and Simon, H.A. 1972] Human Problem Solving. Prentice-Hall.
Figure 3: A Model of Information-Processing Systems
Receptors EffectorsProcessor
Memory
Environment
Part II: (iii) CTP - receptor
• CTP concerns a collection of description of perceptions expressed in a natural language.
• Examples: • It is unlikely that there will be a significant increase in the price
of oil in the near feature.• Diana is young.• Traffic is heavy.• Inflation is low and stocks are a little cheaper.• Most Swedes are tall.• Usually Robert returns from work at abot 6 pm.
Part II: (iii) CTP - receptor
• Natural language involves linguistic variables:• Inflation = [very high, high, not very high, moderate, low,...]• Cost = [expensive, cheap]• Age = [very young, young, middle age, old, very old]• Status = [rich, not so poor, poor]
Part II: (iv) Knowledge Representation Membership Functions of Fuzzy Sets [Zadeh, 1965]
(s)
1
0.5
0 105 Inflation (%)
low high
inflation = {low, high}
(s)
1
0.5
0 200100 Cost (US$)
low high
cost = {cheap, expensive}
R1: If inflation is low THEN cost is cheap
R2: If Inflation is high THEN cost is high
Part II: (v) Computing Perceptions for Decision-Making
• IF-THEN rules:• R1: IF incentives are high AND production is efficient
THEN organisational satisfaction is moderate.• R2: IF incentives are low AND production is efficient
THEN organisational satisfaction is moderate.• R3: IF incentives are high AND production is poor THEN
organisational satisfaction is moderate.• R4: IF incentives are low AND production is poor THEN
organisational satisfaction is bad.
Deriving conclusions from fuzzy rules of inference
Part II: (vi) Conclusions
Fuzzy sets and fuzzy logic are new approaches that explore uncertainties in decision-making processes by using natural language based information;They support CTP and they emerged as a new approach to deal with complex problems as those found in social sciences;They were proposed to fulfil the gap between analyses of non-living (machines) and living systems (behavioural) [Zadeh, 1962].
References1. [Barsalou, L.W., 1999] Perceptual Symbol Systems. Behavioral and Brain Science, 22.2. [March, J.G. 1994] A Primer on Decision Making: How Decisions Happen. The Free Press.3. [March, J.G. and Simon, H.A. 1958] Organizations. 1st Ed. John Wiley & Sons, Inc.4. [March, J.G. and Simon, H.A. 1993] Organizations. 2nd Ed. John Wiley & Sons, Inc.5. [Newell, A. and Simon, H.A. 1972] Human Problem Solving. Prentice-Hall.6. [Simon, H.A. 1997a] Models of Bounded Rationality: Empirically Grounded Economic
Reason. Vol.3. The MIT Press. (1st Ed. publisged in 1947).7. [Simon, H.A. 1997b] Administrative Behavior: A Study of Decision-Making Processes in
Administrative Organizations. The FREE Press.8. [Zadeh, L.A. 1962] From Circuit Theory to System Theory. Proceedings of the IRE, 50:
856-865. 9. [Zadeh, L.A. 1965] Fuzzy Sets. Information and Control, 8: 338-353. 10. [Zadeh, L.A. 1973] Outline of a New Approach to the Analysis of Complex Systems and
Decision Process. IEEE Transactions on Systems, Man, and Cybernetics, 3 (1): 28-44.11. [Zadeh, L.A. 1994] Soft Computing and Fuzzy Logic. IEEE Software, November: 48-56.12. [Zadeh, L.A. 2001] A New Direction in AI: Toward a Computational Theory of
Perceptions. AI Magazine. Spring: 73-84.