1 / 53

Metacognition in Computation: A selected history

Metacognition in Computation: A selected history. Michael T. Cox BBNT Cambridge. Why Metacognition?. “What then can be the purport of the injunction, know thyself? I suppose it is that the mind should reflect upon itself.” -- Augustine, De Trinitate, 16 th century. Why Metacognition?.

jerzy
Télécharger la présentation

Metacognition in Computation: A selected history

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Metacognition in Computation:A selected history Michael T. Cox BBNT Cambridge

  2. Why Metacognition? “What then can be the purport of the injunction, know thyself? I suppose it is that the mind should reflect upon itself.” -- Augustine, De Trinitate,16th century 2

  3. Why Metacognition? • Separates us from the rest of the species • Separates smarter people from less smart • Provides a heuristic basis for decisions • E.g., I am good at home repair, so I can risk embarrassment by volunteering to fix the broken pipe rather than calling a plumber. 3

  4. Metacognition is Ubiquitous 4

  5. Why NOT Metacognition? • Complexity – space and time • Actual human limitations • Easier to show when metacognition does not work rather than how it does • AI hype 5

  6. AI Hype “Once self-description is a reality, the next logical step is self-modification. Small, self-modifying, automatic programming systems have existed for a decade; some large programs that modify themselves in very small ways also exist; and the first large fully self-describing and self-modifying programs are being built just now. The capability of machines have finally exceeded human cognitive capabilities in this dimension; it is now worth supplying and using meta-knowledge in large expert systems.” -- Lenat, Davis, Doyle, Genesereth, Goldstein, and Schrobe 1983 (p. 238) 6

  7. What is Metacognition? • Meta-X is defined as “X about X” • Metacognition is cognition about cognition • Metareasoning is reasoning about reasoning • Metaknowledge is knowledge about knowledge • Metamemory, metarepresentation, metacomprehension, metalogic, metaplans,meta... 7

  8. But what about… • Meta-levels • Reflection • Introspection • Self-awareness • Self-explanation • Consciousness? 8

  9. Outline of Presentation • Psychology, Metacognition, and Human Behavior • Cognition and Metacognition • Problem Solving and Metacognition • Metamemory • Artificial Intelligence, Metareasoning, and Introspection • Logic and Belief Introspection • Knowledge-Based Systems, Metareasoning, and Control • Model-Based Reasoning, Case-Based Reasoning and Introspective Learning 9

  10. What is Not Being Covered? • Social Psychology • Philosophy • Cognitive Neuroscience • Consciousness Studies • Theological Accounts 10

  11. Psychology, Metacognition, & Human Behavior • Earliest Research –circa 1900 • Brown, A. 1987. Metacognition, Executive Control, Self-regulation, and Other More Mysterious Mechanisms. In F. E. Weinert and R. H. Kluwe eds. Metacognition, Motivation, and Understanding 65-116. Hillsdale, NJ: LEA. • Text comprehension and metacognitive activities studied, but under different names. • Not to be confused with introspectionism 11

  12. Cognition & Metacognition • Earliest Research– John Flavell • Flavell, J. H. 1971. First Discussant’s Comments: What Is Memory Development the Development of? Human Development 14: 272-278. • Flavell, J. H. 1976. Metacognitive Aspects of Problem Solving. In Resnick ed. The Nature of Intelligence, 231-235. Hillsdale, NJ: LEA. • Much of the work has been in the child development and cognitive aging research communities 12

  13. Wellman’s Theory • Wellman, H. M. 1992. The Child's Theory of Mind. Cambridge, MA: MIT Press. • Children come to construct a naive theory of mind by observing the difference between self and others • Before age of three – no distinctions • At age of three – difference between wants and actions; between ideas and reality • After age of three – mind as processor and interpreter 13

  14. Metacognitive Variables • Person Variables • Deals with the individual and others • Cognitive psychologists can recall many facts about cognition • Task Variables • Concerns the type of mental activity • Harder to remember nonsense words • Strategy Variables • Alternatives to mental tasks • To remember a list it helps to rehearse 14

  15. Theory-Theories • Theories have • Coherence • Ontology • Causal-explanatory structure • Theories of Mind • Mental constructs self-refer • Distinctions between things and ideas • Desires and beliefs effect thought and action 15

  16. Children’s Theory of Mind • Different than adults • More than just a collection of facts about mental phenomena • Experiment for 3-yr-olds • Sally sees candy put in a bag • Sally leaves room • Candy placed in a drawer • Sally returns, where will she look for candy? 16

  17. Problem Solving & Metacognition • Earliest Research – Dörner • Dörner, D. 1979. Self-Reflection and Problem-solving. In F. Klix ed. Human and Artificial Intelligence, 101-107. Amsterdam: North Holland. • Subjects that introspect versus those that do not with all else the same • Surprisingly very little work altogether 17

  18. Derry’s Theory • Comprehensive model of reflective problem-solving for mathematical word problems • Inspired by ACT* (Anderson) • Four basic phases • Clarifying a problem • Developing a strategy • Executing a strategy • Monitoring/Checking 18

  19. Derry’s Experiment • Computer-based instructional system • Teaches math word problems to military personnel and college students • Assumption: Goal backchaining and MEA form bases for human problem-solving • Gather subject protocols during testing • Categorize protocols and correlate with performance 19

  20. Derry’s Results • Neither group had linear performance • Protocols fell into clarification and execution • Considered checking math answers as metacognitive 20

  21. Swanson’s Research • Separate cognitive abilities from metacognitive abilities • Standardized test scores and meta-cognition questionnaires (Hultch, Hertzog, Dixon, & Davidson 1988) • Measure problem-solving performance • High/Low aptitude interacts with high/low metacognition 21

  22. Metamemory • Earliest Research– John Flavell • Flavell, J. H., and Wellman, H. M. 1977. Metamemory. In R. V. Kail, Jr., and J. W. Hagen eds. Perspectives on the Development of Memory and Cognition, 3-33. Hillsdale, NJ: LEA. • Knowledge of and memory for memory 22

  23. Knowing without Remembering 23

  24. Kausler and Lovelace’s Theory Kausler • Off-line memory evaluation • On-line memory evaluation Lovelace • Pre-performance estimates (predictions) • On-line memory monitoring • FOKs • Postdictions • Reality monitoring • JOLs • Memory performance monitoring 24

  25. Information Processing Models • T. Nelson and Narens • Information processing view • Monitoring component and control component • M. T. H. Chi and K. VanLehn • The self-explanation effect • P. Pirolli and M. Recker • Soar model + GRAPES • Subjects that explained between experiments tended to learn better 25

  26. Reder’s Theory • Game show paradigm • Two stages of memory retrieval exist • Fast familiarity judgment • Slower search stage • Conclusion • Correlated with question terms not answers 26

  27. SAC Model • Source of Activation Confusion • Spreading activation model • Base-line strength change as the power function • Activation change due to neighbors is 27

  28. AI, Metareasoning & Introspection • Earliest Research – Minsky and McCarthy • Minsky, M. L. 1965. Matter, Mind, and Models. In Proceedings of the International Federation of Information Processing Congress 1965 (Vol. 1) 45-49. • McCarthy, J. 1959. Programs with Common Sense. In Symposium Proceedings on Mechanisation of Thought Processes (Vol. 1), 77-84. London: Her Majesty’s Stationary Office. • Models of Models • Declarative Knowledge for the Self 28

  29. Minsky’s Theory • Minsky, M. L. 1965. Matter, Mind, and Models. In Proceedings of the International Federation of Information Processing Congress 1965 (Vol. 1) 45-49. • To answer questions about the world and the self in the world, an agent must have a model it can query • W, M, W*, M*, W**, M** 29

  30. 30

  31. McCarthy’s Theory • McCarthy, J. 1959. Programs with Common Sense. In Symposium Proceedings on Mechanisation of Thought Processes (Vol. 1), 77-84. London: Her Majesty’s Stationary Office. • Knowledge as logic • Logic as thinking • What does it mean for a robot to be conscious? 31

  32. Logic & Belief Introspection • Self-Reference and “aboutness” (Perlis) • Liar’s Paradox from time of Socrates • This sentence is false. • FOL axiomization and possible worlds (Moore) • Belief is different than facts (Hintakka) • Model-Theoretic reasoning • Metalogics and proving provability 32

  33. Konolige’s Deduction Model • Alternative to Possible Worlds Semantics • Deduction Structure is a mathematical abstraction of bounded belief systems • Machines and introspective machines • Intrinsic and extrinsic self-beliefs • Separation of IM from M resolves some problems of self-reference 33

  34. Logical Representations • Is-Complex-wrt (John, ) • How to handle? 34

  35. Knowledge-Based Systems, Metareasoning & Control • Earliest Research – Metaknowledge in expert systems • Barr, A. 1977. Meta-Knowledge and Memory, Technical Report, HPP-77-37. Stanford University, Department of Computer Science, Stanford, CA. • Davis, R. 1976. Applications of Meta-Level Knowledge to the Construction, Maintenance, and Use of Large Knowledge Bases. Stanford HPP Memo 76-7. Stanford University. • Metarules: the redherring of AI 35

  36. Davis’ Theory • Knowledge engineering in MYCIN • Metaknowledge • Schemas • Function templates • Metarules • Rule models • Rule models help interpret what expert asserts TEIRESIAS Performance Program Domain Expert knowledge transfer Inference Engine explanation Knowledge Base 36

  37. Example Rule Model INVESTMENT-AREA-IS • Examples ((rule116 0.3) (rule050 0.7) (rule037 0.8) (rule095 0.9) (rule152 1.0)) • Description • Premise ((returnrate same notsame 3.8) (timescale same notsame 3.8) (trend same) ((returnrate same)(timescale same) 3.8) … • Action ((investment-area conclude 4.7) (risk conclude 4.8))… • More-general (investment-area) • More-specific (investment-area-is-utilities) 37

  38. Model-based Understanding & Learning by Experience Expert (dialog) Knowledge Base (knowledge acquisition) Rule Acquisition (concept formation) (model-directed understanding) Rule Models 38

  39. Problems with Expert Systems • Confuses abstraction with metacognition • Confuses control with metacognition • Self-understanding software tangent? • Explanation is not a rule chain or proof tree • Knows what it does not know? 39

  40. Metareasoning • Earliest Research– Bounded rationality • Simon, H. A. 1955. A Behavioral Model of Rational Choice. Quarterly Journal of Economics 69: 99-118. • Earliest Research – Good’s Type II rationality • Good, I. J. 1971. Twenty-Seven Principles of Rationality. In V. P. Godambe and D. A. Sprott eds. Foundations of Statistical Inference. Toronto: Hold, Rinehart, Winston. 40

  41. Wesfald’s Theory • Treating computation selection as action selection by maximizing expected utility • Cost of time (world changes by itself) • Benefit of better action choices • Execution cost • Resource cost 41

  42. Unifies decision-making systems Decision-theoretic systems Production systems Goal-based systems Reactive systems EBL systems Unifies meta-cognitive systems MRS (Genesereth) TEIRESIAS (Davis) Soar (Newell) System Unification 42

  43. Decision Stages and Shortcuts A Condition (s) B Condition (result(a,s)) C E F Utility (result(a,s), v) DT D Best (a,s) 43

  44. Model-based Reasoning & CBR • Earliest Research – Schank’s emphasis on memory and representation • Schank, R. C., Goldman, N., Rieger, C., and Riesbeck, C. K. 1972. Primitive Concepts Underlying Verbs of Thought (Stanford Artificial Intelligence Project Memo No. 162. Stanford, CA: Stanford University, Computer Science Department. (NTIS No. AD744634) • Using Conceptual Dependency primitives to represent remember, forget, think, expect… 44

  45. Case-Based Explanation • Provides a framework for interpreting Failures • In world actions • In reasoning actions (e.g., memory retrieval) • In social actions • Example: Dog barking story • S1 Police & Dog enter airport baggage area • S2 Dog sniffs luggage. • S3 Dog Barks at luggage. • S4 Police arrests suspect. 45

  46. Computational Introspection • To reason about the self… • When reasoning about the world fails use meta-reasoning to explain the failure • Map from symptom of the failureto the cause of the failure • Learn 46

  47. Symptoms of Failure Expectationexists Expectation does not exist Actual event exists Contradiction Unexpected Success Impasse Surprise FalseExpectation Actual eventdoes not exist 47

  48. Missing Incorrect Correct Causes of Failure 48

  49. Stranded Motorist Example • Planning a vacation • Destination • Reservation • Supplies • Gas • Plan Execution • Goes to store • Buys supplies • Drives to mountains • Runs out of gas • Failure Recovery • Get gas can • Walk to gas station or hitch-hike • Fill can with gas • Return to Fill tank • Continue • Failure Repair • Regress goals to features in initial state • Use features as index to store as new case • Alternate Repair • Reason about the reasoning that led to the failure • Cause was forgetting to fill car with gas at store • Form association between going on long trip and checking gas gauge Causal Possibilities Tree • Sub-goals • Be at store • Make purchases 49

  50. Forgetting to Fill-Up with Gas 50

More Related