Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Overview of the Adaptive Problems (Pp. 300-301) • How do we communicate with others? • How do we classify and categorize objects in our world? • How do we solve problems and reach goals? • How do we make decisions when confronted with a set of alternatives? 2 Communicating with Others: Overview (Pp. 302-303) The Structure of Language • Structure of Language • What sets language apart from other communication systems: Grammar – Language rules: Grammar • Includes phonology, syntax, and semantics – The hierarchical structure of language: phonemes, morphemes, words, phrases and sentences – The structure of sentences – Set of rules that allow the communicator to combine arbitrary symbols to convey meaning – Three aspects: • Surface structure and deep structure • • • • (Pp. 303-304) Language comprehension Language Development in Children Language in Nonhuman Species Is language an adaptation? • Phononlogy: Rules for word sounds • Syntax: Rules for combining words to make sentences • Semantics: Rules used to communicate meaning 3 The Hierarchical Structure of Language (Pp. 304-306) 4 The Units of Language (P. 305) • Phonemes: – Smallest significant sound units in speech • Example: “ee” as in “feet” • Morphemes: – Smallest units of language that carry meaning • Examples: “do,” “un” • Word, phrases, and sentences – Words combine to make phrases • Example: “the interesting class” is a noun phrase Figure 8.3 5 6 1 The Structure of Sentences Surface Structure and Deep Structure (P. 306) (P. 306) • Rules of syntax determine how words combine into phrases, and phrases into sentences – Set of rules used to do this isn’t known • Chomsky’s idea of how sentences work: – Surface structure: Superficial appearance, literal ordering of words – Deep structure: Underlying representation of meaning – Producing sentences requires transformation of deep structure into a surface structure Figure 8.3 7 Language Comprehension (Pp. 306-307) • How do we decide what another person is trying to communicate? 8 Language Development (Pp. 307-308) • Is language a product of genes or experience? – Many researchers believe babies have some inborn preparation for language – Babies follow similar milestones all over the world – Understanding speech depends on top-down as well as bottom-up processing – Communication depends on common knowledge among speakers • Babies are born producing phonemes appropriate for many languages, but soon narrow these down • Pragmatic rules: How practical knowledge is used to comprehend speaker’s intention, produce an effective response – Example pragmatic guidelines (Grice): Be informative, tell the truth, be relevant, be clear – By 3-5 weeks: Cooing – By 4-6 months: Babbling – By 6-18 months: Sounds become more language-like 9 The Beginnings of Communication (Pp. 308-309) 10 Language in Nonhuman Species (Pp. 310-311) • Nonhuman animals definitely communicate • Approximate ages for language milestones: – But recall: not all communication is language – 1 year: Simple words – 2 years: Vocabulary of 200 words • Attempts to teach chimps to speak failed • Signs/symbols communication in chimps: • Comprehension develops even faster; commands and statements are understood – Washoe: Uses about 160 signs – Sarah: Uses plastic shapes to make “sentences” – Kanzi: Understands speech over headphones – 3 years: Telegraphic speech • Grammatical knowledge fine-tuned from 3 to about 6 or 7 • It is really language? – Preschoolers tend to over generalize the rules – Can they generate new combinations? – Can they learn from other chimps? 11 12 2 Is Language an Adaptation? (P. 312) • Why don’t most species use language, if it’s so beneficial? Classifying and Categorizing (P. 313) • Category: Class or objects that most people agree belong together – One view: Natural selection caused this special ability to develop – Another: Developed because we have large brains, generally sophisticated thinking ability – Being able to categorize is adaptive • Allows us to infer properties, even if we can’t see them directly • Important questions about categorizing: • Evidence for adaptation view includes special brain regions for language, specially developed vocal tract – What properties about an object can make it to a particular category? – Do we form abstract category representations? – Are categories organized into hierarchies? – However: Fossil record can’t show how or when it developed, or why 13 Defining Category Membership 14 Natural Categories Do Not have Defining Features (P. 314) (Pp. 313-314) • Example: You know that Monopoly is an example of the category “game,” but why? • Defining features view: Categories defined by features that all members share – But: Many categories don’t have features shared by all; boundaries are fuzzy • Family resemblance view: Members of a category share certain core features, but not all members have to have all these features Figure 8.4 15 Fuzzy Category Boundaries 16 Abstract Category Representations: Prototypes (Pp. 316-317) (P. 315) • Prototype: – Best or most representative member of a category – Example: What is the best example of the category “fruit?” – Could categorize by comparing to prototype • Do we really store prototypes? Figure 8.5 – Alternative: Store all examples of the category • Could categorize by comparing to examples (exemplars) 17 18 3 Prototypes versus Exemplars The Hierarchical Structure of Categories (P. 318) (P. 317) • Most objects can be categorized in several ways – Example: Monopoly > “board game,” “game,” “activity” – How do we organize different categories? • Basic-level category: Used most often, is most useful and predictive • Superordinate categories: More general, less descriptive • Subordinate-level categories: Very specific Figure 8.6 19 Category Hierarchies 20 Solving Problems (P. 318) (Pp. 319-320) • Two kinds of problems: – Well-defined: Goal and starting point are clear; you know when it’s been solved • Researchers mainly study well-defined problems – Ill-defined problems: Goal and starting point are unclear; hard to tell when solution is reached • Many real-life problems are ill-defined – Main question: What thought processes, strategies are used in problem-solving? • Probably many similarities between well- and ill-defined problems Figure 8.7 21 Guidelines to Problem Solving: The IDEAL Problem Solver (Pp. 320-321) • IDEAL (Bransford and Stein): Hypothetical model of problem-solving steps 22 Identifying and Defining the Problem (Pp. 321-322) • Problem representation: Your understanding of what information is given and how that information can be saved – Identify the problem: Recognize the signs that a problem exists – Define (represent) problem in an efficient way: What kind of problem is it, in your mind? – Explore a variety of problem strategies – Act on the problem strategy you choose – Look back and evaluate whether the strategy was effective, – If not, try something new – Must detect which information is relevant, which is not • A problem that can arise: Functional fixedness – Tendency to see objects and their functions in fixed typical ways – Hinders problem solving causing failure to identify and define all the available “tools” to solve a problem 23 24 4 The Maier Two String Problem The Duncker Candle Problem (P. 322) Figure 8.8 25 Exploring and Acting Figure 8.9 Common Heuristics (Pp. 323-324) • Two classes of strategies: (P. 323) 26 (Pp. 324-325) • Means-end analysis – Algorithms: Step-by-step rules or procedures that guarantee a solution – Heuristics: Problem solving “rules of thumb” • Shortcuts that are efficient, but don’t guarantee solution • A problem that can arise: Mental set – Find actions (means) that reduce the gap between the current starting point and goal (end) – Usually requires breaking down problem into sub goals • Working Backwards – Tendency to rely on particular problem-solving strategies that were successful in the past – Start at goal state, move toward starting point • Searching for Analogies • When new problems can’t be solved using old strategies > Failure – Find a resemblance between current problem, one solved successfully in the past – Can be reduced by taking a break from the problem 27 Framing of Decision Alternatives (P. 328) • Framing: Way the alternatives are structured 28 Decision-Making Biases (Pp. 328-329) • Confirmation bias: Tendency to seek out and use information that supports and confirms a prior decision or belief – Example: Is a possible course of action framed as a way to ensure a gain, or avoid a loss? – People tend to avoid risks when gain is emphasized, take risks when loss is emphasized – People avoid seeking out information that might contradict a prior belief • Framing can lead to choices that are irrational from a statistical viewpoint • Belief persistence: Tendency to cling to initial beliefs even when confronted with disconfirming evidence – Example: Doctors are more likely to choose a treatment when they see it as preventing death as opposed to extending life – People tend to try to find reasons why beliefs could still be true, even with contradictory evidence 29 30 5 Representativeness Availability (Pp. 329-330) • When judging likelihood of something falling into a class, compare the similarity of that thing to the average member of that class – Example: Which is probably a random series of coin flips, H H H TTT or TTHTH? (Pp. 330-331) • Base estimates on odds of an event occurring on ease with which examples of the effect come to mind – Example: Diseases that get a lot of publicity are estimated to be more common than other diseases – Example: You believe it’s more likely that you will do the dishes than your roommate – Both are equally likely, but one is more representative • Mistakes that can result from representativeness: – Ignoring the base rate – Conjunction error • You remember all the times you did the dishes, but fewer of the times your roommate did them 31 Anchoring-and-Adjustment 32 Are Decision-making Heuristics Valuable? (P. 331) (P. 331) • Research on heuristics highlights how imperfect we are at decision making • Judgments are influenced by starting points, such as initial estimates – However: Research focuses on special situations where heuristics contradict optimal, statistically based reasoning – Most of the time, they are effective shortcuts – Example: What percent of African countries belong to the United Nations? • “More than or less than 65?” > High estimate • “More than or less than 10%” > Lower estimate • Good things about heuristics: – Often do produce good decisions in real life – Save time and effort over optimal reasoning strategies – Often, we don’t have the statistical information for optimal reasoning anyway – Example: What is a fair price for a new car? • Consider the sticker price first > Higher estimate • Consider what a similar car costs used > Lower estimate 33 Solving the Adaptive Problems 34 Solving the Adaptive Problems (P. 333) (P. 333) • Solving Problems: • Communicating with others: – Humans have developed rules (grammar) for combining sounds, words, and meanings in ways that allow them to communicate wit others – May be adaptation that is unique to humans – Five main steps are involved: identify, define, explore, act, look and learn – Common pitfalls include functional fixedness and mental set • Making Decisions: – Decisions involve choosing from among alternatives, weighing risks – Our choices may depend on how choices are framed – We tend to rely on heuristics including representativeness, availability, anchoring and adjustment • Classifying and categorizing: – People rely on family resemblance to make categorization judgments – Categories are structured hierarchically, and people tend to rely on the basic level most of the time 35 36 6