Asg3-4-01 to -5-04.

Higher mental processes: language; thinking Higher mental processes Language: Reasoning: Concepts: Thinking:

Rule based: Nativist(?) “built in” deductive form al algorithm s

Exam ple based: Em piricist learned inductive “fuzzy” heuristics

Human Language: - flexible rule-governed [?] system - arbitrary sym bols - Phonem es (speech sounds), letters - com bined to m ake words - express indefinite # of m eanings - put together sm all # of units using rules[?] Communication is a social behavior Given-new contract: speaker takes into account what listener knows - context, stress, syntax - description of new objects Levels of language: - physical stim ulus: com plex tim e-varying pattern of sounds - phonem e: sm allest unit of speech that can affect m eaning - syllable: group of phonem es that “go together” - word: One or m ore syllables that stand for som ething - syntactic (‘gram m atical”) structure: set of rules that define acceptable com binations of words. Rules at all levels. Phonetic rules: certain speech sounds go together srodk - srock bm ag - blag bsole - brole - Infants learn word boundaries from rare letter com binations W ordm gram m ar level rules M eaning depends on: - words - gram m ar = set of syntactical rules [?] for arranging words - pragm atics: - prosody: tone of voice, em phasis, etc. - m etaphor: “drowning in m oney” Gram m ar independent of word m eaning "Colorless green ideas sleep furiously." - gram m atical, but sem antically anom alous, sentence Syntactic (“gram m atical”) cues give m eaning - word order The dog bites the m an. The m an bites the dog. - inflection (“endings”) Der hund beisst den m an Den hund beisst der m an

gkuse - gluse

p. 1

Asg3-4-01 to -5-04. - word class (noun, adverb, etc.) content (open class) words: - nouns, verbs, adjectives, adverbs - have referents function (closed class) words - prepositions, articles,“linking” - show relation am ong words differences between content & function words show up in m any ways: - brain dam age effects - Broca’s: gram m ar defect: lose function words - W ernicke’s: vocabulary defect: lose content words - word association learning

People remember meaning: deep structure not form: surface structure

Grammatical structure is psychologically real: - affects other m ental functions: - perception: clicks m oved to gram m atical boundaries - m em ory: recall of com plex sentences fails at gram m atical boundaries - thinking? Sim ulate gram m atical structure with neural nets? Reading - phonetic decoding - whole word recognition eye m ovem ents during reading good readers: - short fixations, few direction reversals - whole word recognition for m ost words - fixate m ore on content words, im portant words - efficient use of short term m em ory Reading problem s: - lim ited vocabulary - inefficient use of phonological working m em ory Dyslexia: reading problem , norm al intelligence - perception? - auditory-visual sym bol m atching? - deficit in phonem e perception? Voice Onset Tim e (VOT) Language development - “shape” of language - phonem es: newborn perceive (all?) phonem es - 6-8 m o.: babbling like speech sounds

p. 2

Asg3-4-01 to -5-04. - add sounds in fixed order: easy first - lose unused phonem es - find syllable boundaries; identify words: statistical regularities - som e sounds often go together in words; other rarely go together - words - 1 year: first words: consonant + soft vowel (/da/) - under-, over-extend m eaning - ~18 m onths: word learning explosion - gram m ar: discovered a. two words + context + gesture: not adult gram m ar b. easiest rules; m ost im portant rules - overextension of rules from regular to irregular Learning the specific vocabulary & gram m ar of your language: - m odeling adult speech - good m odel? - child-directed talk: Rule-based gram m ar build into the brain? - language universals = innate rules? - learn fast from poor[?] m odels - no conscious effort or availability - too com plicated to learn [is a rule-based gram m ar what they learn?] - develop it in sam e order [only very early stages] - new languages develop fast Alternative ideas about gram m ar: - m ap on realities of world - part of “lexicon” (~m ental “dictionary”) - gram m ar not form al (rule-based)but statistical - speech sound parsing - adults unconsciously “learn” sim ple artifical “gram m ars” - adults do help toddlers learn language - neural network Critical periods? - effects of language deprivation Thinking Reasoning: - inductive: exam ple based - deductive: form al = rule based Exam ple: m etals - Inductive: all m etals I have put in fire don’t burn: @ m etals don’t burn - Deductive: if you find a m etal that burns, then the rule is wrong. Deductive reasoning: general principles $ specific instances - syllogism s Inductive reasoning: specific instances $ general conclusion - never know whether there are exceptions to rule Concepts: category, m em bership rules - Concept form ation: - formal concepts: rule-based - test hypotheses (rules)

p. 3

Asg3-4-01 to -5-04.

Problem : are m ost concepts form al, rule based? - concepts develop unconsciously: Hull, Reber - m any everyday concepts are “fuzzy” - system atic, logical analysis is not “natural” way of thinking - m any things are not “com putable” Heuristics: "Natural" (everyday) concepts: - boundaries often fuzzy - exem plar-based - often very efficient, but can m ake m istakes Problem solving and decision making: Formal or heuristic? Formal: - algorithm : set of rules for best decision - if properly developed - if applied to appropriate problem . Problem : Many things cannot be “com puted” Heuristic: m ental short cuts - "rule of thum b" - efficient: can be as good or better than full use of all inform ation - apply to any kind of problem Problem : heuristics often(?) m akes errors Som e heuristics and their problem s: Availability: use what's easily available - recall last few cases þ bias Gambler’s fallacy: Regression fallacy: extrem e events are noticeable, but unusual - "hot hands", "hot bat" are statistical flukes Overestimation, illusory correlation Representativeness Confirmatory bias: - All cards have triangle on one side & circle on other. - W hich cards m ust be turned over to test whether cards with red triangles have green circles on the other side? Does this logic apply in the real world? - crashing noise from kitchen.

Base rate fallacy - Suppose - 1 person in 1,000 has AIDS - a test for AIDS is 95% accurate (5% false alarm s) Ron tests positive. W hat is probability that Ron has aids? 95%? 85% ? 50% ? 15% ? 2% ?

p. 4

Asg3-4-01 to -5-04.

Language and thought: - W horf: linguistic determ inism - language structure shapes perception & thinking - languages have very different words, gram m ars Tests of linguistic determ inism - do differences in vocabulary affect perception? - words for snow & ice - words for color - Language related to color m em ory Different vocabularies, gram m ars im pose different m ental loads - words with several m eanings - word order vs. inflections - English sentences m ore likely to be am biguous out of context W eak version of linguistic determ inism m ay have support: - different gram m ars put different loads on working m em ory

p. 5