• communicative
• arbitrary symbolic
• regularly structured
• structured at multiple levels
• productive/generative/creative
• dynamic (new words added to dictionary)

-production

-comprehension

-complicated sentences can be comprehensible

-simple sentences can be incomprehensible

sentence

phrase

word

morpheme

phoneme 

-smallest unit of speech sound

-diff languages use diff phonemes

-each language has rules of order ("Fp" not okay)

-native speakers of a tonal language (Mandarin) are more likely to have perfect pitch

-maybe evidence that there is a critical period to learning perfect pitch like there is for learning language

-phoneme perception

-language perception is visual in addition to auditory 

-(top down processing)

-various amounts of "voicing" of a phoneme

-"ba" or "pa": percieve phoneme as being one or the other 

-location in the moutn and vowel height/location

-purpose: makes perceiving language possible across speakers and accents

-smallest unit that signals meaning

-prefixes, suffixes, roots

-can be entire words

-may not stand alone

-rules of a language say how they can be combined

-combinations of one or more morphemes

-a single word can have multiple meaning

-rules for word order 

-different languages have diff word order

-related to concepts and categories (LTM)

-words represent or symbolize concepts

-meanings of sentences: words must be grouped into phrases, relationship between words

-noun phrase and verb phrase

-ambiguous: results when the same surface structure can correspond to 1+ deep structures

-word has two different meanings

-"He was bothered by the cold"

-words can be grouped together into more than one phrase structure

"I saw the gorilla in my pajamas

-two different phrase structures

"they are (cooking apples)"

"they (are cooking) apples"

-the same word/phrase can refer to different objects

"John grabbed his lunch, sat on a rock, and ate it"

-When does CONTEXT have its effect?

1. "dumb retrieval": get all possible word meanings from lexicon, context then guides selection of correct meaning

2. "smart retrieval": context guides access to the lexicon, so only retrieve ONE meaning for a word

-"dumb retrieval" proved by Swinney Experiment

-condition 1: lexical decision immediately after ambiguous word (bug)

-results: initially get priming of both meanings

(ant, spy < sew)

-condition 2: lexical decision delayed from time of ambiguous word

-results: after time passes, "bug" primes only the relevant meaning

(ant<spy=sew)

-motor planning area (part of premotor cortex)

-contains the motor programs for words

-programs actually executed by primary motor cortex

-nonfluent, expressive, or motor aphasia

-difficulty in the PRODUCTION of language

-speech is non-fluent, slow, labored

-"telegraphic" speech: only use content words

-word finding difficulty

-comprehension intact

-abnormal intonation

-aware of problem 

aka. sensory or receptive aphasia 

-impairement in the COMPREHENSION of language produced by oneself or others

-speech is fluent but nonsensical (word salad)

-speech is syntactically and intonation correct

-speech full of phonemic and semantic errors

-neoligisms: made up words

-lack of awareness of language impairment 

-comprehend and generate millions of sentences

-know > 100,000 words

-no other species does this

-first job in language acquisition: distinguish language sounds from other sounds 

-then need to parse it into phonemes

-parse phonemes into words 

-once have words, need to assign meanings

-Problems: 

-same sound can refer to diff. things

-diff sounds can refer to same thing

-adults usually do not correct childrens grammar

-correcting grammar does not even help

-kids correct themselves despite little negative feedback

-acquistion task is not well constrained

-irregular mappings from sounds to meanings

-little negative feedback

-all normal humans aquire language

-suggests linguistic universals: general language principals, embodied in every language

-in 1st year, infants can discriminate all phonemes from all languages

-gradually lose discrimination that are not important to own language

-adults help kids by speaking "motherese"

-high pitch, slow rate, exaggerated intonations

-falling pitch and pausing signals phrase boundaries

-holophrastic (one word) stage

-telegraphic (two-word) stage

-learning syntax/rules (syntactic overgeneralization)

-one word utterances

-no syntax, need context to disambiguate

-do understand some phrases

-two word utterances

-correct use of word order:

-subject-action

-action-object

-can convey a lot of info briefly and clearly

-U-shaped curve for irregular past (went)

-initially use appropriate form (went)

-learn rule and overgeneralize (go-ed)

-relearn correct past tense (went)

-learn general rules that apply to new cases

-4-5 yr olds know plural of wug is wugs

-implies language learning is generative, not just imitation

-by 5, kids know 10,000 to 15,000 words

-input gives clues about parts/wholes

-inherent bias that new words refer to shape

-deterministic

-based on rules of logic, conclusions that must be true given the facts

-probabilistic

-conclusions that are possible, but not necessarily true

-normative: how one ought to reason (optimal based on logic or math)

-descriptive: how one actually reasons (describes what people really do)

-linear syllogism

-conditional (if, then)

-reasoning by lining up the facts, checking conclusion

-deduction by comparing how items relate 

(ex: X is taller than Y, Y is taller than Z...)

-normative theory: solve a longer problem by breaking it into smaller pieces

-prediction: longer the syllogism, longer the solution time (RT)

-teach subjects pairs of facts

-subjects respond T/F

-result: further apart the pair, faster reaction time 

-"distance effect"

-subjects dont reason normatively 

-based on if/then conditions

Ex: IF it is raining (the premise, antecedent)...

then the game is cancelled (conclusion, consequent)

Ex: The game is cancelled...what can we conclude?

-2 of 4 rules are VALID

-follow logic, always word

-2 of 4 are NOT VALID

-not based on logic, dont always lead to correct reasoning 

if P, then Q

P

Therefore, Q

(only 5% error rate)

If P, then Q

NOT Q

Therefore, NOT P

(40% error rate)

(conclusion does not follow logically)

If P, then Q

NOT P

therefore, NOT Q

(20% error rate, cant conclude anything)

(conclusion does not follow logically)

If P, then Q

Q

Therefore, P

-concrete condition (bar): 22/24 performed correctly

-abstract condition (postal): 2/24 subjects performed correctly

-possibly the rule needs to be familiar too...

-so concreteness and familiarity are not the whole story 

-we reason based on practical experience

-Ex: Permission Schema

-If a certain action is taken, THEN a certain condition must be met 

(if you are going to drink, you must be 21)

-Ex: Obligation schema

(if its Tuesday afternoon, then you have to come to class)

-standard social contracts

-if you take the benefit, you must may the cost

(cheaters take the benefit without paying cost)

-if you do that, i will kill you

(a threat)

-natural selection made our brains look for cheaters and threats 

-Grover high school problem experiment

-Result: people do better when looking for cheaters 

-in some conditions people do better if theyre mad

-males do better after playing soccer

-theories about why people do better on the real-world versions of the Wason task than abstract

-concreteness

-familiarity

-pragmatic reasoning schemas (permission/obligation)

-social contract schemas (cheaters/threats)

-using Bayes theorum is normative behavior

-Take 3 things into account:

1. the base rate: probability of hypothesis independent of evidence

2. The hit rate: probability of evidence if hypothesis is true

3. the false alarm rate: probability of evidence if hypothesis is not true

-descriptive reasoning

-we often ignore base rates (the overall prevalence of cancer)

-people do well in real world

-Theory: people rely on heuristics in judgements under uncertainty

-rules of thumb are good enough for everyday reasoning

-lab studies exploit weaknesses of heuristics 

-judge whether A has some characteristic by relying on the similarity of A to other things

-problem: focuses too much on current evidence and thus ignores base rates

-recent occurrence of an event reduces the perceived probability of that event 

-betting: wont bet on the lottery number that won in the last drawing

-things that are easily remembered/brought to mind are judged to be more prevalent than events that are less easily brought to mind

-When availability works:

-Subjects given category (Russian novelists vs. flowers)

-Result: correlation between estimate and actual retrieval nearly perfect 

-in general, frequent events are easier to recall or imagine than infrequent ones

But...

-availability is affected by various factors which are unrelated to actual frequency 

-media exposure

-stereotypes and their perpetuation 

-perspective can influence the availability of certain events: we remember what WE do more easily than we remember what OTHERS do

-Ex: Study of married couples

-highly related to availability heuristic

-base judgements on how easily you can imagine how things will turn out in the future, and how things would have turned out under diff. circumstances 

-Ex: Mr. Tees and Mr. Crane airport scenario

-easier to imagine ways to be 5 minutes earlier so Mr. Tees would be more upset

-the ease of imagining cause and effect

-mother/daughter eye color ex. 

-causal link between action and actor

-Helen car accident ex. 

-availability: who gets in more car accidents

-representativeness: in selecting job candidates, maintain the good-old-boys club

-simulation: which suspect can you simulate more easily having committed the crime

-when a choice is framed in terms of gains, people use risk-aversion strategy

-when a choice is framed in terms of losses, people use a risk-taking strategy 

-Framing effect: the same info presented in diff forms can lead to diff decisions

-decision influenced by how the problem is stated

-Ex: makes 75% of free throws > misses 25% of free throws 

-when making estimations, people start from a reference point (anchor) and make adjustments from that anchor

-effect doesnt disappear even with extreme anchors

-influences real-life decisions (jury decisions)

-once you've invested in something, you can't back out even if it's costing you more money

-ex: car/house requires extensive repairs--keep repairing instead of buying a new one

Humans: judgement based on informal contemplation and discussion with others

computers: automated, algorithmic procedures, statistics

-Conclusion: greater accuracy relying on computers rather than on humans 

-speedup with practice 

-rate of improvement diminishes with time

-it takes a VERY long time to achieve the small amount of improvement that separates really good from great

-learner aquires skill by following verbal instruction

-relies on declarative knowledge

-commit instructions to memory 

-often rehearse instructions as you try to perform

-controlled process, requires attention

-strengthen connections that lead to desired result

-get rid of actions that lead to errors

-feedback is very important at this stage

-movements start to transition from declarative knowledge to procedural knowledge 

-movements become part of procedural knowledge

-declarative knowledge about the task is less available

-movements automatic, dont require attention

-speed/accuracy improve

-feedback is less important

-consistent practice is needed to maintain automaticity 

-feedback from one movement triggers the next one

-evidence against:

-speed (movement sequences occur too quickly)

-anticipation movements in typing (keystrokes overlap in time)

-representation of the plan for movement and movement sequences

-fast, doesnt require feedback

-abstract

-hierarchical (abstract high level, specific low level)

-composed of subprograms (less abstract rep. of movement sub-parts)

-Task: learn movement sequences that vary in complexity

-Results: time to release kay increased with the complexity of the movement sequence

-conclusion: we retrieve a seq. of movements (motor program) all at once and then execute them

**response chaining not supported 

-Rosenbaum experiment 

-Evidence for subprograms:

-for longer sequences, more subprograms must be retrieved before being sent to the muscles

-sub programs are not reusable

-to perform, person needs to acquire higher order abstract program ABAA

-describes a specific type of speed-accuracy trade-off that occurs when people make aimed movements

-describes many types of AIMED movements

-RT depends on 2 things

1. size of target

2. current distance from target

-consists of initial state in which a person begins and a goal state that is to be attained

-plus a non-obvious way of getting from one to the other

-completely specified starting conditions, goal state, and methods for achieving goal

-ex: hobbits and orcs

-some aspects of the initial state, goal state, or methods are not completely specified

-ex: finding the perfect mate

-how a problem is initially represented impacts the likelihood of reaching a successful solution

-restructuring of the given info can aid in finding solution

-experts in a given problem solving domain spend MORE time than novices on development of a good problem representaion 

-our initial rep. of a problem may contain both relevant and irrelevant info

-irrelevant info may distract us from attaining a solution

-one aspect of expertise is the ability to pick out relevant info and ignore irrelevant 

-retrieve a rep. of an old problem from memory that is similar to the problem you currently face

-solve current problem using solution from the old problem 

-if subjects know of an analogy, they will be better able to come to the correct solution to a problem

-Duncker's Ray problem

-Military story problem

-can apply the military story analogy to the Duncker Ray probelm and more people can come up with a solution

-Knowledge carryover enhances performance

-ex: knowledge of military story carries over to aid with solution to the Ray problem

-also called entrenchment 

-getting stuck in the rut of using a particular type of problem rep. can hinder you from switching to using a different, more effective type of rep. 

-Ex: Luchins' Water Jar problem

(stuck using equation B-A-2C)

-unnecessarily following rules that are NOT stated in the problem

-ex: nine dot problem, water glass problem, planting 4 trees problem

-seeing an object as having only a fixed, familiar function

-ex: Duncker's Candle problem (subjects were fixed on the function of the box as holder of tacks)

-Maier's two-rope problem (functional fixedness of the wrench)

-failure to notice analogies

-mental set

-imposing unnecessary constraints on problem space

-functional fixedness

-knowledge carry over hinders performance

Ex: mental set, unnecessary constraints, and functional fixedness

-a step-by-step procedure for solving a problem (like a recipe)

-following an algorithm guarantees a successful solution

-short cut/ "rule of thumb"

-not guaranteed to lead to a solution

-difference reduction

-means-end analysis

-working backward

-At any point, select an operator that moves you closer to the goal state

-ex: do not go from ann arbor to chicago to get to paris

-However, sometimes solving a problem requires you to temporarily move away from the goal state

(Hobbits and orcs problem)

(difference reduction PLUS subgoaling)

1. identify the largest difference between current state and goal state

2. set a subgoal to reduce that difference

3. find and apply an operator to reduce the difference

4. if operator can't be applied, new subgoal = remove obstacle that prevents applying the operator

-transform the goal state so it is more similar to initial state

-useful if there are too many paths leading from initial state

-ex: water lily problem

1. Experts develop better problem representations

(spend more time setting up the problem)

2. Experts use analogies

3. Experts knowledge is better organized and elaborated

4. Automatization 

(practice converts controlled processes-> automatic processes)

-experts memory for RANDOM board configurations no better than that of novies

(don't just have greater memory capacity)

-experts memory for ACTUAL board config. much better than that of novices