PSYC224-Exam 3

The fact that people recognize words more quickly when they are in a sentence that makes sense is consistent with what model?

desc how model works

What does the evidence suggest about the use of top-down information during language comprehension in experiments looking at how people resolve ambiguity at the level of sound, word meanings, and syntax?

In an experiment, subjects heard the following sentences in which a particular sound was replaced by a load crash: “It was found that the oat was on the lake. It was found that the oat was on the hanger. It was found that the *oat was on the farm”. What did the experiment find?

In an experiment on word meaning ambiguity, participants first read words that can have multiple meanings (e.g., train) in sentences that are consistent with only one meaning. After reading the sentence, they are then asked to respond to a target word that is related to one of the word’s meanings. Which pattern of results would be predicted by an interactive model of language comprehension?

In an experiment on word meaning ambiguity, participants first read words that can have multiple meanings (e.g., train) in sentences that are consistent with only one meaning. After reading the sentence, they are then asked to respond to a target word that is related to one of the word’s meanings. Which pattern of results would support a serial model?

A key debate in language comprehension centers around whether or not people automatically make elaborative inferences as they read. One view argues that they do not make inferences if they do not have to. What pattern of results would support this view?

What is a key difference between the discrete two-step model and the interactive activation model?

What would the discrete, two-step model predict in a case where A and B participate in an experiment where they are asked to silently read word pairs and are occasionally cued to read the pairs aloud as fast as possible? A is cued to read aloud the following pair as quickly as possible, “darn, bore” where as B is cued to read aloud “dart, board” and the priming words are “ball, dome; beak, doll; bus, door; bell, dark”.

What would the interactive activation model predict in a case where A and B participate in an experiment where they are asked to silently read word pairs and are occasionally cued to read the pairs aloud as fast as possible? A is cued to read aloud the following pair as quickly as possible, “darn, bore” where as B is cued to read aloud “dart, board” and the priming words are “ball, dome; beak, doll; bus, door; bell, dark”.

How do findings from research involving cross-linguistic differences in color and quantity labels support or detract from the strong version of the Sapir-Whorf hypothesis?

How do findings from research involving cross-linguistic differences in color and quantity labels support or detract from the weak version of the Sapir-Whorf hypothesis?

What are specific ways in which language potentially influences our thoughts?

Sapir-Whorf

Language

communicates as well as expresses emotion, thought, and identity; arbitrary, structured, & generative

Arbitrary

made up of learned systems

Structured

governed by a system of rules

Generative

expresses limitless meaning

Phonology

language sounds and how to combine them

Phoneme

sounds that make a difference between meaningful words; certain combinations “allowed” in each language

Dialects

phonological representations vary across languages and can lead to difference in pronunciation in the same language

Morphology

parts of words that have “meaning” (e.g., plurals, tenses)

Syntax

structure, grammar, language rules; ways that words can and cannot be put together

Semantics

understand what words and sentences refer to; relatively easy for some things, and much more difficult to find the meaning of others

Pragmatics

understanding what people mean by what they say; implications (e.g., “can you pass the salt?” vs. “hand it over…”

Understanding language requires computing at multiple levels (e.g., sound, structure, meaning) leading to what two solutions to solve this ambiguity?

serial bottom-up solution and interactive solution

Serial Bottom-Up Solution

we process language using default rules (heuristics) that get things right most of the time; a separate stage of processing corrects errors using context; takes in sound by phoneme → turns to words → then applies grammar

Interactive Solution

language is always processed using multiple sources of available information including context and non-linguist information (e..g, visual cues, person-specific cues, etc.); sounds words and grammar affect understanding bidirectionally

Phonemic Restoration Effect

top-down processing of sounds; people hear a sentence where sound has been replaced by a noise and found that 19 out of 20 subjects said no sound was missing, and that no subject could accurately locate the sound/cough

Cohort

set of words consistent with the first syllable

Cohort Model

top-down processing of sounds as words consistent with the input become active; words in the cohort are eliminated when they become inconsistent with input; other words are eliminated due to contextual incongruity; processing ends where there is one word left in the cohort (syntactic condition - responses slower with sentences with no meaning vs. random condition - responses slowed more with no meaning or syntax)

Syntactics (Attachment) Ambiguity

sentences of phrases that have more than one interpretation, or “parse” (e.g., the women in the chair with the broken leg)

Solutions to Syntactics Ambiguity

Serial: use heuristic to resolve ambiguity, if it wrong, correct interpretation later with context vs. Interactive: using top-down and bottom-up to interpret meaning dependent on context

Word Ambiguity

heuristics and context help to identify words faster, but as separate processes; use knowledge of words and their meanings to disambiguate; also use frequency of how often we encounter a certain meaning to pick the correct meaning of a word

Elaborative Inferences

fill in missing details using context (e.g., paragraph using the word “assaulted” or “stabbed”, if word was “stabbed” followed by “knife”, assume the knife was used)

Speech Errors

“arise from concurrent action or two different intentions”; reveal the processes going on as language output is planned; includes sound exchanges, anticipation errors, perseveration errors, and blends

Morpheme Errors

e.g., “self-destruction instruct” instead of “self-destruct instructions”

Sound Exchanges

occur mostly between nearby words, not across faces, does not respect grammatical category or function (e.g., “rack pat”)

Word Exchanges

span some distance; respect grammatical category and function (e.g., write a mother to my letter)

Anticipation Errors

a type of speech error that arises from a slip of the tongue in which the speaker prematurely uses a sound or syllable from later in the sentence to replace an earlier unit (e.g., take my bike → bake my bike)

Perseveration Errors

a type of speech error where a sound or syllable from an earlier part of unintentionally repeated in a later part replacing the correct intended sound (e.g., beef noodle → beef needle)

Blends

a type of speech error in which two intended words or phrases compete for the same slot and fuse together resulting in a single utterance (e.g., taxi cab -—> tab)

Discrete, Two-Step Model

message → semantic & syntactic retrieval → phonological encoding → articulation; retrieve semantic and syntactic information about words to determine their phonological form

Interactive Activation Model

suggests that semantic and phonological retrieval must interactions; PDP model with layers of nodes corresponding to semantic features, words, and phonemes, all connections are bidirectional and activation spreads through the network; word nodes and phonemes with the highest activation are selected

Lexical Bias

tendency for errors to create words rather than nonwords (e.g., saying “hold card cash” instead of “cold hard cash”); output determined by sleected the phonemes with the highest activation levle

Mixed Errors

tendency to produce errors that are phonologically and semantically related to the intended word (e.g., “apricot” instead of “apple”); receive top-down and bottom-up activation, and are thus more likely to be erroneously (mistakenly) selected

Sapir-Whorf Hypothesis - Strong Version

language determines thought; thoughts are only accessible if you have the words for them (e.g., the Kiriwana word “mokita” means “the truth that everyone knows but no one talks about”, or the idea in English of “the elephant in the room”; evidence being that language does not determine thought

Sapir-Whorf Hypothesis - Weak Version

language favors some thought processes over others; it will tend to think in ways your language suggests. Implications for sexist language and implicit bias; evidence being that language influences thought

Decision-Making

choosing between alternative courses of action

Choice Among Alternatives

you know all the possible outcomes and must choose between them

“Risky” Decision-Making

Compensatory Models

positive features can make up for negative features (e.g., additive model)

Noncompensatory Models

positive features do not make up for negative features; useful when there is no time to make “perfect” choice, only satisfactory (e.g., elimination by aspect, satisficing)

Additive Model

determine the “pros” and “cons”; add up pros & subtract cons and go with what is numerically highest; may not give the answer we “want” as some features may matter more; modifications to model then allow to weight features by priority

Elimination by Aspect

pick one dimension (e.g., price) and a criterion, and eliminate all options that do not meet that criterion. If left with more than one option, repeat the process with a new dimension and criterion; the order of dimensions chosen is important for the final outcome

Satisficing

decide on a set of minimum criteria, look through options and find the first one that meets the criteria; less effort & good for limited time

Normative

pick the choice with the biggest payoff

Rational

make consistent decisions

Expected Value Theory

(value of outcome) * (probability of outcome); motivation to perform a task is determined by an individual’s expectation of success and the subjective value they place on that success

Expected Value

measures the values of outcomes against likelihood, but people often make decisions that go against it

Expected Utility

measure perceived value against likelihood, which can be more valuable for different people or situations; subjective baluevalue

Allais Paradox

proposes two scenarios where people agree on the best decision in each case, then show that the reasoning between these two decisions is inconsistent

Certainty Effect

people often choose certain rewards and avoid certain failures, despite going against expected value

Loss Aversion

people’s tendency to strongly prefer avoiding losses to acquiring gains (some studies suggest that losses are twice as powerful psychologically as gains)

Framing Effects

choices are influences by the way that they are presented (e.g., in disease problem, solutions that are framed as who will be saved are most sought after as they are a certain gain vs. solutions framed as who will die, as people want to avoid a certain loss)

Psychic Budgets

Sunk Cost Fallacy

when past actions affect future choices in an irrational manner (e.g., will walk out of a play is $10/ticket, but not if $50/ticket, even though will not get money back either way)

Heuristics

cognitive “shortcuts”; applied to make a decision; useful because they do not take a lot of time or resources, but do lead to judgement bias

3 Common Heurisitics

anchoring & adjustment, representativeness, availiability

Anchoring & Adjustment

how do we determine what a “reasonable” estimate is? (e.g., price of an time, probability of an event, year of a historical event)

Anchoring

a strategy in which estimation becomes with an intital anchor (or value) and the estimate is then adjusted in lgiht of incoming information

Representativeness

estimate that something is more probable if it has many features that you associate with a particular one

Typicality

e.g., Julie is 26-years-old, has a degree in PE, has been physically fit since childhood, and loves the outdoors, so the probability she is a ski instructor who teaches aerobic is high

Conjunction Fallacy

two conditions can never be more probable than the probability of one of them alone (e.g., probability of a heart attack is overall higher than heart attack in those over 55 who smoke)

Availability

items that are easier to remember are also thought to be more common (e.g., death by asthma or tornado, 58% of people picked tornado despite it being less common)

Solo Member Effect

the more distinctive a feature is (e.g., 1 Asian in a group of 6 people vs. 3 Asians in a group of 6 people), the more available the feature is and therefore estimated to be more probable

Sample Size

how many observations are made (e.g., a particular vet treats 20 cats/day); the smaller the value, the more likely to deviate from the average

Base Rate

how often an event occurs in the population (e.g., 70% of cats are tabbies); if value is low, then more likely to be false positives than corrective diagnosis in the medical field

False Positives

an error in date modeling, testing, or screening where a result incorrectly indicates the presence of a condition that is not actually there

Reasoning

given some information, decide what else is true

Inductive Reasoning

takes an example of a small set of examples and infers what might be true of the group

Deductive Reasoning

takes the general principles and determines what must follow them

Logic

a formal system that gives us rules for determining the truth of conclusions based on a set of premises

Premise

something we assume/know to be true (e.g., a lecture posted to the PSYC224 Canvas page will be about psychology)

Conclusion

something we derive from the premise (e.g., this lecture is posted to the PYSC224 Canvas, so it will be about psychology)

Antecedent

the “if” part of the premise (e.g., if a lecture is posted to the PSYC224 Canvas…)

Consequent

the “then part” of a premise (e.g., …then it will be about psychology)

Logic tells us that…

affirming the antecedent should be true, and denying the consequent should be true

Wason Selection Task

four cards (E, 4, 7, F), to confirm the rule that if a card has a vowel on one side, it has an even number on the other side, need to turn E (affirm antecedent) and 7 (deny consequent)

Confirmation Bias

tendency to look for evidence that confirms our hypothesis rather than looking for falsifying evidence

Cheating Schema

when people think someone is cheating the system (e.g., drinking alcohol when not supposed to), they reason logically due to the social tendency to not want people to get away with things vs. if not cheating, do not reason logically

Problem-Solving

representing and aligning the world and our goals to allow us to, for example, develop a plan to study for college, find a job, or manage relationships

a problem has…

givens, states, goal, obstacles, and operators

Givens

the initial state (current situation, available objects, and information)

Goal

the desired “final state’ (outcome or information)

Obstacles

what is keeping the current state different from the desired final state

Operators

what allows us to change the current state; need this otherwise stuck

State(s)

current descriptions of the elements of the problems

Problem Space

a “map” that shows all possible moves and how they are linked; with this information, can determine the most efficient mapping from initial state to goal state with as few possible moves as possible; used with well-defined problems

Well-Defined Problems

parts of the problem are clearly specified (e.g., games, math, Tower of Hanoi, Cannibal vs. Missionaries, Chess); know the initial state, what the final state should be, and what all the allowed moves are

Ill/Poorly Defined Problems

uncertainty in givens, obstacles, operators, and even goals (e.g., healthcare decisions, innovation in business, relationship issues, tumor/military problem, insight problem)

Intermediate Products