Cog Psych Exam 3

Concepts, Language, Visual Knowledge, Judgement and Decision Making, Problem Solving

Perception

inferences guided by knowledge

Attention

anticipate inputs guided by knowledge

Memory

connect and organize information

fill in gaps based on knowledge

Prototypes

idealized member (average)

an average of the category members that have been encountered

May differ across individuals

May differ across countries/cultures

Possesses all the characteristic features

Usually there is no real world encounter of prototype

Exemplars

category members (examples)

Early learning involves exemplars.

Experience involves averaging across exemplars to get prototypes

Concepts

mental representations

concepts as theories

like schemas

Allow people to generalize

Permit use of knowledge to fill in gaps and make inferences

Make inference based on understanding of knowledge that specifies category membership, not similarity

Theories can explain cause and effect relationships

Categorization

utilizing our concepts

can help but it can also lead to systematic errors

stereotyping

groups as seen as more similar to each other and we exaggerate the difference between groups

typicality effects

The more prototypical category members are “privileged” in rating tasks

intercategory

compare between groups

intracategory

compare within group

cognitive economy

willing to sacrifice some accuracy in order to be efficient

Rosch’s prototype theory

rather than focusing on definitions that define the boundaries of a category, the category is characterized by a central member that possesses all the characteristic features (the prototype

categories have fuzzy boundaries, with no clear specification of membership and nonmembership

graded membership

the idea that some members (those closer to the prototype) are “better” members of the category than others

distance reflects how good a member of the category an exemplar is

Posner & Keele

View dot patterns

Judge whether each belongs to category A or category B

Guess at first, but will get better with feedback

Prototype typically easy to classify and remember

Even if the prototype is never seen during learning

sentence verification task

True or false?

Robins are birds.

Penguins are birds.

A sentence like “Robins are birds” can be verified faster than a sentence like “Penguins are birds.” This is because robins share more features with the prototypical “bird” than penguins do.

production task

Name as many fruits as possible.

People typically start with category members that are closest to the prototype

If we ask people to name as many birds as they can, they typically start with category members that are closest to the prototype (e.g., robin). For fruit they are likely to start with bananas, apples, or oranges.

limitations of prototypes

Too much information discarded

Correlation

Context dependence

economical but less flexible

categories have fuzzy boundaries, with no clear specification of membership and nonmembership

limitations of exemplars

More flexible but less economical

May keep too much info – but helps represent variability w/i category

Similarity

current view

heuristic for categorization

Similar things share properties and tend to be in same category

Can account for most typicality effects and graded category membership.

Quick but imperfect strategy

Too variable

Any two things can be arbitrarily similar

Judgments of resemblance depend on other knowledge, such as which attributes are important and which are not.

knowledge network

conceptualized web of ideas

Knowledge represented via a vast network of connections and associations between all the information we know.

Revealed by the sentence verification task.

hierarchical network

Advantages:

Supports inferences

Hierarchies allow inheritance

Cognitive economy

problems:

Typicality effect

Robin is a bird < Penguin is a bird

Robin more accessible from Bird than Penguin

Nonredundancy

Having feathers associated with bird

Quicker to say peacock has feathers than sparrows have feathers.

Nonhierarchical information can affect category judgments (e.g., typicality effects)

Adaptive control of thought (ACT) associative network

Network of propositions and labeled links

represent more complex relationships

Nodes represent concepts

Links associate related concepts

Like a feature net (on steroids)

Necessary to represent propositions (simplest truths), not just associations

However, we can think about a much larger set of relationships than just equivalence and possession

propositions

the smallest units of knowledge that can be true or false

ACT network characteristics

Nodes have activation

thresholds

Linked nodes sum activation

Links vary in strength

frequency

recency of prior encounter

Similar to feature net model

more frequently/recently encountered information more strongly linked, easier to activate

ACT Solution

Links vary in strength, meaning that some information can be accessed quickly

Spreading activation as retrieval process

Should not see typicality or redundancy

One thing makes you think of another, or at least prepares you to think about it.

propositional networks

local representation each node is equivalent to a concept

connection networks

Distributed representation—pattern of activation across units represents a concept

use distributed representations, where information is represented by a pattern of activation across the network.

Because this processing is also parallel (and not serial), connectionist networks are said to involve parallel distributed processing (PDP).

Connectionist network

Model cognitive processes using interconnected networks of simplified, neuron-like, units

The units (nodes) send and receive signals analogous to neural communication in the brain

The strength of an output depends on the strength of the input signal from the sending unit combined with the strength of the connection between the units (called a connection weight)

marr’s input

take input transform it based on learning to a response

The input layer codes a pattern of activation and sends it to a hidden layer, which, in turn produces a pattern of signals at the output layer.

Units producing an output (depicted as filled circles), and units that are not active enough to meet the threshold for sending a signal (an off neuron with an output of 0, depicted as open circles).

By adjusting the connections weights(links) between the input layer and the output layer, the network learns to associate specific input patterns with specific output patterns.

The system acts as an associative memory that takes a memory cue and outputs (retrieves) an associated pattern.

distributed representation scheme

Use of a pattern of activity across a set of units to represent information

connectionnism

Set of approaches in cognitive psychology, AI, and neuroscience, which models mental phenomena as emergent processes from interconnected networks of simple units

Related concepts:

Neural networks

Machine learning

applications fall into 3 categories

Pattern classification

Prediction

Control and optimization

pattern classification

Solving these problems entails "learning" patterns in a dataset and constructing a model that can recognize these patterns

prediction

Build a forecasting model from the historical data set to predict future data points

optimization

Finding solutions for a set of very difficult problems

concepts overview

We can reason about concepts using prototypes and exemplars, particularly in cases where quick judgments are required.

But for more sophisticated judgments, we can bring theories to bear, represented by networks of interrelated conceptual knowledge.

Computational networks have been used to explain this capability.

broca aphasia

lesion frontal

speech nonfluent

comprehension

wenicke aphasia

lesion temporal

speech fluent

comprehension poor

organizaiton of language

sentence

phrase

word

morpheme

phoneme

This organization is hierarchical because sentences are composed of words, words are composed of morphemes, and morphemes are composed of phonemes. The hierarchical organization allows parts to be recombined in novel ways.

phoneme

smallest unit of sound within a language

building blocks of speech sounds

Born to hear all phonemes, but lose this ability after about 6 months old

Lack of invariance (co-articulation, categorical perception, top-down assistance, McGurk effect)

morpheme

smallest unit of meaning

phonology

sequence of phonemes that make up a word

lexicon

words within a language

syntax

structure and grammar

how to combine a word with other words to form sentences

A set of rules that describes the legal sentences that can be constructed in a language

Generativity

Infinite number of sentences by combining finite set of words

Phrase-structure rules specify deep structure

Transformational grammar specifies possible surface structures

semantics and comprehension

meaning

what the words (the sentence) mean and how relates to other things we know

Immediacy and eye-mind assumption

Prosody, pragmatics

discourse

larger units of analysis (e.g., conversations, stories)

properties of language

Ambiguity rampant

Arbitrary – The relationship between the elements of the language and their meaning is arbitrary

Duck doesn’t look or sound like what it is—arbitrary symbol

Generative – The basic units of language (words) can be used to build an unlimited number of meanings

Combine symbols to make novel ideas—atypical, das Stachelschwein

Metacommunication

generativity

New words can be formed

“Hardware” and “software” lead to “spyware” and “malware”

Any political scandal is “…..Gate”

Words can take on new meanings.

“I have been hacked by a hacker.”

Infinite number of sentences by combining finite set of words

voicing

Whether vocal folds vibrate ([z], [d], [b], [v])

Or not ([s], [t], [p], [f])

manner of production

Whether air is fully stopped ([b], [p], [d], [t])

Or merely restricted ([z], [s], [v], [f])

place of articulation

Where in the mouth the air is restricted

Closing of lips ([b], [p])

Top teeth against bottom lip ([v], [f])

Tongue behind upper teeth ([d], [t], [z], [s])

fragile aspects of language aquisition

phohonology (production) and grammar are age sensitive

resilient

semantics/ vocabulary learning can be easily learned later in life

speech segmentation

the process of “slicing” the speech stream into words and phonemes.

We know how to segment our own language only. This is why foreign speakers sound like they are talking fast

coarticulation

the blending of phonemes at word boundaries

-Helps speech be more fluent and quick.

-think about bad AI speech production it cannot consider one letter at a time it has to consider the surrounding letters to make it sound natural

problem of invariance

one phoneme can actually take many different forms (use top down to fix)

to decipher the variable, ambiguous signal (the data) we benefit from:

Top-down support

Categorical perception

categorical perception

speech sounds vary continuously but we percieve them in categories

parse

Process of assigning words to proper phrase structure

So you can figure out the subject and object of a sentence

When we perceive a sentence, we must parse the sentence’s syntactic structure, or

assign the words to a phrase structure

garden path sentences

We parse sentences as we hear them which can lead to errors

A garden-path sentence initially suggests an interpretation that turns out to be wrong

minimal attachment

simplest phrase structure

deep structure

The meaning of a sentence

If transformations are applied, the sentence can be turned into a question, phrased in the passive voice, etc.

surface structure

The exact wording and syntax of a sentence

movement rules (transformations)

shift elements to produce different surface structures

immediacy

try to interpret each content word as encountered

eye mind assumption

word fixated as long as that word being processed.

If you are not looking at a word you are not processing it.

extralinguistic context

factors outside of language itself affecting comprehension

maxim of relation

things should be relevent

maxim of quantity

not more informative than necessary

prosody

Refers to the patterns of pauses and pitch changes that characterize speech production. It is used to:

Emphasize elements of a sentence

Highlight the sentence’s intended structure

Signal the difference between a question and an assertion

disfluency and memory

Contained “uhs” and “ums” or was completely fluent

Better retellings if stories contained disfluencies

Is this because disfluencies serve as guideposts to major plot points?

Placed disfluencies in typical or atypical locations

Memory better for the both; disfluencies increase general listener attention

whorfian hypothesis

Wording impacts thought which impacts the memory

absolute

Does not change no matter how the reference changes

t in absolute frame of reference east remains east even when the person turns around

egocentric

Describe things in relation to the person you are referring to

it is to the left of him

Person turns around their right becomes their left

Moderated by attention => experience. Indirect influence

Language => Thought

Language => Attention => Experience => Thought

Mentalese

deductive reasoning

Problems to which one can apply formal logic and derive an objectively correct solution

Premise: statement of fact taken to be true for the purposes of a logical problem

Conclusion: A statement of fact derived by logical processes

syllogisms

conditional statements

syllogisms

logical composed of three statements of fact: 2 premises and conclusion

conditional statements

a logical form composed of 3 statements

1) “If condition p is met, then q follows.”

2) p or q is true.

3) A conclusion about p or q

conversion errors

a person reverses one of the premises.

“All As are Bs” and believes that it is also true that “All Bs are As.”

belief bias

belief syllogism’s conclusion is something people already believe is true, they are more likely to judge the conclusion as following from the premises

atmosphere errors

People also incorrectly rely on a low-level matching strategy between the words in the premises and those in the conclusions

Two premises of a syllogism are both either positive or negative or use the same quantifier.

People biased to accept a conclusion that maintains the same atmosphere

truth

depends on consistency with the world

validity

depends on the form of the statements

does it follow logical progession

do people think logically

(some) Philosophers say “yes” and errors are due to carelessness

Cognitive psychologists say “no”; errors tell us how people really think in everyday situations

Atmosphere effects

Belief bias

Atmosphere – using consistent wording to gauge validity

Belief – if you believe the conclusion is the truth then you are less likely to check the validity

modus ponens

valid

mood of affirmation (confirming anticident)

If P is true, then Q is true.

P is true.

Therefore, Q must be true.

If I drink beer then I will get fat

I drink beer

therefore, I will get fat

Denying the antecedent

error

If A is true then B is true

A is not true

Therefore B is not true

If its a robin it is a bird

It is not a robin

Therefore it is not a bird

Modus Tollens

valid

If P is true, then Q is true.

• Q is false.

• Therefore, P must be false.

If I drink beer then I will get fat

I did not get fat

therefore, I did not drink beer

Affirming the consequent

error

If a is true then b is tru

B is true

Therefore a is true

If its a robin then its a bird

It is a bird

Therefore it is a robin

four card rule

2 flips

check to confirm

check to falsify

permission schemas

one type of pragmatic reasoning schemas

If a person satisfies condition A, they have permission to carry out action B

Activating a permission schema can improve performance in a four-card task

Pragmatic reasoning schemas

Used to solving real-world versions of a task (ways of thinking about cause and effect that we learn in daily life)

reasoning

going beyond the information given

transitivity

If you prefer A to B, and B to C...

Then you should prefer A to C

Decision making example?

what is a good decision

Rational or normative decision making

A set of rules by which some choices are better than others and one choice can be said to be optimal

expected value theory

the best choice is the one that offers the largest financial payoff

problem: Not every dollar has the same subjective value

expected utility theory

The best choice is the one that offers the reward with the greatest subjective or personal value (not necessarily the greatest financial reward or expected value)

framing effects negate expected utility theory

description invariance should hold

Consistently make the same choice irrespective of how the problem is described, as long as the basic structure is the same