Cognitive🔢 Flashcards
Processes of memory
Encode (info enters memory or slips away)
Storage (preserved for recollection in future or forgotten)
Retrieval (info recollected on tip of tongue)
Memory taxonomies
Multiple models that categorise memories
Modal model of memory
TIME
Input➡️sensory memory ➡️STM ➡️⬅️LTM
Sensory memory
Research
The Senses, modal special 250-500 ms iconic
Spelling partial report method- shown part of whole grid, report one row or whole grid. Recall perfectly with tone, late capacity
STM
Increased by rehearsal and chunking
30-60s duration
Miller- 7 +/- 2 capacity
LTM
Transferred STM to LTM by rehearsal loop
Retrieved LTM to STM
Unlimited duration and capacity
Working memory
Combines Modal Model and LTM systems
Focus attention of different info and processing to anticipate next steps
Manipulate info in STM: active rehearsal and chunking
Maintain info while distracted
Reading span task- true or false sentence and remember last word in sentence: duration 3s (shorter than STM) and 3-5 capacity
Long term memory systems CONTENT
LTM (Decarative and Non declarative)
Declarative explicit - episodic and semantic
Non declarative implicit- procedural and other e.g. conditioning
Declarative and non declarative
Declarative- aware of, can express
Non-declarative- difficult to bring to awareness and express
Multicomponent model
CE (coordinates storage systems)
VSS PL (separate storage for visual/auditory info) EB (bind multimodal info to episodic memories)
LTM (transfer between WM and LTM)
Memory in brain areas (Lashley)
Search for engram (neural representation of a memory)
Side of lesion predicted memory performance
Distributed memory areas
Areas of the brain and specific memory functions
Frontal cortex- working memory
Cortex-distributed memory storage
Cerebellum- form implicit memory
Hippocampus- form explicit memory
Amygdala- form implicit and emotional memory
Temporal- spatial memory and episodic storage
Mental time travel
Re construct personal events from the past, imagine possible future scenarios, early memories
Chronesthesia
Reminiscence bump
Superior memory for events from adolescence
Childhood amnesia less than five years
Measuring memory:
Types of recall and recognition
Recall:
FREE (as many as can)
SERIAL (in order)
CUED (help of cues)
Recognition (verify of probe matched memory)
Measuring memory:
Indirect memory task
Incidental encoding (not conscious) and implicit retrieval (seemingly unrelated)
Improving encoding:
3 Levels something can be encoded
Structural (what looks like)
Phonetic (if word rhymes)
Category (type of word)
LTM is coded deeper, connects with knowledge already stored
Improving encoding:
Spacing effect and varying cues
Better memory after distributed practice
Longer spacing-better memory
Varied retrieval cues changes circumstance (mood, environment) stored with material
Improving encoding:
Serial position
Info encoded first or last remembered better
Primacy- no interference, deeper processing
Recency- STM retrieval
Improving retrieval: Testing effect (re study vs recall)
Recall after 5 mins better with re study
Recall after 2 days/week better with recall
Improving memory:
Encoding specificity
Better retrieval when context at encoding and retrieval is same
Improving memory:
Transfer appropriate processing
Match processing of encoding and retrieval aids episodic memory
Semantic memory network:
Collins and Quillian’s hierarchical network
Not economical to separately store representations
Representations: activation between nodes and their connecting paths
Superordinate, basic, subordinate
Semantic memory network:
Collins and Loftus network
Semantic relatedness between words
Theories of categorisation:
Prototype theory
and criticisms
Metal representation average of all members. Common and distinctive features
Cannot explain how people can tell the size of categories e.g. dog breeds
Can not explain how members are added to category
Theories of categorisation:
Classical
And Criticisms
Semantic memories form representations
Make predictions, extract features
Family resemblance- different members of category can share different features
Central tendency-categories have an averaged ideal but some deviate
Graded membership- some members more typical
Theories of categorisation:
Exemplar theory
And Criticisms
Separate representations of physical features of examples (seen before) Can tell category size can add new members
Theoretical circularity-how can people retrieve all members to define category it’s based on category membership initially
Can not explain how categories without features are formed
Theories of categorisation
Explanation based theory
Common characteristics rather than physical features
Can be created ad hoc (waterfowl can swim so have webbed feet)
Measuring categorisation tests
Typicality rating- how good of an example from category it is
Exemplar production-recall as many items from category as you can
Category membership verification-asked if fits the category or not
Schemata
Capture commonly encountered aspects of life, predictions
Schemas= order of events temporally ordered
5 primary schema processes
Selection Abstraction Interpretation Integration Reconstruction
Schema processes: selection
Select info central to schema
Study text without context, with context before or with context after
Those told context before recalled the most
Encoding of schema relevant info
Schema processes: abstraction
Objects consistent to schema are retrieved
7 or 4 looks like either
Ambiguities are converted to abstract representation that captures the meaning and consistent with schema
Schema processes: interpretation
Filled gaps with schema consistent info
“Dropped the glass”
Most assumed dropped glass dropped glass was broken
Schema processes: integration
Permissible inferences
Chair on box to right of tree, infer chair to right of tree
Infer chair to right of tree
Integrate schema, schema consistent holistic representations
Schema processes: reconstruction
Meet own schema
Reconstruct Native American tale to meet own cultural schema
Details reconstructed, simplified
Participants recalled items in waiting room of postgrad that wasn’t there
Schacter’s seven sins of memory
Transience Absentmindedness Misattribution Blocking Bias Amnesia Persistence Suggestibility
Schacter’s seven sins of memory: TRANSCIENCE
Decay and research
DECAY- forgetting over passage of time
Thorndike’s law of disuse: memory decays with longer time it isn’t used
Rows of nonsense trigrams, most forgetting is early and continues slowly
BUT passage of time causes nothing itself, correlated with forgetting
Schacter’s seven sins of memory: TRANSCIENCE
Interference and research
INTERFERENCE- forgetting from memory competition
Proactive- older memories impair retrieval of new
Retroactive- newer memories impair retrieval of old
Brown-Peterson paradigm- learn trigrams then recall after distracting task (varying hours) better memory with less retroactive interference, remembered if go straight to sleep
Schacter’s seven sins of memory: ABSENTMINDEDNESS
Lapses of attention that affect memory
In lecture almost half has task unrelated thoughts, caused poorer results in test
Schacter’s seven sins of memory: BLOCKING
Info present but temporarily inaccessible
Resolving tip of the tongue may prevent it recurring later
Schacter’s seven sins of memory: MISATTRIBUTION Source monitoring (where memories come from)
Attribute memories to incorrect source
Internal-real/imagined
External- who told you
Reality- real or heard about
Schacter’s seven sins of memory: MISATTRIBUTION
Source info types
Perceptual- detail highly if memory actually experienced
Contextual- context memory acquired is consistent with expected source
Affective-emotional reaction
Cognitive- mental processing of info
Schacter’s seven sins of memory: MISATTRIBUTION
Deese-Roediger-McDermott paradigm
People falsely recall semantically similar words that were never presented
Schacter’s seven sins of memory: SUGGESTIBILITY
Implement memory which never occurred
25% falsely remember lost in a mall
50% falsely remembered doctored photograph
More false beliefs when encouraged to guess, 20% forced confabulation, especially with confirmatory feedback
Schacter’s seven sins of memory: BIAS
Distort past memories based on current knowledge and beliefs
Hindsight bias- misremember memories as being more similar to current knowledge
Recall German election result as more similar to real results that what predicted at first
Schacter’s seven sins of memory: PERSISTENCE
Unwanted recollections that cannot be forgotten
Schacter’s seven sins of memory: AMNESIA
Deficit in memory caused by brain damage, disease or trauma
Can encode new memories but not explicitly retrieve them (HM)
Retrograde-cannot access memories prior to event
Anterograde- loss of ability to store new memories after the event
Eyewitness memory (recognition/recall)
Memory for crime or other dramatic or witnessed event
Recognition- identify from mugshots
Recall- statement or testimony
Loftus and Palmer
‘Smashed’ 2x more likely to say yes to broken glass than hit
Post event information
3 explanations
Memory replacement theory- overwrites original
Blocking theory-original info blocked but still present, retroactive interference
Source monitoring- info from other sources mistaken for real memory
Loftus- Questioning with consistent misleading info (stop/yield sign) asked with sign seen vs other sign
Misleading info alters memory
Effects of implicit social demands on memory
Watch video with tonic vs alcohol placebo, read info with or without misinformation
Those thinking they drank alcohol were more susceptible to misinformation
Effects of stress on memory (navy personnel)
Navy personnel, group speech with or without doctored video
Interrogate with or without photo
Questionnaire with or without misinformation
All led to false memories, 84% misidentified unrelated photo as interrogator
Vulnerable eyewitnesses
Children etc more errors
Reinforcement/feedback Repeat questions Co-witness information/pressure Invite speculation Introduce new info
Children eyewitness research
Read picture book
Closed and unanswerable questions= less correct responses
Answer given even if don’t understand
False confessions research
Plant memory over 3 interviews
Context accurate cues, incontrovertible evidence, social pressure, suggestive retrieval prompts, rapport and facilitators, pages and silence, disappointment and sympathy
70% had false memories of committing crime with police contact but reduced to 26-30% when explicit statement of memory
Arousal effect
Yerkes-Dodson U curve- medium stress for optimum memory
Easter brook hypothesis- attention and arousal interact to determine cue utilisation more for central details
Weapon focus- increased memory for weapon, others decline
Unusual item hypothesis- unusual objects attract attention inconsistent with schema
Eyewitness confidence and accuracy correlation
Accuracy-confidence correlation r=.29
Affected by:
Confirmatory feedback
Repeated questioning- less accurate, more retrieval fluency
External motivation-neutral motivation correlated higher with confidence r=.44 than money
Improving eyewitness memory
Aware of effects on eyewitness memory/ confidence
Minimise contamination of memory
Double blind procedures minimise unconscious biases
Only one suspect with at least 5 known innocent fillers
Minimise motivation, tell witness suspect may not be in lineup
Assess confidence at identification
When can eyewitness memory be reliable
Memory probed for first time, not contaminated yet
Interview protocols avoid suggestive questions and minimise guessing
Investigator sensitive to witness’ level of confidence
Cognitive interview
Reinstate context (encoding specificity principle) Recall from different points Report everything Take different perspectives Don’t interrupt witness
Increase in correct details highest for older adults (more influenced by misleading info and prone to source monitoring details)
Categorisation theories
Emeplar
Classical
Explanation based
Prototype
Language
Exchange of info, acquired from the womb
Speak fluently by 3 years
Spoken, written or signed words to communicate meaning, requires understanding of syntax
Hockett’s design features of language
Speaking (vocal-auditory)channel-humans and animals
Combine units to create meaning-humans
Speech organ, adapted-humans
Comprehension (interchangeability)-humans
INPUT (speech) ACTIVATE (mental representations of sounds) and LINK TO (meaning) OUTPUT (comprehension)
Sapir wharf theory
Way we think is affected by our language
Colour perception and word order affected by language
Recognise words immediately and link to mental representation and meaning
Main language functions
Speaking and Writing-production
Reading and understanding-comprehension
Speech production processes
Structure to communicate (subject-verb-object)
Activate phonological representation
Say out loud (Semantics-syntax-morphology-form)
How is Lexicon ordered
Ordered by associative semantic network (activate things of similar meanings) linked to form and syntax to express concept
What is lexical assess based upon (what are words activated on)
Mechanisms activate words on frequency and semantic and phonological relatedness
Sentences and phrases
Sentence =Noun phrase + verb phrase
Noun phrase= determiner + noun
Verb phrase= verb
Morphemes
Morpheme-smallest meaningful unit of sound
Free morphemes-e.g. cat
Suffix-catS
Preffix-UNderstood
Phonemes
Roughly 40
Graphemes represent phonemes
Pragmatics
Context
Grice’s Maxims= quality, quantity, relation, manner
Spreading activation
Similar meanings and similar sounds compete for selection
Spreading activation evidence
More hesitations humanities lectures, synonyms compete for attention
Slip of tongue from same category
Phrases planned, words slotted in with syntactic/morphological elements left in place
Speech production errors
semantic blend, semantic and morpheme exchange and word and phoneme exchange
Semantic blend error “spork”
Semantic and morpheme exchange “slicely thinned”
Word and phoneme exchange “Lork yibrary”
Explain tip of the tongue
Research
Activation and competition between related items. Blocked retrieval of target word
Fragment competition task-asked to fill in blanks, related words compete for selection
Italian speakers-know noun gender but can’t access (syntax and form processed differently)
Coarticulation
Sounds overlap, variations in how phonemes are produced
Sound changes when followed with different articulation (‘n’ in thin pronounced differently when proceeds a ‘b’ or ‘c’)
Comprehension challenges
Problems understanding- fail to link phonemes with meaning
Reverse process for speech
Comprehension input and output
Input (speech)
Activate (mental representations of sound) and Link to meaning
Output (comprehension)
Comprehension: speech, form, syntax and morphology, semantics
Ambiguity at different levels
SPEECH (acoustic input)-hard to tell where one word ends and begins
WORD LEVEL-similar sounds or spellings with different meanings PHONEME LEVEL- words change the way they sound depending on context
Word level ambiguity (homonyms/homographs/holographs)
Homonyms- words and sound spelt the same (Bank and bank)
Homophones-words sound the same (mussel and muscle)
Holographs-words spelt the same but different sound (read and read)
What causes phoneme level ambiguity
Different constants-air hits point in mouth
Place of articulation-causes obstruction, constant sound
Coarticulation- sounds change in sentence when followed by sound with different articulation
Sound perception Categorical perception (distinguishing sounds)
Perceive sounds at one end of continuum or another (instead of processing all variation between two sounds) infants can distinguish between phonemes
Sound perception
Defining sounds
Point at which vocal chords vibrate relative to release of closure (voice onset time)
Syntax and phonology order of processing
Syntax of a phrase is processed before phonology
P B vibrations
P vibration begins instantly
B pause after, vibration then begins
Regional accents (invariance problem)
Unable to define categories of sounds, difficult to identify common features that correspond to phonemes (in accents)
Understanding speech with cough
Phonological representation of word allows us to still understand
Top down processing
Match input to existing mental representations to understand oncoming language
More frequently used words are recognised faster, multiple morphemes take longer to recognise. Similar spelling slower to recognise
Frequency in lexical access
Easier/quicker to access words used frequently with fewer neighbours
Testing lexical access
Press for ‘word’ or non word and assess speed
Quicker reaction time for words with fewer neighbours
Impact of meaningful context (word monitoring task research )
Say word quickly as possible when heard in sentence
Measure delay period until word is said
If made sense ( meaningful context): say word quicker, comprehend meaning before next word is heard-
phonological representation already activated
Priming paradigm
Cross modal and findings
CROSS MODAL: primed with visual and auditory target at same time, asked if visual word is real or not
CAPT (captain or capture)
Related to either ship or slave or unrelated
Quicker reaction to both related words than unrelated
If presented half way through word=primed for both outcomes. If primed at end of word, unrelated concepts are deactivated
Conditioning importance
Fundamental to learning, explain behaviours and habits
Relevant for drug addiction e.g. cues and relapse
Precise stimuli, well controlled experiments
Can generalised to more complex learning
Apply to AI in teaching
Types of learning is similar across species
Functional analogies: same learning function implemented differently in structural brains
Classical conditioning
US- UR
US+CS-UR
CS-UR
Aversive conditioning (phobias)
Unconditioned stimulus can also be aversive, understanding phobias, anxiety
One shot (single trial) learnt quickly and long lasting Evolutionary principle to think stimulus caused response to avoid it for safety
Extinction
Remove unconditioned stimulus to see how strong the associations are.
Present conditioned stimulus alone
Reduction in response to CS is extinction
new associations learnt much faster however
Spontaneous recovery
Resting period between last day of extinction and test trail
Response to CS returns
Original learning not completely erased
Generalisation
Introduce another stimulus (shares resemblance)
Respond with CR to new stimulus
We are not overtrained in specific examples
Discrimination
Similar stimulus but do not respond with CR to new stimulus
Pay attention to differences
Second order conditioning
Pair similar conditioned stimulus with conditioned (CS1 + CS2) - CR
CS1 acts as US
CS2 also triggers conditioned response
Learning in longer chains of associated stimuli
Trial, block and session definition
Trial-single presentation of a CS-US
Block-several traits, specific parameters
Session- one or more blocks, sessions usually separated by intervals (hours/days)
Partial reinforcement
Random trials where CS is not followed by US. Slowed acquisition and extinction (difficult to link)
Partial reinforcement of trials (fixed ratio, variable ratio, fixed interval, variable interval)
Fixed ratio-behaviour reinforced after specific number of responses
Variable ratio-behaviour reinforced unpredictable number of responses (high responses despite low enforcers)
Fixed interval-behaviour reinforced after specific amount of time
Variable interval-behaviour reinforced after unpredictable amount of time
Temporal sequences
trace and delay conditioning
Which easier to learn?
Trace conditioning-gap between CS offset and US onset
Delay condoning-CS offset overlap with US onset
Delay is easier learn (no time gap)Time gap requires hippocampus
Blocking
Pair CS1 with US and add in CS2 alongside
Presenting CS2 alone does not trigger UR
Operant conditioning
US presented when animal performs behaviour, outcome reinforced or punished
Thorndike’s law of effect- more likely to repeat behaviour that has a pleasant outcome in similar situation
Positive reinforcement-reward
Negative reinforcement- remove unpleasant
Positive punishment-unpleasant is present
Negative punishment-remove pleasant reward
Reward magnitude
Pigeons peck keys for large or small reward
Bigger reward=faster learning initially but levels off
Shaping
Divide learning goal into stages, reinforce individual steps
Primary and secondary reinforcers
Primary reinforcer- stimuli naturally preferred
Secondary reinforcer-neutral event associated with primary reinforcer e,g, bell for food
Cognition
Basis for intelligent behaviour, overrides reflective habitual responses for more complex goals
Gathers info from memory systems and interacts with motor systems
Top down processing-higher order goals
The cortex and birds
Birds do not have neocortex
Pallium more like a cortex (grey and white matter covering cerebellum)
Nidopallium (NCL) is functionally compared to mammalian prefrontal cortex, contains caudolateral
Scala naturae
Living hierarchy, simple to more complex
Assumes only humans have insight related cognition (neocortex)
Working memory and monkeys
Delayed response task, left or right light for 1 sec, 2-3 sec delay before has to move lever in direction of light
Neurons increase activity in delay period to hold in working memory or decrease when doesn’t need to
Dopamine and classical conditioning
Before learning-spike in dopamine after reward
After learning- spike after conditioned stimulus ( not reward)
Reward omission (decrease in responses after but spike still after conditioned stimulus)
Dopamine and working memory
Dopamine tag sensory cues as relevant
Neurons show delay activity while item kept in working memory
Delay between stimulus and choice may be basis for working memory
Pigeons and working memory and findings
Directed forgetting, record activity in NCL
Remember: pecks sample, remember cue, delay and test period
Forget: animal pecks sample cue, forget cue for 2 secs
Activity increase in remember cue phase and delay (holds in WM) and test period
Forget trials: activity activated but decreases as doesn’t hold info in WM
Crows working memory and findings and neurons
Delayed match to sample task, choose from 4 options after delay
Selection performance above chance
Sample selective neuron (selective during samples) distinguishes samples with firing rate
Delay selective neuron (selective during delay)
More than half neurons were either sample or delay selective or both
Firing rate of neurons does not distinguish different stimuli as well as in correct trials
Bird brain conclusions
Delay activity in NCL bridges time gap between stimulus and choice, neural basis for working memory
Neural circuits for WM have probably evolved several times and do not require neocortex
NCL and prefrontal cortex might be functionally analogue
Bees working memory and results
Delayed match to sample task in a tunnel
short delay: high performance
WM up to 6 seconds
Incorrect pattern:still identify correct sample
Rule learning crows and results
Pick stimulus from sample or pick stimulus not from sample
Visual/auditory cue determines which
Measure NCL activity
High NCL firing for non match, low for match rule
Rule learning bees
Two samples in tunnel
First indicates correct pattern OR second indicates correct pattern
Can learn both
Bees generalisation
New visual stimuli in tunnel
Still performs above chance
Indicates bees can generalise to more abstract task rule
Learning abstract concept bees
Delayed match to sample
Match odour at entrance to one in decision chamber
Perform above chance, cross modal transfer of learning
Higher order cognitive functions
Self recognition
Theory of mind
Object permanence
Self recognition: birds and mirror task
Yellow dot on bird, looks in mirror
Mark directed behaviour (touch mark with foot or beak)
Self directed behaviour (touch other body parts)
Activity towards mark (mark directed behaviour divide by self directed behaviour)
Birds show significant MARK DIRECTED BEHAVIOUR
Self recognition and evolution
Neural mechanisms enable self recognition, evolved several times in vertebrates
Laminated cortex not a necessity for self recognition
Evidence against scala naturae and cortex is special
Theory of mind: ravens
Focus on ground
Raven on ground (focus) given food and hides it. Another raven observes it (visible to focus) and control cannot see where food is (blind)
Either sighted or blind observer taken out with focus
When focus is with observer, focus hides food again twice as often. Aware the bird has knowledge of where the food was
Theory of mind: ravens
Focus as observer
Focus observes ground bird hide food, has a blind co-observer and sighted co-observer
Focus bird taken out with blind OR the other sighted bird with hider
Focus runs to food when with co-observer (knows he saw too) much faster than with blind. Take longer with blind time make it seem like he doesn’t know either
Distinguishes between different levels of knowledge
Object permanence in birds
Higher Piagetian stages develop gradually as bird ages. Step from 4 to 5 is challenging- A not B error before stage 5 (search where object was hidden previously)
Can reach full object permanence
Comparing rodents vs birds in higher order cognitive function
Rodents have cortex unlike birds
Rodents easy to maintain, birds hand reared
Rodent episodic memory, birds several forms of insight related cognition
Rodents strong olfaction, birds visual and auditory (like humans)
Rodents bad for cognitive ageing vs birds live long time
Rodents can use ontogenetics unlike birds
