Cognitive Psychology Flashcards
1
Q
Memory
A
- Memory is the faculty, by which the mind stores and remembers information (Oxford dictionary).
- Memory involves acquisition (encoding), storage, retaining, retrieval, and forgetting.
- Memory processes are closely related to learning.
- Three major processes:
∙ Encoding.
∙ Storage.
∙ Retrieval.
2
Q
LTM: Declarative vs. Procedural
A
- Declarative (explicit):
∙ Knowing “what”. Memory for facts and events. These memories are consciously (explicitly) recalled (E.g. your address or what you wore on a night-out)
∙ Relational processing: Items are bound by association to the experience in which they occurred, related events, and other memories. This can result in creating of false memories when relations are retrieved that did not actually occur in the given instance but are common in the network of relational memories. - Procedural (implicit):
∙ Knowing “how”. Memory of skills, how to do things, the use of objects, or movements of the body. These memories are unconscious (implicit) (E.g. playing a guitar or riding a bike).
∙ Procedural memories are accessed and used without conscious control or attention. Often investigated via priming.
3
Q
Declarative Memory: Episodic vs. Semantic
A
- Episodic memory: Memory of past events (e.g. times, places, context).
- Semantic memory: Memory of meanings, concepts, facts, and general knowledge.
4
Q
The Multi-Store Model
A
- Sensory store: Modality specific, brief, subject to rapid decay.
- Short-term store: Limited capacity, fragile. Forgetting from STM is often explained by trace decay and/or displacement.
- Long-term store: Unlimited capacity. Forgetting from STM is explained by interference, cue-dependent forgetting, and/or problems with consolidation.
5
Q
Sensory Stores
A
- Iconic (visual) storage
∙ Information decays in iconic memory after about 0.5 seconds.
∙ With reduced task demands iconic memory lasts up to 1.6 seconds - Echoic (auditory) storage
∙ Unattended auditory information lasts about 2-4 seconds in echoic storage.
6
Q
Short-term memory
A
- Supports brief storage and immediate recall.
- Stores information that is currently activated (i.e. attended).
- Activation decay and switching attention cause information to leave the short-term store.
- Decay can be prevented by rehearsal (re-activation).
7
Q
STM: Memory Span
A
- Miller (1956) “Magic Number Seven”
∙ The maximum number of units recalled with minimal error is seven plus or minus two.
∙ Memory span is not limited to a certain number of stimuli, but by the number of chunks. - Chunking
∙ Grouping a series of apparently random items into a smaller number of meaningful segments enhances recall.
8
Q
Serial Position Effects
A
- Primacy effect: The tendency to better recall the first few items on a list.
- Recency effect: The tendency to better recall the last few items on a list.
9
Q
Decay Over Time in STM
A
- Decay of information in STM depends on whether information is rehearsed.
- Without rehearsal (e.g. masked by counting backwards by threes), people forget items in the list almost completely after 18 seconds.
10
Q
The Multi-Store Model:
Evidence from Neuropsychology
A
- Are STM and LTM independent? What do we consider classical direct evidence for the distinct nature of brain systems?
- Double dissociation! Patients with distinct LTM and STM impairments (double dissociation)
- Spiers et al. (2001)
∙ Retrograde amnesia: damage to the medial temporal lobe.
∙ Impaired long-term memory.
∙ Intact short-term memory. - Challice & Warrington (1970)
∙ Anterograde amnesia: damage to the parietal and temporal lobes.
∙ Intact long-term memory.
∙ Impaired short-term memory.
11
Q
Baddeley’s Working Memory System: Central Executive
A
- Central executive – responsible for directing attention to relevant information, suppressing irrelevant information, effortful control, and for coordinating cognitive processes in multitasking.
- Phonological loop and VSS are central executive’s “slave systems“ responsible for maintenance of information
- Central executive “supervises” and “coordinates” the slave systems.
12
Q
Implicit Learning
A
- Implicit learning is a form of learning that occurs when the learner is unaware of what they learned.
1. in the absence of intention to learn
2. in the absence of awareness of learning
3. the resulting knowledge is difficult to express explicitly - Real-life learning nearly always involves both implicit and explicit mechanisms.
- Implicit learning may be most important during early development.
13
Q
Methods: Artificial Grammar Learning
A
- Protocol:
∙ Training phase
∙ Testing phase
∙ Debriefing - Results:
∙ Participants perform above chance at testing
∙ Participants are unable to explicitly describe the learned rules
14
Q
Evidence for Implict Memory from Brain-Damaged Patients
A
- Double dissociation between performance on explicit (impaired) and implicit (unimpaired) tasks in amnesiacs
- Patients with damage to striatum show impaired implicit performance
∙ Siegert et al. (2006)
∙ Saint-Cyr, Tyalor, and Lang (1988)
15
Q
Why do we forget?
A
- Decay theories. Memories fade as a function of passage of time .
- Interference theories. Memories are “forced out” by preceding and succeeding events.
- Cue-dependent forgetting. Memories fade due to lack of supporting contextual cues.
- Memory repression. Memories are (un)consciously blocked due to the high level of associated stress.
- Motivated forgetting. Conscious memory repression.
16
Q
Saving Method: Ebbinghaus (1885)
A
- Experimental protocol:
∙ Learn a list of nonsense letter combinations (CVC trigrams).
∙ Recall them at various intervals.
= Forgetting is fastest shortly after learning and the rate then decreases with time.
17
Q
Forgetting: Implicit vs Explicit Memory
A
- Forgetting is slower for implicit than for explicit memory
∙ Tulving, Schacter, & Stark (1982): Memory for a word list was poor after one week (explicit memory) but word-fragment completion (implicit memory) was unaffected.
∙ Mitchell (2006): Participants still had implicit memory for pictures 17 years after the initial experiment!
18
Q
Cue-Dependent Forgetting
A
- Cue-dependent forgetting – failure to recall a memory due to missing stimuli or cues that were present at the time of encoding (learning context).
- Tulving (1976) :
∙ Retrieval success is directly related to the degree of overlap between the information presented at retrieval and the information stored in memory, including its context. - Thompson and Tulving (1970) - Recall performance is:
∙ Best when strong cues appear at study and recall
∙ Worst when there is a mismatch of cues at study and recall
19
Q
Memory Repression
A
- Freud (1915/1963)
∙ Threatening or traumatic memories are often inaccessible to conscious recall. - Most evidence is based on patients with recovered repressed memories.
- Non-experimental evidence, in which adults recover repressed memories is controversial.
∙ Lawsuits sparked a “memory war” over whether recovered memories are reliable.
✳︎ Lief and Fetkowicz (1995): self-report data suggest that recovered memories recalled inside therapy may be more likely to be false than those recalled outside therapy.
20
Q
Recovered Memories
A
- Clancy et al (2000) induced false memories in the lab.
✳︎ People are given lists of semantically related words to remember.
✳︎ Later, they often “remember” semantically related words that were not on the list.
✳︎ Those who reported recovered memories of abuse showed higher levels of false recognitions.
21
Q
Memory Consolidation
A
- A process that fixes information in long-term memory
∙ The hippocampus is assumed to play a vital role in consolidation process
∙ Prior to memories being stored, unconsolidated memories are highly vulnerable
✳︎ “New memories are clear but fragile and old ones are faded but robust” (Wixted, 2004)
22
Q
Standard Model of Consolidation
A
- Information is consolidated by transferring it from one brain region to another (system consolidation)
1. The hippocampus detects and stores novel information
∙ Takes place over hours
2. The information is gradually transferred from the hippocampus to neocortical areas
∙ Takes place over days
∙ Sleep plays an important role in memory consolidation
23
Q
How do we know STM and LTM are distinct systems?
A
- Anterograde amnesia: Problems remembering new information learned after the onset of memory loss
- Retrograde amnesia: Problems remembering events old information – acquired prior to the onset of memory loss
24
Q
Declarative memory
A
- Episodic memory: Events, Relations
- Semantic Memory: Objects, Word meanings, Facts, People
25
Q
Are Episodic and Semantic Stores Independent? Evidence from Amnesia
A
- Anterograde Amnesia: Patients with anterograde amnesia due to damage to hippocampus often have episodic impairments but little or no semantic impairments (Spiers et al. 2001; Vargha-Khadem et al. 1997).
- Retrograde Amnesia: Retrograde amnesia for episodic memory is also more severe than for semantic memory (Tulving 2002; Yasuda et al. 1997).
26
Q
Constructive Nature of Episodic Memory
A
- Episodic memory is (partially) a constructive system. Prone to errors and illusions (remember false memories?).
- Why?
►Too much resources necessary to produce a permanent record
►We usually access the gist rather than details. This tendency increases with age (Brainerd & Mojardin 1998)
►Constructive processes involved in episodic memory might be related to prediction - forming future plans
✳︎Amnesiacs with episodic memory loss also have impaired ability to imagine future events (Hassabis et al., 2007)
27
Q
Episodic memory: How is it studied?
A
- Recall
- Recognition
28
Q
Semantic memory
A
Semantic memory stores facts, categories, and concept. Like a database, thesaurus, or a dictionary
29
Q
Hierarchical Network Models
A
- Semantic memory is organized into a series of hierarchical networks (like a taxonomy)
►Concepts are represented as nodes
►Features are associated with each concept
►Each concept possesses features of higher-level concepts in addition to its own features - Hierarchical Distance Effect: Sentences should take more time to process as the number of levels between the tested concept and specific features increases
30
Q
Limitations of Hierarchical Network Models
A
- Familiarity effect:
►Hierarchical distance is often confounded with familiarity
►Controlling for familiarity reduces the hierarchical distance effect (Conrad 1972) - Typicality effect:
►Some category members are more typical than others
►Verification is faster for features of more representative (typical) members (Rips, Shoben, & Smith, 1973) - E.g., “PENGUIN is a bird” slower than “CANARY is a bird”
31
Q
Feature Models
A
- A feature-based approach assumes:
►Living things are distinguished based on their visual/perceptual features (i.e., what they look like)
►Non-Living things are distinguished based on their functional properties (i.e., what they’re used for)
✳︎ Many brain-damaged patients exhibit category-specific deficits, more commonly for living rather than for non-living, objects (Martin & Caramazza, 2003)
32
Q
Limitations of Feature Models
A
- Dividing information into just two basic categories (sensory and functional) may be too simplistic
- Multiple-property approach (Cree & McRae 2003)
►Further subdivisions of the two categories:
∙ Perceptual properties divided into: Visual, auditory, taste, and tactile sensations
∙ Functional properties divided into:
∼ Entity behaviours (what a thing does)
∼ Functional information (what humans use it for)
33
Q
Simulation Models
A
- Barsalou (2008):
►Semantic memory (and retrieval from it) is based on simulation: “The re-enactment of perceptual, motor, and introspective states acquired during experience” (p. 618) - Hauk, Johnsrude, and Pulvemüller (2004):
►Verbs activate parts of the motor strip corresponding to the areas associated when performing an action - Pulvemüller et al. (2005):
►Arm-related words are facilitated when TMS is applied over the arm site on the motor strip
►Leg-related words are facilitated when the leg site is stimulated
34
Q
Procedural memory
A
Procedural memory does not involve conscious recollection but reveals itself through behaviour
35
Q
Procedural Memory: Priming
A
- Priming refers to an increased sensitivity to a stimulus due to prior exposure (Tulving & Chacter 1990)
- A way to access implicit memory
- May be a major vehicle of implicit learning
- Priming is often preserved in amnesic patients with impaired declarative knowledge
36
Q
Perceptual Priming
A
- Occurs when repeated presentation of a stimulus leads to facilitated processing of its perceptual features
- The prime and the target share perceptual properties
- Patient evidence:
►Intact conceptual priming
►Impaired perceptual priming
37
Q
Conceptual Priming
A
- Occurs when repeated presentation of a stimulus leads to facilitated processing of its meaning
- The prime and the target are related semantically
- Patient evidence:
►Intact perceptual priming
►Impaired conceptual priming
38
Q
Domains of Psycholinguistics
A
- Acquisition/learning ►How do children acquire language? - Comprehension ►How do people form interpretations from speech? ►Reading - Production ►How do people generate utterances?
39
Q
Levels of Representation
A
- Semantics: Meanings of words and utterances
- Syntax: Combinatorial rules, word order
- Lexicon: Words, Morphology
- Phonology: Sounds, phonemes, prosody, intonation
40
Q
Reading vs Speech Perception
A
- Reading: Evolutionarily recent, Provides punctuation marks, More ambiguous
- Speech Perception: Evolutionary older, Uses prosody instead of punctuation, Less ambiguous
41
Q
What helps to avoid ambiguity?
A
- Prosodic cues in speech perception ∙ Pitch ∙ Intonation ∙ Stress - Main cues to sentence structure in text are punctuation marks
42
Q
Research Methods in Reading
A
- Lexical decision task (LDT) ► Is “WUG” a word (yes/no)? - Naming task ► Say a printed word out loud as quickly as possible. - Eye tracking - Priming - Neuroimaging
43
Q
Models of Phonological Access in Reading
A
- Weak Phonological Model: Coltheart et al. (2001): Phonological processing of visual words is inessential for word identification
- Strong Phonological Model: Frost (1998): Phonological coding during reading is essential for word identification
44
Q
Support for Strong Models: LDT: Van Orden (1987)
A
- Used homophones in LDT
∙ Homophones: Words that sound the same but have different meanings (e.g. bank). They are also often spelled differently (e.g. rows/rose). - Participants made more errors when asked questions like:
∙ “Is ROWS a flower?”
than
∙ “Is ROBS a flower?” - Conclusion:
∙ Participants engaged in phonological processing, mistaking “ROWS” for “ROSE”
45
Q
Support for Weak Models: Phonological Dyslexia: Patient PS (Hanley & McDonnell, 1997)
A
- Understood meanings of words
∙ Could give accurate definitions of printed words - Could not pronounce words accurately
- Evidence that access to word meaning is independent of phonological access
46
Q
Does Phonological Processing Always Occur?
A
- Use of phonology depends on the nature of the words and participants’ reading ability.
∙ Phonology is used more when accessing the meaning of low-frequency words than high-frequency words.
∙ Poor readers are more likely to access phonology. - Accessing phonology in order to understand a word during reading is not all-or-nothing process.
47
Q
Phonological Processes in Reading: Evaluated
A
- Weak phonological models are probably too weak
∙ Phonological processing often occurs during reading
∙ It can be quite rapid and automatic - Strong phonological models are probably too strong
∙ Phonological processing is not always necessary
∙ Phonological dyslexia
►Difficulties processing phonology
►Intact ability to access lexical meaning
48
Q
Automatic (Holistic) Word Processing
A
- Rayner (1998): People often skip (don’t fixate) words when they read suggesting that word identification can be automatic
- Cheesman & Merikle (1984): Stroop effect (remember what it is?) suggests word meaning is automatically extracted even when participants try not to process the visual form (i.e. letters)
49
Q
Word Superiority Effect
A
- Faster RTs and fewer errors when letter is presented in a word.
- Word accessed as a whole, letter identification happens later (Reicher 1969)
50
Q
Research at Cambridge Uni
A
- Rayner et al (2006) found that
∙ All kinds of letter transpositions slow reading speed
∙ Matters which letters are transposed
∙ Matters in which words the transposition occurs - Suggests that letter-level and word-level processing occur in parallel
51
Q
Speech Processing Stages
A
- Stage 1: Phoneme identification: key = /k+i/
- Stage 2: Syllable identification. E.g. water = wa + ter
- Stage 3: Word identification.
- Stage 4: Utterance comprehension.
- Stage 5: Meaning integration.
52
Q
Problems Faced by Listeners
A
- Segmentation problem: Difficulty separating out or distinguishing words from the pattern of speech sounds due to the continuous nature of speech signal
- Individual differences in rate of speaking
- Degraded speech
∙ Environmental noise
53
Q
What Helps Segmentation?
A
- Certain sounds never appear together within a syllable. This cues a likely boundary between words: e.g., [m,r] in English
- Possible-word constraints: e.g., a stretch of speech without a vowel (in English) couldn’t be a word.
- Syllabic stress: e.g., in English, initial syllables of most content words are stressed. In some languages the position of the stress in a word is fixed.
54
Q
Speech perception
A
focuses on lower-level processes (e.g. phonological and lexical access).
55
Q
Sentence processing
A
focuses on higher-level processes (i.e. syntactic and semantic analysis).
56
Q
Two main levels of analysis
A
- analysis of the syntactic structure (parsing)
- analysis of the resulting sentential meaning
57
Q
Pragmatics
A
studies how we understand the intended (vs. literal) meaning
58
Q
Discourse and text studies
A
focus on how the meanings of individual utterances/sentences are integrated at the level of passages and the whole story.
59
Q
Sentence processing (2)
A
- takes place when a reader or listener attempts to understand a sentence
- occurs: in isolation, or situated in context (linguistic, pragmatic, visual, etc.)
60
Q
Parsing
A
the process of building a syntactic interpretation during sentence processing. Leads to assigning the incoming words to sentence roles (e.g. Subject, Object).
61
Q
Parser
A
abstract sentence processor.
62
Q
Parse
A
a temporarily built interpretation (structure) that may or may not be the final/correct one.
63
Q
Global Ambiguities
A
More than one interpretation possible for the sentence as a whole
64
Q
Local Ambiguities
A
- persist only temporarily
- multiple interpretations are possible only at initial processing stages
- once the remainder of the sentence is known, only one interpretation remains
65
Q
Local Ambiguities: Garden-path sentences
A
- The garden path effect occurs when the parser commits to a wrong (more likely) interpretation leading to a “dead end” when later input is inconsistent with the initially committed interpretation.
- In this case, the structure needs to be revised and repaired. In reading, this results in longer reading times, more interpretation errors, and reverse eye-movements (regressions).
- Understanding how we resolve GP and other syntactic ambiguities helps us understand how we build syntactic structure when we process sentences.
66
Q
Models of Sentence Processing
A
- Serial (two-stage):
1. Initially, reader commits to a single interpretation.
2. Reader then realizes this interpretation is incorrect
3. Reader uses available disambiguating information in order to resolve the conflict
4. The initial interpretation is then abandoned, the new correct interpretation is formed - Parallel (one-stage):
- Reader simultaneously considers multiple interpretations until disambiguating information is available
- After that, only the correct interpretation is maintained.
- Mixed:
- Reader may use either serial or parallel strategy depending on what’s easier:
- to commit to the most likely initial interpretation hoping it’s correct
- to construct multiple interpretations & store them to avoid reanalysis
67
Q
Serial Models: Frazier (1987); Frazier & Rayner (1982)
A
- STAGE 1: SYNTACTIC ANALYSIS The simplest/most likely syntactic structure is selected. Only syntactic information is used at this stage.
- STAGE 2: SEMANTIC ANALYSIS The syntactic structure is interpreted for its meaning. In case of ambiguity, go back to Stage 1.
68
Q
Parallel Models: MacDonald et al (1994)
A
- Parallel models don’t contradict incremental nature of sentence comprehension. However, not one but many possible interpretations are considered in parallel.
- Context (lexical, semantic, etc.) is used as soon as possible in order to constrain syntactic possibilities and resulting interpretations.
69
Q
Serial vs Parallel Models: Evaluated
A
- Serial Model:
- Strengths:
-Simple account
-Minimal set of principles
Limitations:
-Inflexible
-Context can influence interpretation early
-Doesn’t account for cross-language differences - Parallel Model:
- Strengths:
-Use all relevant information to for quick interpretation
-Assumes flexibility in parsing decisions - Limitations:
-Not all relevant constraints are used at the same time
-Doesn’t apply easily to complex sentences
-Little direct evidence of fully parallel representations
70
Q
Good-Enough Models: Ferreira, Bailey, & Ferraro (2002), Ferreira (2003)
A
- The typical goal of comprehension is to parse the input well enough to meet task demands
- Heuristics (“rules-of-thumb”) are applied to simplify and to speed-up the interpretation task
- The balance between syntactic and semantic analyses can be determined by these heuristics
71
Q
Discourse and Story
A
- The highest levels of meaning integration
- Discourse logically and inferentially connects several sentences
- Story processing involves relating overall text information to relevant knowledge in long-term memory
- Discourse processing: Inferences and Schemas.
72
Q
Schemas
A
are integrated packets of knowledge about the world, events, people, and actions.
73
Q
Schemas: Evaluated
A
- Strengths:
-Schematic knowledge aids text comprehension
[Also responsible for many memory errors and distortions]
-Double-dissociation in the neuropsychological literature between schema-based and lower-level knowledge impairments - Weaknesses:
-Too abstract. Difficult to identify precise characteristics of individual schemas
-Most schemas are largely untestable
-Conditions determining when a given schema is activated are unclear
