Reading...
Play button
Play button
Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
150 Cards in this Set
- Front
- Back
- 3rd side (hint)
|
Categories
|
Organizational groups for semantic memory
|
|
|
|
Examplars and rules
|
Exemplar - Multiple individual category members that we can compare instances to. These category members are specific real examples.
Rule - A rigid set of requirements about the properties required for membership in a category — e.g An animal is a builder if it has long legs, angular body, spots; otherwise it is a digger. (verbal) |
Allen and Brooks 1991 study on 2 imaginary animals – builders and diggers.
Participants were shown pictures of animals sequentially. Once they could apply the rules researcher introduced a surprise test on an animal they have not seen before. Researchers suspected that even though knew a rule for the categories; they might nevertheless be storing exemplar memories and using them in categorization. Earlier research showed that the brain automatically stores and uses exemplar memories even when it is not necessary. They introduced 2 new builders. One differed from previous builders in only one characteristic (positive match) and the other new builder differed from previous digger by one characteristic (negative match) . If participants don’t store exemplar knowledge and use only the rule then can easily categorize animals shown. But if they did store exemplar knowledge then hard to categorize the negative match animal as the counterpart was in the other category (digger) previously; leading to a rule and exemplary memory conflict. Outcome – Exemplar category stored and impacts categorization. Do we then store exemplars but not rules? The researchers used a second group of participants but this time didn’t tell then the rules categorizing builders and diggers. Rule participants had a stored a rule in their category knowledge that made them less vulnerable to exemplar memories than were no rule participants. By applying the rule on some occasions, rule participants were more likely to categorize negative match animals correctly. Different brain systems become active to represent exemplars and to represent rules. Furthermore, the particular systems that become active support the notion that category knowledge is represented in modality-specific areas: visual areas represent the content of exemplars; motor areas implement the process of rehearsing rules. |
|
|
Prototypes
|
A best average example.
|
|
|
|
Typicality effects
|
Typical category members are instances that are more similar to the prototype/exemplar, whereas atypical category members are instances that are more different from the prototype/exemplar.
If the prototype indicates that birds fly than most flying birds are typical along the typically gradient but the ostrich is atypical because it can’t fly. Many theories assume that prototypes develop to represent categories. Typicality effects can arise, even if no prototypes have been stored for a category and only exemplar memories represent the category. (Medlin &: Schaffer,1978). Typicality is on a gradient from highly typical to highly atypical. |
|
|
|
Basic level categories
|
The level of a taxonomy used most often, learned most easily, and processed efficiently - usually middle level categories
Across all cultures, traditional and developed , there is high agreement in middle-level categories: names for middle-level categories such as “deer,” “eagle,” and “alligator” tend to be found wherever these animals are part of the environment. “Deer,” for example, refers to pretty much the same creatures across cultures. Cultures typically have names for such categories even when they play no role in people’s lives. Many plants and animals that have no function still feature in the middle level of cultures e.g wild animals. Middle levels in taxonomy are most important because their members (unlike members of categories at higher and lower levels) typically share a common configuration of physical properties, and this configuration differs that of other categories at the same level. For example, deer have four legs, two pointed hooves, a tail, and other physical properties in a particular arrangement. Butterflies have head, body, antennae and large flat wings. Why is middle level taxonomy so salient? Our visual feature-detection systems have become tuned over evolutionary time to morphological features that distinguish different categories from one another. However researchers don’t quite accept this basic level because middle-level categories are not always the dominant ones. In western culture, for example, the middle level for categories of plants and birds are not dominant because many don’t know much about plants and birds to name them at mid-level taxonomy; they use a higher level. Similarly an expert can process lower levels as well as he can middle levels. |
Names for categories at higher and lower levels show much less agreement across cultures ( Malt, 1995). For example, cultures that eat butterfly larvae have many categories for different kinds of larvae; cultures that don’t, don’t.
When similar categories at higher and lower levels do exist in two cultures, they often don’t refer to the same sets of things in the environment. One culture might have a category for “trees,” another might have a category for “firewood” that includes only trees that are burnable. Furthermore, both high- and low-level categories are likely to deviate from scientific taxonomies than are middle-level categories. There is no common morphological structure within a category in the higher levels, not all mammals are built like deers. To discriminate the difference between categories, it is necessary to learn subtle visual features. |
|
|
Types of Long Term Memory
|
Explicit memory: semantic and episodic.
Implicit memory: priming, procedural, associative, non-associative. |
|
|
|
Explicit Memory
|
Declarative memory that is long term memory that ordinarily can be consciously recollected and declared.
It includes episodic memory and semantic memory. Episodic memory – memory of events in our own personal past. Temporally dated, spatially located, and personally experienced events or episodes. Has context. Semantic memory – general knowledge of things in the world and their meaning. Knowledge of words and concepts, their properties, and interrelations. Lacks context. Tested with explicit memory tests that require the retrieval of explicit description or report of knowledge from memory |
|
|
|
Implicit Memory
|
Nondeclarative memory that is nonconscious forms of long term memory that are expressed as a change in behavior without any conscious recollection.
Tested with implicit memory tests that do not require descriptions of the contents of memory but reveal memory implicitly through observed changes in performance. Includes priming, procedural, associative, non-associative. |
|
|
|
H.M. case study
|
Surgery removed hippocampus, amygdala, and much of the surrounding medial temporal cortices.
Working memory, intelligence, and long term memory acquired before surgery intact. Demonstrated anterograde amnesia, that applied to both episodic and semantic knowledge. Demonstrated some temporally grade retrograde amnesia. The closer an event to his surgery, the more likely it is to be forgotten. Indicates that memory is not permanently dependent on medial temporal lobes but is lodged somewhere else over time. Medial temporal lobes are critical to declarative long term memory. Nondeclarative memory remained intacts – he was able to learn mirror tracing skill. Amnesiac patients like H.M. can also prime. Warrington and Weiskrantz showed amnesiac patients a list of words (e.g. ABSENT, INCOME, FILLY), then told them to complete word stems from the original list withour referencing the original list. Priming effect was demonstrated, i.e. there was an increased likelihood of generating a response related to a stimulus previously presented (the original list). |
|
|
|
Anterograde amnesia
|
The inability to consciously remember information encountered after brain damage. Inability to form, retain, and retrieve new episodic memories.
|
|
|
|
Retrograde amnesia
|
The forgetting of events that occurred before the damage to the brain.
|
|
|
|
Brain regions involved in different types of memory.
|
Medial temporal lobes are critical to declarative long term memory.
Greater engagement of the frontal lobe attentional mechanism increases encoding efficacy with the left frontal lobe supporting the encoding of words, and the right frontal lobe supporting the encoding of nonverbal stimuli. Medial temporal lobe (especially the hippocampus) is convergence zone where multiple features (can be from any modality or depth be it perceptual or semantic information) can be bound into an integrated memory representation. It is necessary for episodic memory. Lesions of the right hippocampus give rise to greater deficits in nonverbal episodic memory while lesions of the left hippocampus give rise to greater deficits in verbal episodic memory. Right hippocampus activates more during the encoding of unfamiliar faces while left hippocampus activates more during the encoding of words. Verbal and nonverbal representations are bound togerther within the medial temporal lobes Consolidation is the process that modifies representations such that they become more stable over time and ultimately exist independently of the medial temporal lobes. Interactions between the medial temporal lobes and various lateral cortical regions are thought to store memories outside the medial temporal lobes by slowly forming direct links between the cortical representations of the experience Neurological studies support the notion that recollection differentially depends on hippocampal memory mechanisms while familiarity depends differentially on perihinal memory mechanisms. Hippocampus neurons differentially signal memory for the conjunction between stimulirather than for individual stimuli while perihinal neurons differentially signal stimulus familiarity |
|
|
|
Levels of processing
|
Levels of processing – Different aspects of stimulus processing correspond to different levels of analysis, ranging from a “shallow” level of perceptual analysis to a “deep” level of semantic analysis.
The deeper, the level of analysis the better you remember. Episodic memory substantially benefits from elaboration of the meaning of a stimulus or event at time of encounter. Problems: No independent measure of depth of processing required by a particular encoding task other than its impact on memory, making it hard to test. |
|
|
|
Incidental learning
|
Learning does not have to be intentional. It can be done as a byproduct of completing a task.
|
|
|
|
Transfer appropriate processing/Encoding specificity principle.
Morris et al. study on pages 206-207. |
Processing at encoding is most effective to the extent that processing at encoding overlaps with processing to be performed at retrieval.
Encoding specificity principle – Our ability to remember a stimulus depends on a similarity between the way a stimulus is processed at encoding and the way it is processed at retrieval. Morris and colleagues (1977) Participants had to encode words by making a rhyme decision or a semantic decision about each word. During retrieval memory was retrieved by a task requiring recognition of words that had previously been studied (semantic, levels of processing) or a task requiring recognition of words that rhymed with previously studied words. Encoding rhyme tested with rhyme test and encoding semantic tested with standard test both performed better than encoding rhyme with standard test and encoding semantic with rhyme test. Processing at encoding is particularly effective to the extent that it overlaps with processing at retrieval. Level of processing may work because it influences what is encoded instead of influencing the strength or durability of encoding. |
|
|
|
Enhancers of encoding: generation effect, spacing effect (massed vs. distributed practice).
|
Generation effect – you are more likely to remember information you retrieve or generate during study than information you simply receive and attempt to “memorize”
Spacing Effect – Distributed practice where trials with the same stimulus are separated by other stimuli is more effective for encoding than massed practice where many trials with the same stimulus are undertaken without interruption. In massed practice, less likely to attend fully to each presentation. With each subsequent presentation we are deluded into thinking we’ve learnt an item an allocate increasingly less attention to it. In addition in massed practice, repeated trials are likely to be highly similar, so the same aspects of stimulus are selected for encoding each time. Distributed practice fosters greater encoding variability because the context is more likely to have change more between trials, so different aspects are selected for encoding at each trial, resulting in a richer memory representation and additional routes back to memory. |
Slamecka & Graf, 1978
Reading condition – word pairs presented and participant must decide if 2nd word is synonymous (semantic) to or rhymes with the 1st word. Generate condition – participants asked to generate a synonym or a rhyme from first word. Memory was better after semantic encoding than after phonological encoding (levels of processing) Process engaged during initial generation at encoding are likely to overlap with those required to generate the information at retrieval (transfer –appropriate processing) Memory was better for the generate condition than the read condition |
|
|
Convergence Zone (in the brain).
|
Area of brain where multiple features (can be from any modality or depth be it perceptual or semantic information) can be bound into an integrated memory representation
Medial temporal lobe (especially the hippocampus) is a convergence zone |
|
|
|
Role of the hippocampus/medial temporal lobe in memory consolidation and pattern completion/recapitulation.
|
Consolidation is the process that modifies representations such that they become more stable over time and ultimately exist independently of the medial temporal lobes.
Interactions between the medial temporal lobes and various lateral cortical regions are thought to store memories outside the medial temporal lobes by slowly forming direct links between the cortical representations of the experience Hypothesis 1: memory consolidation in the cortex occurs through process of reinstatement or recapitulation, wherein during sleep and during remembering the medial temporal lobes recapitulate the pattern of activation during learning, thus strengthening the direct connections across the relevant lateral cortical regions. Pattern completion – an episodic memory which is a conjunction of linked features is built from one feature. Retrieval also entail recapitulation which is a reinstatement of pattern activations that was present during encoding. In retrieval a partial cue to the hippocampus triggers pattern completion, and the hippocampus projects back to cortical areas and replays the pattern of activation that was present during encoding. Medial temporal lobe activation precedes recovery of episodic knowledge. The pattern of cortical activation during retrieval has been found to be similar to the pattern of activation during encoding. Interference between competing memories affects retrieval. Retrieval is cue dependent – it is stimulated by hints and clues from the external and internal environment. When cues are absent or unused, pattern completion is less likely. Context and states are particularly strong cues. Context dependent effect – retrieval is typically better when the physical environment at retrieval matches that at encoding State-dependent effects – better retrieval when internal states at retrieval match those at encoding Dual process theories of recognition assert that both recollection and familiarity can support recognition. Recollection relies on the same pattern completion mechanisms that allow recall of episodic info associated with a retrieval cue. Familiarity emerges from a different process that takes account of overall similarity instead of detail. Recollection is slower than familiarity, so in situations requiring rapid recognition we rely on familiarity. Recollection is dependent on attention at the time of encoding and retrieval. If attention is divided, the use of recollection in recognition decisions decreases. |
|
|
|
Reconstructive Memory
|
Memory is reconstructive. We reconstruct the past during retrieval rather than reproduce it. When our memories are not clear, we may infer the way things “must have been” from our current thoughts and expectations.
Biases imply that memory is reconstructive. Bias – the inclination toward a conclusion not justified by knowledge or logic. Belief bias - background knowledge about the way of the world and personal beliefs unconsciously influences memory to reshape it in a form consistent with expectations Consistency bias – biases resulting from the often erroneous belief that one’s attitudes are stable over time, therefore unconsciously adjusting memories to bring the past in line with the present |
|
|
|
Consistency Bias
|
Biases resulting from the often erroneous belief that one’s attitudes are stable over time, therefore unconsciously adjusting memories to bring the past in line with the present
|
|
|
|
Misattribution and Misinformation Effects
|
Misattribution – Ascribing a recollection to an incorrect time, place, person, or source.
False recognition in word lists could result from related words being activated by reading words on word list and misattributing the memory of having thought the word to having seen the word. False recognition occurs when we encounter a stimulus that although not previously encountered is semantically or perceptually similar to previously encountered stimuli Structures in medial temporal lobes responsible for accurate episodic memory are also involved in storing and retrieving the info that leads to false recognition. This is shown by amnesiac patients showing lower levels of false recognition than healthy participants. Misinformation distorts memory through a combination of misattribution and acceptance of suggested misinformation when memory is weak Suggestion – suggestion of false information can 1. lead to the overwriting of previously encoded information with the new misinformation 2. lead to misattribution: false information is presented an encoded. When later tested remember both actual information and false information but can’t remember which is the actual information and which is the misinformation 3. be accepted as truth simply because we can’t remember and submit to the authority of the source of misinformation We may not only accept suggested misinforomation as accurate but may also remember other details beyond those suggested. Inducing people to visualize experiences that never occurred can sometimes lead them to conclude that their representations for what they imagined were memories of real events. Neuroimaging data shows that we are more likely to falsely claim to have seen an object our earlier imagination of the object elicited robust activation of brain regions supporting the perception of the object |
|
|
|
Working Memory
|
Short term mental storage and manipulation operations
|
|
|
|
Retroactive and Proactive Interference
|
Interference theories hold that if the same cue is bound to multiple representations, those representations compete during retrieval resulting in interference.
Retroactive interference – new information disrupts ability to remember older information Proactive interference – old information disrupts ability to remember newer information |
|
|
|
Blocking and Suppression
|
Blocking – The representation with a stronger association to a cue than the representation-cue associations of other representations bound to the same cue “wins” the competition and is remembered. Emphasizes that forgotten information still resides in memory but access to it through a particular cue is temporarily blocked by a dominant competing representation. Weaker representations can be unlocked with alternative cues that are more strongly associated with it. Blocking also explains output interference where the strengthening of memories provided by the act of internal retrieval blacks retrieval of other memories. Internal retrieval of representations for a cue strengthens associations with retrieved items making it harder to retrieve other representations associated with that cue.
Suppression – the active weakening of a memory that occurs because the act of retrieval is competitive. Retrieving involves both strengthening of association between cue and target representation and suppression of the representation’s competing associates. This suppression is not unique to the cue-reprsentation association strength of associates. A suppressed representation is harder to retrieve using alternative cues as well to the extent that the representation was suppressed. |
|
|
|
Types of non declarative memory: priming, conceptual priming, skill learning, habit memory, conditioned associations.
|
Priming
Perceptual priming results in an enhanced ability to identify a stimulus. It is observed in all modalities: vision, audition, and touch. Highly dependent on the degree of perceptual overlap between the initial encounter with the stimulus and repeated ones. The overlap is greatest when initial and subsequent encounters are in the same modality. Perceptual priming does little or nothing if the initial encounter is in a different modality from subsequent ones. Emerges from learning within the sensory cortices. Sensory specific cortical processing is necessary for perceptual priming. Perceptual identification task – words are presented on a screen for a very short time and task is to identify each flashed words. When given a study list of words including test word(s), probability of participants identifying test words that were in study list increases. Measure of priming is difference between accuracy of studied and unstudied stimuli. Neuroimaging shows that visual priming is accompanied by decreased activation in regions of the visual cortex that were engaged during initial processing of stimuli. Why? One hypothesis suggests that the responses of neurons that are not essential for identification are dampened down, resulting in a sparser and more selective neural representation (fewer neurons fire in response to stimulus) and enhanced identification. Conceptual priming Conceptual priming results in facilitated processing of the meaning of a stimulus or enhanced access to concept. Demonstrated by the categorical exemplar generation task. Participants are presented a category cue and asked to generate the first few exemplars of the given category that comes to mind. Probability of spontaneously generating a given exemplar is higher if that word had appeared on an earlier unrelated study list. Conceptual priming is associated with decreased activation in the left inferior frontal lobe and left lateral temporal cortex during repeated conceptual processing of a stimulus. The left frontal lobe is thought to contribute to semantic retrieval when the sought information does not immediately come to mind upon cue presentation. Increased accessibility of sought semantic information due to conceptual priming decreases demands on retrieval process, and decreases the cognitive effort required to retrieve relevant information. Skill learning What are the stages of skill learning? The stages of skill learning are the cognitive stage, the associative stage, and the autonomous stage. Cognitive stage – knowledge is declaratively represented, often in verbal code, and attentional demands are high. Associative stage – Behavior begins to be tuned and error rates abd “verbal mediation” (talking to yourself as you learn) decline as the associations in memory for a skill are formed and strengthened. Autonomous stage – behavior is highly accurate, rapidly executed, and relatively automatic. Knowledge may be expressed without awareness of the underlying memories that make it possible. Skill learning generalizes to new instances or exemplars that were not encountered during learning (unlike priming). It emerges primarily from the basal ganglia. Some skills place additional demands on the cerebellum and cortical regions. Basal ganglia dysfunction as seen in Parkinson’s and Huntington’s patients impairs skill learning. Neuroimaging shows activation in the caudate and putamen portions of the basal ganglia as a skill is acquired. Habit memory Stimulus-response habits are habits that emerge through the slow accumulation of knowledge about the predictive relation between a stimulus and a response. Tested with probabilistic classification task, in which participants learn to predict 1 of 2 possible outcomes from a set of cues, each cue bearing a probabikistic relation to the outcome. Basal ganglia is also involved in habit learning. Brain damaged patients with damage to basal ganglia are severely impaired in probabilistic classification task, and healthy participants show increased activation in basal ganglia and decreased activation in the medial temporal lobes across the course of habit learning. Conditioned association Pavlov’s classical conditioning depends on the degree to which the presence of one stimulus predicts the occurrence of the other. Effective learning occurs when one stimulus reliably and predictably signals the occurrence of the second stimuli. The cerebellum is thought to be the site at which perceptual inputs are associated. Can be tested with conditioned eye blink responses where the repeated pairing of a tone and a following puff of air to the eye, which causes one to blink (UCR) causes one to associate the tone (CS) with the puff of air and the act of blinking your eye (CR) for healthy people and amnesiac patients. Cerebellar lesions disrupt conditioned eye blink responses. |
|
|
|
Baddeley-Hitch model
|
Short term storage is to enable complex cognitive activities requiring integration, coordination, and manipulation of multiple its of mentally represented info (BH), not primarily a way station for info to reside en route to long term memory (AS)
In BH there is an integral relationship between a control system (central executive) that governs the deposition and removal of info from short term storage and storage buffers themselves BH proposes at least 2 distinct short term memory buffers – one for verbal info (phonological loop) and 1 for visuospatial info (visuospatial scratchpad) |
|
|
|
Chunks/chunking
|
Single items can be grouped into higher levels of organization called chunks
|
|
|
|
Phonological similarity
|
When items simultaneously stored in working memory have to be serially recalled, performance is significantly worse when items to be maintained are phonologically similar (sound the same)
Possibly caused by confusions that arise when similar sound based codes are activated for the different items in the phonological loop |
|
|
|
Central executive
|
This part of the model
· Determines when info is deposited into the storage buffers · Determines which buffer is selected for storage (phonological loop for verbal info vs visuospatial sketchpad for visual info) · Integrates and coordinates info between the 2 buffers · Provides a mechanism by which information held in the buffers can be transformed, inspected and otherwise cognitively manipulated § Central executive controls and allocates attention, determines both how to expend cognitive resources and how to suppress irrelevant info that would consume those resources § Dual task coordination – the process of simultaneously performing 2 distinct tasks, each of which typically involves storage of information in working memory § Coordination of storage demands requires the engagement of the central executive · Alzheimer patients performed worse in the dual task condition than healthy subjects, suggesting that at least some of the cognitive impairment in Alzheimer’s is due to a dysfunctional central executive § Neuroimaging studies suggests that executive functions can be distinguisged from short term storage · Maintenance vs manipulation o Manipulation (dorsal areas of prefrontal cortex) causes more brain activation than maintenance (ventral regions of prefrontal cortex) |
|
|
|
Phonological loop
|
Verbal working memory involves a “mind’s ear” and a “mind’s voice”
o 2 subcomponents § phonological store § articulatory rehearsal process · active process to refresh information to prevent complete decay of auditory phonological code · articulatory rehearsal o voice internally the sound you heard internally o similar to shadowing - repeating quickly something we hear even without understanding; suggests involvement of phonological loop in language learning o Steps § When visually presented verbal info is encoded, info is transformed into sound based or auditory-phonological code · Code is like an internal echo box – a repository for sounds that reverberate briefly through before fading away · Only for visually presented material; auditory info has automatic initial access to phonological store § It is refreshed through articulatory rehearsal § Once the verbal information is spoken internally in rehearsal by the mind’s voice, the mind’s ear can hear it and maintain it in the phonological store o Verbal working memory capabilities should depend on the level of difficulty of phonological processing (translate verbal info into sound code) and articulatory processing (turn verbal info into speech based code) o Because working memory is flexible, performance on working memory tasks will not be catastrophically disrupted if for some reason the phonological loop is unuseable; in that case the central executive and visuospatial scratchpad kick in o 2 primary components of working verbal memory (phonological storage and articulatory rehearsal) are subserved by functionally independent systems and should be dissociable o Phonological similarity effect § When items simultaneously stored in working memory have to be serially recalled, performance is significantly worse when items to be maintained are phonologically similar (sound the same) § Possibly caused by confusions that arise when similar sound based codes are activated for the different items in the phonological loop o Word length effect § Performance on a task is worse when the items to be maintained are long words § Key factor is time it takes to pronounce all syllables; number of syllables § Assumes that pronunciation time affects speed of silent rehearsal, which requires speech based processing. The longer it takes to rehearse the more likely it is for it to be dropped by phonological store o Articulatory suppression § The condition of phonological processing and information rehearsal interference due to some task · E.g. producing overt and irrelevant speech while maintaining visually presented words in working memory § Decreases performance significantly § No phonological similarity effect or word length effect because it takes out the phonological loop o Visuospatial scratchpad is unhelpful for auditorily presented info o Studies of brain damaged patients indicate that the phonological store relies on the left inferior parietal complex o Storage and rehearsal are fundamentally independent processes § Magnitude of phonological similarity effect was not affected by word length and the word length effect was not affected by magnitude of word length effect § Articulatory suppression impaired memory performance for English letters but not for letters of an unknown foreign language · Activation was observed in brain structures associated with the motor component of speech, which was thought to represent internal speech or sub vocal rehearsal § Rhyme tasks engage rehearsal but not storage § Behavioral and neuroimaging evidence converge to establish the dissociability of the storage and rehearsal components of verbal working memory § What is the true function of cognition in the phonological loop? · Important for learning a new language but not for comprehension of a familiar language, and can be employed in a wide range of verbal memory tasks o Developmental data: level of child’s ability to repeat non words strongly predicts size of vocab a year later o P.V. (brain damaged patient) can’t learn Russian (can’t store phonologically unfamiliar words in short term memory) but can learn novel associations between familiar words (learning ability intact) |
|
|
|
Visuo-spatial scratchpad
|
The ability to develop, inspect, and navigate through a mental image is a cardinal function of visuospatial working memory
§ Since classifications decision depended on visuospatial representation, requiring the pointing (a visuospatial response) interfered with performance. § Mental navigation is an inherently spatial process § Visuospatial working memory depends on brain systems that plan movement of the eyes (or other body parts) · This movement planning system may be the basis for spatial rehearsal, the process of mentally refreshing stored locations to make them highly accessible § Spatial rehearsal may involve covert shifts of attention to memorized spatial locations § Predictions · Paying attention to a spatial location will enhance perceptual processing at that location · Maintaining a location in working memory facilitates the orienting of attention to that location · Rehearsal in spatial working memory and spatial selective attention draw on at least some of the same processes · They both rely on the same right hemisphere frontal and parietal brain regions · Spatial working memory is accomplished by enhancing processing in brain regions that support visual perceptual processing of those locations o Maintaining a spatial location in working memory produces enhanced brain activity in visual cortex regions of the opposite hemisphere, which is expected because of the contralateral organization of these brain regions § 2 sorts of info processed by visuospatial scratchpad · spatial · visual § Different types of codes are required to maintain the 2 different types of info on the visuospatial scratchpad § Visuospatial working memory may be composed of 2 distinct systems – 1 for maintaining visual object representations and the other for spatial ones § Distinction between object and spatial processing is supported by distinct neural pathways involved in processing spatial and object visual features – dorsal where and ventral what pathways · Neuroimaging studies and brain damaged patients also support this |
|
|
|
Word length effect
|
Performance on a task is worse when the items to be maintained are long words
Key factor is time it takes to pronounce all syllables; number of syllables Assumes that pronunciation time affects speed of silent rehearsal, which requires speech based processing. The longer it takes to rehearse the more likely it is for it to be dropped by phonological store |
|
|
|
Speed of pronunciation effect
|
Related to word length effect
Pronunciation time affects speed of silent rehearsal, which requires speech based processing. The longer it takes to rehearse the more likely it is for it to be dropped by phonological store |
|
|
|
Articulatory suppression
|
The condition of phonological processing and information rehearsal interference due to some task
E.g. producing overt and irrelevant speech while maintaining visually presented words in working memory Decreases performance significantly No phonological similarity effect or word length effect because it takes out the phonological loop |
|
|
|
Brooks study as evidence for two separate storage systems (phonological and visuo-spatial)
|
Experimental evidence favors the distinction between verbal and visuospatial working memory
When the interference with a verbal task is verbal, performance was more impaired than when the interference was spatial, implying that competition between 2 verbal (or 2 spatial) tasks produced more impaired performance which is evidence for separate resources or stores for each info type |
|
|
|
Dual task coordination
|
Dual task coordination – the process of simultaneously performing 2 distinct tasks, each of which typically involves storage of information in working memory
Coordination of storage demands requires the engagement of the central executive. Alzheimer patients performed worse in the dual task condition than healthy subjects, suggesting that at least some of the cognitive impairment in Alzheimer’s is due to a dysfunctional central executive |
|
|
|
Frontal lobe syndrome
|
Distractibility, difficulty concentrating
Problems with organization, planning Perseveration: fail to stop inappropriate behavior |
|
|
|
Phineas Gage Case Study
|
Gage was a railroad construction foreman
An 1848 explosion forced a 3 1/2 foot long steel tamping rod through his head. Following frontal lobe damage, behavior changed from responsible, trustworthy, hardworking, and calm demeanor to irresponsible impulsive, and given to fits of temper and profanity. |
|
|
|
Dr. P Case Study
|
Successful middle aged surgeon who liked to travel and play sports.
Suffered brain damage from surgery complications. After trauma, IQ was still relatively high. But could not handle many simple day to day activities. Difficulty adapting to changes. Difficulty planning. Could only carry out the most basic routines. Couldn’t work as a surgeon, worked as delivery driver for brother’s business, could handle only one delivery at a time. |
|
|
|
Frontal Executive Hypothesis
|
Every executive process is primarily mediated by the PFC (prefrontal cortex)
Overstated |
|
|
|
Stroop task
|
The Stroop Task is a psychological test of our mental (attentional) vitality and flexibility. The task takes advantage of our ability to read words more quickly and automatically than we can name colors. Incompatible condition is when a word is printed or displayed in a color different from the color it actually names. Conversely in the compatible position, a word is printed or displayed in the same color that the word names.
Executive attention is needed on the Stroop task because it involves selective attention to ink color, and inhibition of color name. We expect PFC patients to perform worse on incompatible tests (ink color and color name are different) than normal participants because they have difficulty in selectively attending and inhibiting, which is true. PFC patients are unable to use the Attention controller (in dorsolateral PFC) to strengthen excitatory connections based on goals. Therefore, regardless of conflict, well learned and hence strong connections will override weaker connections. This results in the word node – response node connection often winning over color node –response node connection. Word node – response node is stronger because it is well learned. Normal participants have healthy PFCs, where the attention controller is able to according to goals, strengthen goal relevant connections. Hence in the stroop task, the attention controller will strengthen color node – response node connections in accordance with goal of responding on the basis of ink color, such that those connections are stronger than the well learned word node – response node connections. In the neural network model of the Stroop task, what 2 components come into play on incompatible trials? The 2 components are the Attention controller (described above), and the conflict monitor (in anterior cingulate) which monitors the amount of conflict between nodes at the response level , and engages the attention controller as conflict increases. In the compatible Stroop test, the conflict monitor doesn’t activate the attention controller because there is no conflict. Stimulus response compatibility is a measure of degree to which the assignment of a correct response to a stimulus is consonant with the way people would act naturally. For the stroop task, stimulus response assignments are more natural for the compatible condition than for the incompatible one. Accordingly, participants in stimulus response tasks respond faster to compatible stimuli and responses than to incompatible stimuli and responses. |
|
|
|
Wisconsin card sort task
|
The Wisconsin Card Sorting Test (WCST) is a neuropsychological test of "set-shifting", i.e. the ability to display flexibility in the face of changing schedules of reinforcement. Initially, a number of stimulus cards are presented to the participant. The shapes on the cards are different in color, quantity, and design. The person administering the test decides whether the cards are to be matched by color, design or quantity. The participant is then given a stack of additional cards and asked to match each one to one of the stimulus cards, thereby forming separate piles of cards for each. The participant is not told how to match the cards; however, he or she is told whether a particular match is right or wrong. During the course of the test the matching rules are changed and the time taken for the participant to learn the new rules, and the mistakes made during this learning process are analysed to arrive at a score.
|
|
|
|
Tower of Hanoi
|
The Tower of Hanoi is a puzzle. It consists of three rods, and a number of disks of different sizes which can slide onto any rod. The puzzle starts with the disks in a neat stack in ascending order of size on one rod, the smallest at the top, thus making a conical shape.
The objective of the puzzle is to move the entire stack to another rod, obeying the following rules: * Only one disk may be moved at a time. * Each move consists of taking the upper disk from one of the rods and sliding it onto another rod, on top of the other disks that may already be present on that rod. * No disk may be placed on top of a smaller disk. To perform the tower of Hanoi problem, requires attending to some disks while ignoring others, switching attention between each mental move, and updating working memory. The task also involves goals and subgoals. Goals are decomposed into subgoals and a big problem is broken down into smaller parts. These are all executive processes so as expected PFC patients are take many more moves than normal participants to complete the tower of Hanoi. |
|
|
|
Executive processes or functions
|
selective (or executive) attention
task switching inhibition scheduling monitoring. |
|
|
|
selective (or executive) attention
|
How do patients with frontal damage differ from normal participants on the Stroop task?
PFC patients are unable to use the Attention controller (in dorsolateral PFC) to strengthen excitatory connections based on goals. Therefore, regardless of conflict, well learned and hence strong connections will override weaker connections. This results in the word node – response node connection often winning over color node –response node connection. Word node – response node is stronger because it is well learned. Normal participants have healthy PFCs, where the attention controller is able to according to goals, strengthen goal relevant connections. Hence in the stroop task, the attention controller will strengthen color node – response node connections in accordance with goal of responding on the basis of ink color, such that those connections are stronger than the well learned word node – response node connections. In the neural network model of the Stroop task, what 2 components come into play on incompatible trials? The 2 components are the Attention controller (described above), and the conflict monitor (in anterior cingulate) which monitors the amount of conflict between nodes at the response level , and engages the attention controller as conflict increases. In the compatible Stroop test, the conflict monitor doesn’t activate the attention controller because there is no conflict. Stimulus response compatibility is a measure of degree to which the assignment of a correct response to a stimulus is consonant with the way people would act naturally. For the stroop task, stimulus response assignments are more natural for the compatible condition than for the incompatible one. Accordingly, participants in stimulus response tasks respond faster to compatible stimuli and responses than to incompatible stimuli and responses. Criticisms of monitor plus attention model Attention can be needed either at response level or earlier in processing. Milham and colleagues used incongruent eligible and incongruent ineligible trials to show that the dorsolateral PFC was active in both incongruent trials but the anterior cingulate activated only on incongruent-eligible trials which require inhibition at the response level and hence the sensing of conflict by the conflict monitor and the activation of the attention controller. These results challenge the anterior cingulate as the site of the conflict monitor, suggesting that it mediates response inhibition instead PFC patients perform poorer on reason based categorization than on similarity based categorization (Pizza vs Quarter). In pizza vs quarter, participants must consider fixed and variable categories in considering whether an object that is 3 inches in diameter is a pizza of a quarter. Quarter is fixed category (smaller than 3 inches) while pizza is in variable category (can change size). Difference between reason based and similarity based categorization is the engagement of executive attention (in pizza case executive aattention should be on size dimension of the 2 categories). |
|
|
|
task switching
|
What is involved in switching attention?
Switching attention brings more processes into play. Behavioral experiments show that there is almost always a cost to switching, regardless of the task. This cost is presumably due to engaging additional cognitive processes; most of the neuroimaging results on switching attention indicate that additional neural areas are involved during the switch. Some of these additional neural areas are the same ones involved in executive attention—dorsolateral PFC and the anterior cingulate. ( In the relevant studies on switching, these two areas were not involved when performing one task for an entire block.) Other additional switching areas are in the parietal How can a switching cost be demonstrated? In switching attention, the focus of attention is moved from one entity to another. Rogers and Monsell 1995 got participants to perform pure tasks (same task) and alternating blocks ( switch from digit to letter) tasks. They found that participants require more time to respond on alternating blocks than on pure blocks. This time difference, usually on the order of 100 to 300 milliseconds, is often referred to as the switching cost. Criticism: Often do task only on a need basis not switch from one task to another as it is called for. You don’t for eg alternate between cooking and talking on the phone. So Garavan 1998 got trial participants to update cumulative counts of the appearance of each type of arrows. Presumably, the two cumulative counts are held in a state of readiness in working memory. Once the stimulus is presented and the appropriate count updated, the participant pushes a button to bring on the next stimulus. Findings: Accuracy is typically high which indicates that the switching process works but at what cost? When the researchers alternate the right and left arrows, the order of switching cost went to 500 to 600 milliseconds. Thus switching costs are evident even in tasks that have the same unpredictable structure as natural situations. Rubenstein et al.(2001): Strong evidence that switching attention is a metaprocess—a process that coordinates other processes—that also provides an information-processing framework for understanding task switching. Participants were required to shift attention between different attributes of multidimensional stimuli. The task, based on the idea of alternation, was modeled after the Wisconsin Card Sort task. Designs on four target cards vary in the number, shape, color, and size of the pictured objects. Participants must sort each test card in a deck into one of four piles according to a particular attribute, and they always know what that attribute is. In the alternation condition, participants sort on the basis of one attribute, for instance, shape, on the first trial—that is, for the first card then on the basis of a different attribute, for instance, number, on the second trial; switch back to shape on the third trial; switch back to number on the fourth trial; and so on for an entire block of trials. In other conditions, the participants sort on the basis of the same attribute for the entire block (that is, in pure blocks) The time to sort a card in the alternation condition takes longer than the comparable time in the various pure block conditions. The researchers drew a diagram from their findings regarding information-processing during switching tasks. Important – see Fig 7-14 pg 303 Executive processes: Goal shifting – rule activation Task processes: Stimulus – stimulus identification - response selection – movement production - response Figure shows a double dissociation between the task-processing and executive-processing levels. In this double dissociation, a given variable affects one level of processing but not the other one, whereas another variable shows the opposite pattern. The fact that one variable affects task processing but not executive processing implies that there must be some mechanism in the former that is not present in the latter. Likewise vice versa. Hence, a double dissociation is strong evidence that two different mechanisms are involved. Executive level: In addition the researchers used 2 arithmetic tasks (+ and - ) to influence the executive level. They wanted to facilitate changed goals by including the operator sign (+ or -) with the problem or not. The sign freed the participants from remembering which task they should be performing and thereby speed up their goals. They found that in the pure conditions, inclusion of the operator sign had no effect; so this variable had no effect on the task performing level. But it did affect the executive-processing level. This is half the double dissociation. Task processing level: To complete the double dissociation, the researchers manipulated a factor that in theory, should have an effect on the task-processing level but not on the executive processing level. That variable was simply the discriminability of the numbers (easy or hard to see) . As expected, low discriminability increased response time in all conditions but had no effect on switching time (because it had no effect on goal switching or rule changing). This is the other half of the double dissociation. The Neural switcher hypothesis Neuroimaging. Rubeinstein used PET while participants were sorting by shape, number, shape etc The patterns of activation for the pure conditions were subtracted from those for the alternation condition, the result—known as a subtraction image showed a very substantial activation in the PFC, particularly dorsolateral PFC, as well as clear activation in the parietal cortex (Meyer et1998). The attribute-shifting task is a close cousin of the Wisconsin Card which is used to detect frontal damage—and no wonder, given that the card sort involves switching in dorsolated PFC frontal damage cant do. Questions raised 1. The neuroimaging implies that imply that regions of the parietaJ cortex are involved suggesting that the neural mechanisms mediating switching are not confined to the frontal cortex. This is evidence against hypothesis that executive processes are mediated only by the frontal cortex. 2. To what degree are neural regions dedicated to specific cognitive processes. One region associated with switching —dorsolateral PFC—is also involved in executive attention. How is it possible for one region to have two functions? Perhaps it is because PET cannot differentiate between a spatial resolution of 10 millimeters. But MRI who is better machine than PET also has same results. A possibility is that the same neural mechanism mediates executive attention and attention switching, but more of that mechanism’s resources are typically needed when switching attention than just attending. Does this means that there are neural mechanisms dedicated solely to switching? Sylvester etc 2003 found that areas distinct to switching include areas in the inferior parietal lobe and the extrastriatal visual cortex, whereas areas distinct to executive attention include two frontal areas, one in the anterior PFC and the other in the premotor cortex. What gets switched around? Switch of task, of representation and of attended attribute. There are others but question is does it matter what gets switched? Researchers looked at meta analysis – results of test pools. They put together on one schematic brain the peaks of the activations from switching studies that either attended attributes or tasks. Then they applied a clustering algorithm to see whether the peaks fell in clusters or are evenly distributed. They wanted to see whether (1) whether the peaks for switching attributes fell into a few separate clusters; (2) whether the peaks for switching tasks fell into a few clusters; and (3) the extent to which these two sets of clusters overlap. The results show substantial overlap between clusters for switching tasks and attributes. This means switching and tasks and attributes involve the same neural mechanisms and many of the relevant neural mechanisms are located in the parietal cortex which against argues against the frontal executive hypothesis. These results are relatively new. According to the model of attention switching discussed, what are the underlying causes of a switching cost? This cost is presumably due to engaging additional cognitive processes; most of the neuroimaging results on switching attention indicate that additional neural areas are involved during the switch. Some of these additional neural areas are the same ones involved in executive attention—dorsolateral PFC and the anterior cingulate. |
|
|
|
inhibition
|
What is response inhibition, and what is distinctive about it?
Response inhibition, the suppression of a partially prepared response, can be isolated experimentally by tasks such as go/no-go and stop-signal. These task show that participants are sensitive to the probability of a go response (more probable = harder to inhibit go response when X requiring no go response appears) or the time to delay a response (The longer the delay between the question and the stop signal, the more processing the participant has completed and the harder it is to inhibit the response). At the neural level, these tasks activate the dorsolateral PFC and the anterior cingulate. However, these tasks often bring an additional area on line, the orbitofrontal cortex. In addition, there is a remarkable parallel between the participant’s ability to inhibit a response and the development of the PFC. What evidence indicates that response inhibition is a distinct executive process? Response inhibition is mediated by the PFC which occurs in executive attention processing tasks. Inhibition involves some task switching which requires executive attention. Jean Piaget 1954 got infants to look for a desired object in same place and was rewarded for finding it. When the under one year old infant see the same object put in some other place they still return to the first place to look for it. This is because they cannot inhibit the previously rewarded response and select the new correct one. Some say they lack a working memory to maintain the location of the object but not so for they look at new location but reach for old location. At one they do better . By age 3 to 5 they can handle the go/no go task. At age 7 to 12 their neural patterns of activation are like the adult. 3-5 kids do well on the tapping test. Participants response once if the tapping is twice and twice if experimenter taps once. Kids 3 to 5 show improvement as tapping task activates the dorsolateral PFC and adults with lesions do poorly on this test. Why is response inhibition so important in daily life? Response inhibition is the suppression of a partially prepared response. If you said everything that come to mind or perform every action you thought you will be friendless or worse. Some psychiatric disorders appear to be marked by the lack of response inhibition. E.g bizarre speech of schizophrenia or behavior of obsessive compulsive disorders. Go/no go tests assesses frontal lobe functioning. Participants push a button for every letter other than X (no go). They found that the longer the sequence of go response the more difficult for a participant to start an inhibitory response. MRI shows anterior cingulate also has a part to play together with dorsolateral PFC. In inhibition an extra PFC area is activated – the orbitofrontal cortex which is below the dorsolateral PFC. Those with more activation in the orbitofrontal cortex made less errors on the no go tasks. Logan developed the stop signal task. The longer the delay between the task and stop signal the more likely the response tendency cannot be aborted so more likely to make mistakes. EREPs ever related potentials showed inhibitions could occur at any point in the preparation and execution of a response. |
|
|
|
scheduling
|
What mechanisms are used for sequencing information?
3 different mechanisms may be used in sequencing unrelated items, including associations between adjacent items, direct coding of temporal order, and using relative familiarity to code order. Neuroimaging studies support the idea of different mechanisms: the dorsolateral PFC may mediate direct coding of order whereas the parietal cortex may mediate the use of continuous information - such as the relative strength or familiarity—to code order. In sequencing related items, there is an important distinction between constructing familiar vs novel sequences; the operation of the PFC is sensitive to this distinction, as indicated by the fact that frontal-damage patients are far more impaired on generating a novel sequence than a familiar one. What are the different ways in which order information may be represented? 1. By direct associations. E.g. the participants are asked to remember a series of items say JGXRLB followed by 2 probe letters GL. Participants have to say whether GL is in same order in memory set. By association say J comes before G and L after G. 2. By attaching order tags to the items. Eg J is tagged as first item, G the second and so on. 3. Code on the basis of familiarity. If asked whether G comes before I participant check for familiarity of 2 items, order is decided based on how much a representation has decayed. More decay = earlier; less decay = later. In coding, order information is continuous not discrete as for 1 & 2. Above are arbitrary items. What is the evidence that the PFC is involved in processing ordered sequences? People with lesions in the PFC , particularly the dorsolateral PFC; patients with lesions in posterior cortical regions; and normal controls. They were tested on script sequences, such as starting a car, and novel sequences, such as opening a beauty parlor. In one study they got the participants to mime starting actions on starting a car. There was no difference in all 3 groups for numbers or types of actions. But when they had to put the actions into an appropriate sequence, none of the normal or patients with posterior lesions made any errors. The frontal damage ones had difficulties. Same when used unfamiliar theme like opening a beauty parlor. This shows that PFC is involved in sequencing actions as well as generating them. The posterior patients had extensive brain damage yet they did ok so it matters where the damage is. By sequencing we mean, in part, coding information about the order of events in working memory. You can’t form a plan to accomplish a goal without coding the order of actions or events involved. Coding the order of an item requires different mechanism than coding the identity of a problem. Neuroimaging shows that people use 2 different representations of order: a familiarity based representations that rely on the intraparietal cortex and a direct temporal coding representation that is mediated by the dorsolateral PFC. |
|
|
|
monitoring
|
What is involved in monitoring our performance on line?
Work on monitoring as an executive process has followed two lines of research. One focuses on monitoring the contents of working memory; the PFC is clearly involved. The other focuses on monitoring for errors, a process that has been well studied in behavioral experiments. In addition, ERP and fMRI studies suggest that such monitoring is extremely rapid, occurring within 100 milliseconds of the response, and that the process is mediated by the anterior cingulate. In tasks requiring the monitoring of working memory, what neural structure is routinely activated? Presumably the requirement to monitor the contents of working memory is mediated by the PFC, the left PFC in particular (Petrides & Milner,). This result has wide generality. It holds with verbal items as well as visual ones and essentially the same deficits can be produced in monkeys that have been lesioned in region of the dorsolateral PFC (Petrides, 1986). Furthermore, when normal human participants perform the task while having their brains imaged (by PET), one of the major regions activated is the dorsolateral PFC. Figure 7-19 6 item self ordered pointing task. The task involves the updating and monitoring of working memory. We are quickly ‘aware’ of making an error.” What is the evidence for this assertion? Error detection can be quite slow. Studies show that it takes 700milliseconds. Work using event related potentials that participants detect errors very soon after making them. Gehring found a wiggle in the ERP response wave following an error. ERN (error related negativity ) is a negative in the ERP wave detected at start of error and peaks 100 milliseconds after the error. This shows some kind of internal monitoring process. ERN generated by a midline structure in the frontal cortex presumably in the anterior cingulate. Some say possibly what is being monitored is not errors per se but conflict at the response level. Neuroimaging studies show that trials that engender substantial conflict , yet are responded to correctly, lead to increased activation in the same region of the anterior cingulate activated in ERP studies (Carter etal.,L, 1998) Findings from both ERP and MRI proof that such an executive process exists. |
|
|
|
Role of PFC (prefrontal cortex) in executive functions
shown in stroop task and wisconson card sort task. |
Executive attention is needed on the Stroop task because it involves selective attention to ink color, and inhibition of color name. We expect PFC patients to perform worse on incompatible tests (ink color and color name are different) than normal participants because they have difficulty in selectively attending and inhibiting, which is true.
PFC patients are unable to use the Attention controller (in dorsolateral PFC) to strengthen excitatory connections based on goals. Therefore, regardless of conflict, well learned and hence strong connections will override weaker connections. This results in the word node – response node connection often winning over color node –response node connection. Word node – response node is stronger because it is well learned. Normal participants have healthy PFCs, where the attention controller is able to according to goals, strengthen goal relevant connections. Hence in the stroop task, the attention controller will strengthen color node – response node connections in accordance with goal of responding on the basis of ink color, such that those connections are stronger than the well learned word node – response node connections. In the neural network model of the Stroop task, what 2 components come into play on incompatible trials? The 2 components are the Attention controller (described above), and the conflict monitor (in anterior cingulate) which monitors the amount of conflict between nodes at the response level , and engages the attention controller as conflict increases. In the compatible Stroop test, the conflict monitor doesn’t activate the attention controller because there is no conflict. Stimulus response compatibility is a measure of degree to which the assignment of a correct response to a stimulus is consonant with the way people would act naturally. For the stroop task, stimulus response assignments are more natural for the compatible condition than for the incompatible one. Accordingly, participants in stimulus response tasks respond faster to compatible stimuli and responses than to incompatible stimuli and responses. To perform the tower of Hanoi problem, requires attending to some disks while ignoring others, switching attention between each mental move, and updating working memory. The task also involves goals and subgoals. Goals are decomposed into subgoals and a big problem is broken down into smaller parts. These are all executive processes so as expected PFC patients are take many more moves than normal participants to complete the tower of Hanoi. |
|
|
|
“Switching cost” in task switching paradigm.
|
How can a switching cost be demonstrated?
In switching attention, the focus of attention is moved from one entity to another. Rogers and Monsell 1995 got participants to perform pure tasks (same task) and alternating blocks ( switch from digit to letter) tasks. They found that participants require more time to respond on alternating blocks than on pure blocks. This time difference, usually on the order of 100 to 300 milliseconds, is often referred to as the switching cost. Criticism: Often do task only on a need basis not switch from one task to another as it is called for. You don’t for eg alternate between cooking and talking on the phone. So Garavan 1998 got trial participants to update cumulative counts of the appearance of each type of arrows. Presumably, the two cumulative counts are held in a state of readiness in working memory. Once the stimulus is presented and the appropriate count updated, the participant pushes a button to bring on the next stimulus. Findings: Accuracy is typically high which indicates that the switching process works but at what cost? When the researchers alternate the right and left arrows, the order of switching cost went to 500 to 600 milliseconds. Thus switching costs are evident even in tasks that have the same unpredictable structure as natural situations. Rubenstein et al.(2001): Strong evidence that switching attention is a metaprocess—a process that coordinates other processes—that also provides an information-processing framework for understanding task switching. Participants were required to shift attention between different attributes of multidimensional stimuli. The task, based on the idea of alternation, was modeled after the Wisconsin Card Sort task. Designs on four target cards vary in the number, shape, color, and size of the pictured objects. Participants must sort each test card in a deck into one of four piles according to a particular attribute, and they always know what that attribute is. In the alternation condition, participants sort on the basis of one attribute, for instance, shape, on the first trial—that is, for the first card then on the basis of a different attribute, for instance, number, on the second trial; switch back to shape on the third trial; switch back to number on the fourth trial; and so on for an entire block of trials. In other conditions, the participants sort on the basis of the same attribute for the entire block (that is, in pure blocks) The time to sort a card in the alternation condition takes longer than the comparable time in the various pure block conditions. The researchers drew a diagram from their findings regarding information-processing during switching tasks. Important – see Fig 7-14 pg 303 |
|
|
|
Inhibition of response (Go/No Go task)
|
Response inhibition, the suppression of a partially prepared response, can be isolated experimentally by tasks such as go/no-go and stop-signal. These task show that participants are sensitive to the probability of a go response (more probable = harder to inhibit go response when X requiring no go response appears) or the time to delay a response (The longer the delay between the question and the stop signal, the more processing the participant has completed and the harder it is to inhibit the response). At the neural level, these tasks activate the dorsolateral PFC and the anterior cingulate. However, these tasks often bring an additional area on line, the orbitofrontal cortex. In addition, there is a remarkable parallel between the participant’s ability to inhibit a response and the development of the PFC.
|
|
|
|
Sequencing Tasks (keep track of order information)
|
What mechanisms are used for sequencing information?
3 different mechanisms may be used in sequencing unrelated items, including associations between adjacent items, direct coding of temporal order, and using relative familiarity to code order. Neuroimaging studies support the idea of different mechanisms: the dorsolateral PFC may mediate direct coding of order whereas the parietal cortex may mediate the use of continuous information - such as the relative strength or familiarity—to code order. In sequencing related items, there is an important distinction between constructing familiar vs novel sequences; the operation of the PFC is sensitive to this distinction, as indicated by the fact that frontal-damage patients are far more impaired on generating a novel sequence than a familiar one. What are the different ways in which order information may be represented? 1. By direct associations. E.g. the participants are asked to remember a series of items say JGXRLB followed by 2 probe letters GL. Participants have to say whether GL is in same order in memory set. By association say J comes before G and L after G. 2. By attaching order tags to the items. Eg J is tagged as first item, G the second and so on. 3. Code on the basis of familiarity. If asked whether G comes before I participant check for familiarity of 2 items, order is decided based on how much a representation has decayed. More decay = earlier; less decay = later. In coding, order information is continuous not discrete as for 1 & 2. Above are arbitrary items. |
|
|
|
Sensory synesthesia, “to perceive together.”
|
Strong synesthesia is characterized by a vivid image in one sensory modality in response to stimulation in another one.
Weak synesthesia is characterized by cross-sensory correspondences expressed through language, perceptual similarity, and perceptual interactions during information processing. |
|
|
|
Sensory leakage hypothesis.
|
The sensory leakage hypothesis claims that information leaks from one sensory channel into another, producing strong synesthesia.
|
|
|
|
Inducer stimulus (pain, for Carol) and induced percept (color). Both idiosyncratic (associations differ for each synesthete) and systematic (e.g., higher pitch sounds=brighter light sensations).
Induced images are often visual, while the inducers are often auditory, tactile, or gustatory (taste) stimuli. |
An association or correspondence exists between an inducer in one modality (e.g., pain in Carol’s case) and an induced percept or image in another (e.g., color). These correspondences have several salient characteristics.
Simultaneously idiosyncratic and systematic. Correspondences are idiosyncratic in that each synesthete has a unique scheme of associations. Middle C on the piano may be blue to one color-music synesthete and green to another. Yet both synesthetes will reveal a systematic relationship between color brightness or lightness and auditory pitch: The higher the pitch of the sound, the lighter or brighter the color of the image. There are other systematic associations from other forms of synesthesia including auditory-visual synesthesia where the association is of pitch to shape and size: The higher the pitch of the sound, the sharper, more angular, and smaller the visual image. Synesthetic images are typically simple (e.g., consist of a color or shape), but dynamic (e.g., as the inducer waxes and wanes, so does the image). Induced images tend to be visual, whereas inducing stimuli tend to be auditory, tactile, or gustatory. Don’t know why. Because strong synesthesias are noticed in early childhood, it is possible they are inborn. Higher incidence of synesthesia in females speaks against synesthesia being learned. The connection between the inducer and induced is so entrenched that the image is considered part of the percept’s literal identity. Synesthetic perception is highly memorable and genuine, but it is not clear whether to attribute the synesthetes’ superior performance to their synesthesia or perhaps to better memory for word-color pairings in general. With regard to production, the relation between the inducer and the induced is typically unidirectional. That is, although a voice induces a yellow image, a yellow percept need not induce an image of a voice. |
|
|
|
In weak synesthesia, the associations are context dependent, not absolute.
|
Laboratory experiments square with the idea that most people can appreciate cross-modal associations. In such studies, participants are asked to pair a stimulus from one sensory modality to a stimulus from another. These studies show that pairings are systematic. For example, given a set of notes varying in pitch and a set of colors varying in lightness, the higher the pitch, the lighter the color paired with it. In weak synesthesia, the correspondences are systematic and contextual, so that the highest pitch is always associated with the lightest color but in strong synesthesia, the correspondences are systematic and absolute.
Cross-modal correspondences both inborn and learned. Infants who have not yet learned language show a kind of cross-modal “matching” of loudness-brightness and pitch-position. Other correspondences develop over time. For example, 4-year-old children can match pitch and brightness systematically, but not pitch and visual size. By age 12, children perform these matches as well as do adults. |
|
|
|
Encoding
|
Processes used to store information in memory
|
|
|
|
Storage
|
Processes used to maintain information in memory
|
|
|
|
Retrieval
|
Processes used to get information back out of memory
|
|
|
|
Recall vs. recognition tasks
|
Recall Task: You have to generate an answer.
In TOT example, I gave you definitions of obscure words, you had to recall the word. Recognition Task: don’t need to generate the answer. Shepard’s study of visual memory, viewed 612 pictures, then shown two pictures and asked to indicate which one they had seem previously. |
|
|
|
Free recall
|
Recall all the words you can from the list you saw previously.
|
|
|
|
Serial recall
|
Recall the names of all previous presidents in the order they were elected
Need to recall order as well as item names |
|
|
|
Cued recall
|
Give participants some clue to trigger recall.
Paired associates -locomotive, switch-paper, etc. dishtowel |
|
|
|
Explicit vs. implicit memory tasks
|
Explicit memory tasks
Involves conscious recollection Participant knows they are trying to retrieve information from their memory Implicit memory tasks Require participants to complete a task The completion of the task indirectly indicates memory. |
|
|
|
Procedural memory (how) vs. declarative memory (what)
|
Procedural: Knowing how to do something
Ride a bike Skateboard Skiing Declarative memory: Memory for facts (semantic) or events (autobiographical) |
|
|
|
Atkinson & Shiffrin’s 3 Stage Model of Memory
|
|
|
|
|
Sensory memory (iconic, echoic memory)
|
Get information in
Very limited duration Keep only what is processed Visual=iconic Auditory=echoic |
Averbach & Coriell (1961) Iconic Memory Research
|
|
|
Short term memory (STM)
|
limited capacity (7 +/-2) items
take in from sensory memory and long-term memory persists as long as it is rehearsed |
Duration of Short-Term Memory (Peterson & Peterson, 1959)
|
|
|
Long term store (or LTM)
|
Fed by short-term memory
Virtually unlimited capacity Virtually unlimited duration Getting into LTM takes effort |
|
|
|
Sperling partial report procedure
|
Whole report procedure
Flash a matrix of letters for 50 milliseconds Identify as many letters as possible Participants typically remembered 4 letters Partial Report Procedure Flash a matrix of letters for 50 milliseconds Participants are told to report one row at a time (e.g., bottom row). Participants were able to report any row requested Sperling showed that we could perceive a lot more with the sensory buffer than it seemed. |
|
|
|
Chunking
|
Strategy for holding more in your memory
|
|
|
|
Expertise and chunking (chess study)
|
You have 5 seconds to memorize as much as you can
Then, draw an empty chess board and reproduce the arrangement of pieces |
|
|
|
Vogel et al. study on visual WM
|
Vogel, Woodman & Luck (2001)
Used colors and orientations (lines). For example, show display with six colored squares Then see same or new display with one changed. People can remember about 3-4 visual objects. Store integrated objects, not just features. |
|
|
|
Permastore or Very long term memory (Bahrick’s research)
|
High school year books containing all of the names and photos of the students were used to assess memory.
392 ex-high school students (17-74) took 4 different memory tests: Free recall of the names A photo recognition test where they were asked to identify former classmates A name recognition test A name and photo matching test For some of the participants, it was as long as 48 years since they graduated from High school |
|
|
|
Craik & Lockhart’s Levels of Processing Model (LOP)
Levels of processing effect (deep vs. shallow processing) Self reference effect. |
Different ways to process information lead to different strengths of memories
Deep processing leads to better memory Elaborating according to meaning leads to a strong memory Shallow processing emphasizes the physical features of the stimulus the memory trace is fragile and quickly decays Distinguished between maintenance rehearsal and elaborative rehearsal |
|
|
|
Craik & Tulving (1975)
Support for Levels of Processing Effect |
Craik & Tulving (1975)
Participants studied a list of words in 3 different ways Structural or Physical: Is the word in capital letters? Phonemic: Does the word rhyme with dog? Semantic: Does the word fit in this sentence? The ______ is delicious. Or is it a type of plant? A recognition test was given to see which type of processing led to the best memory |
|
|
|
Self reference effect.
|
Subjects have to determine if words in a list describe them.
Smart, talkative, diligent, shy… High levels of recall, even if you thought shy, diligent, etc., did not describe you. Even higher recall for words that did describe you. |
|
|
|
Criticisms of LOP theory, Morris study: transfer appropriate processing.
|
Circular definition of levels
Transfer appropriate processing effect Morris, Bransford, and Franks (1977) Two processing tasks: semantic vs. rhyme Two types of tests: standard yes/no recognition vs. rhyme test Memory performance also depends on the match between encoding processes and type of test |
|
|
|
Baddeley’s Working Memory Model
|
Working Memory: refers to the system or systems involved in the temporary storage of information in the performance of cognitive skills such as reasoning, learning and comprehension
|
|
|
|
Phonological storage capacity: chunks.
|
Evidence for Articulatory/Phonological Loop
Used to maintain information for a short time and for acoustic rehearsal. Word length effect Effect of articulatory suppression Speed of speech effect Acoustic similarity effect |
|
|
|
Visuospatial sketchpad: Brooks letter-scanning task or sentence task coupled with pointing responses or vocal responses. Pattern of interference.
|
The component of WM devoted to visual imagery and spatial processing
Information can enter the buffer either directly from visual perception from long-term memory Information can then be treated like a percept: scanned, rotated, enlarged, etc. |
|
|
|
Word length effect
Effect of articulatory suppression |
Subjects can generally remember about as many words as they can say in 2 seconds.
Memory span for short words “sum, wit, harm” is better (larger) than for long words “opportunity, individual, university” Effect disappears with articulatory suppression - repeatedly say “the” when viewing the list (e.g.“the the the the the the the the”) which prevents formation of phonological code |
|
|
|
Speed of speech
|
Memory span is better:
For words that are pronounced quickly (bishop vs. harpoon ) For people who speak quickly In languages where words are pronounced quickly (digits span) (Chinese better than English better than Welsh) |
|
|
|
Acoustic similarity effect
|
Span test
|
|
|
|
Central executive in Baddeley’s model
|
Supervise attention -
Focuses attention on relevant items and inhibiting irrelevant ones (e.g., Stroop) Planning/Coordination - Plans sequence of tasks to accomplish goals, schedules processes in complex tasks, often switches attention between different parts Monitoring of mental activity - Updates and checks content to determine next step in sequence of parts |
|
|
|
Mental workspace.
|
Purpose of working memory
Whenever you need to retain some information while processing other information |
|
|
|
Frontal cortex: prefrontal cortex; dorsolateral prefrontal cortex, orbitofrontal cortex, anterior cingulate.
|
|
|
|
|
Delayed response paradigm: How do we actively maintain information?
|
Cue is presented.
Delay (must remember cue during this delay). Response. Task used with monkeys See cue in one of 8 locations in a circle around a fixation point (a plus sign). Needs to remember that cue during delay of 2–30 seconds. Then sees “go” signal – monkey looks at location of the cue. |
|
|
|
Memory “blindspots” in delayed response paradigm.
|
Creating Memory Blindspots:
Train animal to perform the delayed response task. Method 1: Then lesion brain areas responsible for active maintenance. Animal loses ability to remember cue over the delay. Method 2: Created transient deficits in delayed response task by cooling neurons so they can’t function. Restore ability by letting neurons return to normal temperatures. |
|
|
|
N-back task.
|
Subject sees a stream of letters presented one at a time.
D F F B C F B B Task is to decide if current letter matches the letter N back. If N =1 D F F B C F B B No No Yes No No No No Yes If N=2 D F F B F B C B No No No No Yes Yes No Yes |
|
|
|
Reverberatory loops in the brain, they keep information active.
|
Active maintenance seems to involve sustained firing by certain neurons in prefrontal cortex after the stimulus is taken away.
|
|
|
|
Selective (or executive) attention
|
Acts on the contents of working memory and directs subsequent processing to achieve some goal (enter tomorrow’s schedule into WM and focus on that schedule so you can figure out if and when you can do your friend the favor).
|
|
|
|
Task switching
|
Switches attention from one task to another (from the conversation back to watching the stove, then back again).
|
|
|
|
Inhibition
|
Ignore or inhibit processing of irrelevant or distracting information (the music).
|
|
|
|
Scheduling
|
Planning timing/ordering of a sequence of activities (can you postpone coffee date in order to...)
|
|
|
|
Monitoring.
|
Are you achieving your goals?
|
|
|
|
Stimulus-response compatibility tasks: role of executive attention.
|
Stimulus-Response Compatibility: The degree to which the “correct” response to a task is consistent with what people would do naturally.
Stimuli appears on left or ride side of screen: Compatible Response: Press key with left hand for stimuli on left side/press key with right hand for stimuli on right side. Incompatible: reverse hands, left hand response to right sided stimuli, etc. Compatible responses are faster and more accurate than incompatible responses. |
|
|
|
“Switching cost” in task switching paradigm.
|
Costs of Switching Attention: Juggling multiple tasks is always less efficient than doing each one separately one at a time.
Studied in task switching paradigm Make one of two simple judgments Task 1: Is number above/below 5? 2, 3, 7, 8, 4, 9, etc. Task 2: Is a letter a vowel or consonant? A, B, N, E, G, etc. Pure blocks vs. Alternating blocks. RT is longer for alternating blocks. Referred to as “switching cost”, extra time needed for alternating blocks. |
|
|
|
Inhibition of response (Go/No Go task)
|
Response inhibition is the suppression of a partially prepared response.
Go/No Go task: Press a button (as quickly as possible) whenever you see a letter, unless it is an “X”. More “go” trials in a row, the harder it is to inhibit the response when an “X” is shown.FMRI (brain imaging) studies seem to show that brain regions in Go and No go trails seem to be different. Dorsolateral PFC active in Go trials, but anterior cingulate is active on No Go trials. And orbitofrontal PFC. Lesions to orbitofrontal PFC in animals impair response inhibition. Same with brain damage in people. |
|
|
|
Sequencing Tasks (keep track of order information)
|
Coding information about the order of events.
If you need to carry out a sequence of actions to accomplish a goal you need to code the order of events or actions. Coding for item identity vs. coding for item order. Sternberg study. Patients with frontal damage are impaired in coding order information. |
|
|
|
“Random” number generation.
|
Task: Generate a string of N random numbers.
People try not to repeat numbers in close succession (so, no 2,7,4,4, or 2,7,4,8,4) and to have no discernible patterns in the sequence (no 3,4,5,6 or no 2,4,6,8, etc). They have to monitor the string of numbers to do this: Uses PFC (compared to saying numbers in order). |
|
|
|
Self-ordering task
|
Self-ordering task (Petrides et al., 1993).
See a page with six different pictures on it. Point at one of the pictures. Next page, same pictures arranged differently. Point to a different picture, and so on for each page. You have to store each selection (e.g., tea pot) in WM and monitor that information before making next selection. Then you update the information in WM by adding next selection to it. Self-ordering task is easy for normal subjects as long as the number of items is relatively small (7 or less). Frontal patients have hard time with this task. PFC is activated when doing task (for normals). Harder for both pictures and for verbal information. Lesions in PFC affect monkeys ability to do this type of task. |
|
|
|
ERP and monitoring: the “error” signal.
|
ERP evidence (Gehring et al. 1993).
Event Related Potentials: Using EEG to detect some event related activity in the brain. He found evidence for an error related negativity. If someone makes an error on some task, there is a negative ERP wave that starts at around the time of the error and peaks within 100msec of the error. |
|
|
|
Monitoring
|
Monitoring: Assessing one’s performance in real time
as you are performing tasks. |
|
|
|
Standing’s studies of visual memory
|
Subjects in Shepard’s and Standing’s studies of visual memory recognized from 600 to 10,000 pictures.
|
|
|
|
Bahrick’s “permastore” or very long term memory idea
|
Bahrick’s research on very long term memory.
Permastore – subjects remembered high school Spanish or the names and faces of their high school classmates even 30, 40 or 50 years later. |
|
|
|
Encoding specificity effect
|
Our ability to remember a stimulus depends on the similarity between the way a stimulus is processed at encoding and the way it is processed when tested (Tulving & Thompson, 1971).
|
|
|
|
Baddeley forgetfulness survey.
|
Baddeley, 1990, Forgetfulness survey
Asked people questions like how often do you: “Forget important details about yourself, such as your birthday or where you live?” Respond on scale of 1 to 9 1 is “never in the last 6 months” 9 is “more than once a day” Average answer was 1 (never in the last 6 months). |
|
|
|
Ebbinghaus’ forgetting studies/curve.
|
Studied the learning and retention of nonsense syllables.
He memorized lots of nonsense syllables until he them all with 100% accuracy. Koj, Bov, Wug, etc. Meaningless stimuli – according to LOP theory can’t process them deeply. 2,300 of these words, put them all on cards in a box, he would pull out a random set for a study. |
|
|
|
Why do we forget?
|
Failing to encode the information in the first place.
Depth of processing: shallow processing will not lead to stable memories. |
|
|
|
Improving retention
|
Spacing Effect (Massed vs. distributed learning)
Organization of information State/context dependent memory Encoding specificity effect: our ability to remember a stimulus depends on the similarity between the way a stimulus is processed at encoding and the way it is processed when tested (Tulving & Thompson, 1971). |
|
|
|
Spacing Effect (Massed vs. distributed learning)
|
Also discovered by Ebbinghaus.
Massed vs. Distributed Practice If you study for 3 hours one evening (massed). vs. Or study same material 1 hour each time for three evenings (distributed). Distributed works better |
|
|
|
Organization of information
|
Bower, Clark, Lesgold, and Winzenz (1969)
Participants remembered 65% of the organized list, but only 19% of the random list Thus, organization helps memory retrieval Impose structure on material, recall it better. |
|
|
|
State dependent memory
|
Godden & Baddeley (1975)
Learn words underwater while diving or while on surface. Tested later underwater or on surface. Do better if test context matches learning context. (40% more words). Listening to cafeteria noise on headphones (Grant et al., 1998). Recall is improved if internal physiological or emotional state is the same during testing and initial encoding State-dependent - internal, physiological factors |
|
|
|
Context dependent memory
|
After learning to move a mobile by kicking, infants had their learning reactivated most strongly when retested in the same rather than a different context (Butler & Rovee-Collier, 1989).
Context-dependent - external, environmental factors |
|
|
|
Proactive interference
|
Old memories interfere with recall of new information
The experimental group remembers less material from the tested list B than the control group Information previously learned (list A) interferes with retrieval of List B |
|
|
|
Savings in relearning (easier to relearn “forgotten” material then it was to acquire it the first time).
|
Nelson (1971) Critical Manipulation
|
|
|
|
Decay vs. interference in forgetting
|
Decay theory
Memory is weakened with disuse Simply passage of time… Interference theory Proactive: old memories interfere with recall of new information Retroactive: new memories interfere with recall of old information |
|
|
|
Retroactive interference
|
New memories interfere with recall of old information
The experimental group will remember less material from the tested list A compared to the control group Information learned afterwards interferes with retrieval of List A. |
|
|
|
Proactive interference
|
Old memories interfere with recall of new information
The experimental group remembers less material from the tested list B than the control group Information previously learned (list A) interferes with retrieval of List B |
|
|
|
Nelson’s study of paired associates.
|
Evidence Supporting “Still There” Theory Nelson (1971)
|
|
|
|
Savings in relearning (easier to relearn “forgotten” material then it was to acquire it the first time).
|
Nelson (1971) Critical Manipulation
|
|
|
|
Anterograde vs. retrograde.
|
Anterograde amnesia
Inability to learn new explicit information after trauma Patient H.M. The guy in Memento Retrograde amnesia Inability to retrieve explicit information prior to trauma Temporally-graded Memory for old information typically intact More recent information more vulnerable |
|
|
|
Patient H.M. Case Study
|
HM had surgery in 1953 to relieve epilepsy.
surgeons removed medial temporal lobe on both sides. Unable to learn most new information: severe anterograde amnesia. But he could recall facts from before surgery, preserved language skills, recognized people (although not as well as before). Couldn't recognize new people, remember salient events (e.g., mother's surgery). Normal working memory. Can still carry on a conversation, but won’t remember it if you leave the room and come back (i.e., impaired long term memory) Can still learn some things. Poor on explicit tests of memory. But can show learning in implicit memory tests. Classically conditoned: Blink (CR) to tone (CS), paired with puff of air (UCS) into eye. Studies with animals show that the learning is being done by the cerebellum. Can form preferences for new music, but does not recognize the melodies. |
|
|
|
Hippocampus,/Medial temporal lobe.
|
Medial temporal lobe is critically important for new explicit long term memory formation
Memory are not stored permanently in hippocampus. H.M. remembers old events (stored in cortex). Hippocampus and related structures seem to be critical for initial encoding –they “bind” together activity in different parts of cortex. Once memories are “consolidated”, don’t need the hippocampus to retrieve them. H.M. can remember things from his past – they’ve been consolidated and stored in cortex. |
|
|
Tower of Hanoi
|
|
|
|
|
Spared implicit memory
. |
H.M. can still learn some things.
Poor on explicit tests of memory. But can show learning in implicit memory tests. Classically conditoned: Blink (CR) to tone (CS), paired with puff of air (UCS) into eye. Studies with animals show that the learning is being done by the cerebellum. Can form preferences for new music, but does not recognize the melodies. Warrington, E. K. & Weiskrantz, L. (1970). Amnesic syndrome: Consolidation or retrieval? Nature, 228, 629-630. |
|
|
Mirror reading
|
Participants
Korsokoff’s amnesics , and Patient N.A., versus normal subjects Severe Anterograde Amnesia (Can’t learn new things) Methods Experiment included 50% repeated words across 4 days Non-repeated words: implicit Repeated words: implicit + explicit Results For new words, Normals and Amnesics improved about the same (implicit only) For old words, Normals were better than amnesics (implicit + explicit). |
|
|
|
Taxonomy of Long Term Memory
|
|
|
|
|
Categorization by Pigeons
|
Categorization allows inferences about members of a class
|
|
|
|
Semantic memory
|
General knowledge, memory of facts
Not tied to specific time or place |
|
|
|
Concepts and categories
|
Concept: A mental representation of an item and associated knowledge and beliefs about that item (cat, tools, furniture).
Category: A mental grouping of persons, ideas, events, or objects that share common properties. Concepts and categories, what’s the difference? Concepts are the members of a category. A category is a collection of related concepts. Apple can be a category – there’s different kinds of apples – More or less prototypical apples. Apple is also one concept – in another category fruit. Same thing could be a concept or a category. Natural Concepts/categories Occur naturally (e.g. plants, trees, cats) Share physical traits, appearances, behaviors, history, etc. Artifact Concept/categories Created by humans (e.g., hammers, computers) Function is key. |
|
|
|
Defining Features (Classical View)
|
Must defining feature(s) this to be considered a member
Concept is defined by a set of necessary of jointly sufficient features. Bachelor: human, male, unmarried, adult. Problems: 1. Difficult to specify necessary features of some concepts What is the defining feature of a monster? 2. Some things do not have necessary and sufficient conditions. Games – What makes something a game? Basketball vs chess 3. Exceptions Is monk a bachelor? Is a game show “reality TV”? 4. Typicality Effects Some things are better examples of a concept than others Robin is a more typical bird than a ostrich Zebra painted all black – still a zebra. Robin with clipped wings – still a robin. |
|
|
|
Prototypes
|
Abstracted representation of a category containing salient features that are true of most instances
Characteristic features which describe what members of that concept are like Monster prototype has these characteristics: Scary, pale, has sharp teeth, is evil, lives in odd place (coffins, closets, or graveyards) Vampires, Zombies, and Bogeymen all fit that prototype well, Can a green, grumpy, lives in a garbage can monster also fit? Yes, but less well. A “typical” member of a category, one that has most of the defining features of a category. In some cases, can think of the prototype as the average. Deals well with fuzzy concepts Fuzzy concepts are categories that cannot be easily defined (Monster, Games) To categorize something, simply compare to prototype Evidence: Typicality ratings Picture Verification: See a picture of something. Is it a dog? Y/N? Faster if it is prototypical dog. Draw a bird. People draw robin, not ostrich. Verify Statements: True or False? A robin is a bird. A chicken is a bird. An ostrich is a bird. Faster if more typical Production: List all the birds you can think of. Robin, bluejay first, list non-protoypical birds later in list. Making Inferences If told a new fact about a prototypical bird (robins), like “robins like to drink milk.” Subjects are willing to extend that to all birds. Cardinals like milk, too. But people will not do the same for a non-prototypical member (chickens or ducks) of category. General process of forming concepts and categories with prototypical members is probably innate but the details of the concepts and prototypes we develop are based on experience. My Japanese GSI said Robin was not a prototypical bird for her. Prototypical house is different in different cultures. Expertise can restructure a category and result in different types of prototypes (bird watcher, or ornithologist). |
|
|
|
Exemplars
|
No single prototype but rather multiple examples convey what the concept represents
Vegetable Concept = Peas, Carrots, or Beans The more similar a specific exemplar is to a known category member, the faster it will be categorized Similar to Prototype View in that representation is not a definition Different from prototypes in that representation is not abstract Descriptions of specific examples To categorize, compare to stored examples |
PET study of Explicit & Implicit Memory
|
|
|
Combined classical/prototype theory
|
Evidence for both, so combine
Introduce the idea of the “core”: Defining features that item must have Prototype: Characteristics typical of examples – but not necessary |
|
|
|
Theory based views
|
Knowledge of the world informs and shapes our predictions about concepts
Features in a complex network of explanatory links indicate: 1. Relative importance of features 2. Relations among features Objects classified into concept that best explains the pattern of attributes |
|
|
|
Sorp/doon study
|
Sorp/Doon Story
Manipulated if the change was caused by an accident, a change in nature, or a control group reading about sorps Participants were then asked Is it more similar to a bird or an insect? Is it more likely to be a bird or an insect? |
|
|
|
Geometric approach
|
Concepts may lie in a geometric space
Subjective ratings can be used to place the concepts in a geometric space (Using “Multi-dimensional scaling”). Similarities may lie in a geometric space “Watermelon” and “Honeydew” are closer than “Honeydew” and “Tomato” Metric Axioms If concepts are really represented in a geometric space, similarities should satisfy certain properties (axioms) of (geo)metric space: 1. Minimality: The dissimilarity between a concept and itself must always be the smallest possible. But, a highly familiar concept is rated as more similar to itself than one that is less familiar. Apple-Apple more similar than Pomegranate-Pomegranate New York-New York more similar than Butte-Butte 2. Symmetry: The similarity between two concepts must be the same regardless of the order. How similar is an apple to a plum? How similar is a plum to an apple? Experiments on people’s ratings of similarity have found that: An unfamiliar category is judged more similar to a familiar category than vice versa. “Pomegranate” judged more similar to “apple” than “apple” is to “pomegranate.” This violates the symmetry axiom. 3. Triangle Inequality: If one concept is similar to a second concept, and the second concept is similar to the third concept, then the first and the third must be reasonably similar. Lemon is similar to Orange Orange is similar to Apricot Lemon is not similar to Apricot |
|
|
|
Feature-based measure;
Tversky's feature comparison (contrast) model |
Feature-based approaches look at features in common and different features.
Feature-based similarity approaches do not require metric axioms Tversky’s Contrast Model Similarity of L and O = (features that L and O have in common) minus (features of L that O doesn’t have) minus (features of O that L doesn’t have) L is the instance you need to categorize O is an example from the category. Similarity (L,O) = a*f(shared) - b*f(L but not O) - c*f(O but not L) a, b, and c are weights f is a function Explanations for axiom violations: Minimality: things you know well have more features, so the (a) (shared) part of the equation will be higher for familiar things than for unfamiliar (when you compare a concept to itself). Symmetry: (b) and (c) weights can be different, so the order of the comparison makes a difference. Triangle Inequality: two concepts can be similar to a third for different reasons, but have little in common themselves. |
|
|
Collins and Quillian semantic network model (1969)
|
Structure is hierarchical
Time to retrieve information based on number of links Cognitive economy Properties stored only at highest possible level Inheritance Lower-level items also share properties of higher level items Sentence Verification True or False A robin is a bird All chickens are birds All birds are chickens A cat is a bird Time to answer should depends on number of links according to TLC. This is not always the case like in reverse distance effect and typicality effects. Distance effect More links = More time to process Problems: Reverse distance effect 1. A dog is an animal. 2. A dog is a mammal. Respond faster to 1 than to 2, but not according to TLC. According to TLC, a dog is an animal should be slower than a dog is a mammal because animal is higher than mammal. Typicality effects A robin is a bird. A chicken is a bird. Robin-bird and chicken-bird time should be the same since they are both 1 level down from bird but in reality, Robin-bird time is more typical and has a faster processing time than chicken-bird time. Basic level When asked to (1) name features or (2) say what features objects have in common Easy with basic-level (chairs), hard with other levels (furniture). People will name an object by the basic level term. Children learn basic-level words before more specific subcategories (lower) or more general categories. (higher) TLC doesn’t account for this |
|
|
Revised model with spreading activation
|
Not hierarchical
Links vary in strength Explicitly shows information about relations A node is activated when person sees, reads, hears, thinks about a concept. Activation spreads to adjacent nodes. Spread of activation permits sentence verification. When activation intersects, decide whether relationships make statement true. Explaining Reverse distance and Typicality Explains typicality and reverse distance effect with length of connecting lines. Shorter line = more typical; Longer line = more atypical Dog is connected to animal by a shorter line than the line attaching dog to mammal. Explaining Priming Respond “yes” if both are words Respond “no” if one is not a word Doctor-Nurse: 855 fastest; doctor primes nurse Butter-Nurse: 940 No priming Wine-Plame: 980 No Priming and non-word When doctor node is activated, activation spreads to nurse node which is attached to doctor. |
|
|
|
In-class demonstration of sentence memory and data.
|
Read a number of sentences several times and asked us to memorize them. Later showed a list of sentences and asked which sentence was in the earlier read sentences. Results varied, but it turned out that none of the sentences were from the reading earlier. They just contained ideas from the reading.
Bransford and Franks Read story to participants. Later asked, was this probe sentence in the story? The more ideas from the story is in the sentence, the more confident they are that it was in the story. We remember meanings and ideas better than syntax and verbatim |
|
|
|
Episodic Memory
|
Episodic: Memory of events, personal experiences.
Tied to a specific time and place Personal perspective (your POV) First kiss, walking to class this morning, etc. |
|
|
|
Remembering verbatim vs. gist information
|
Memory for gist information is better than verbatim information
See Class demonstration of sentence memory and Sachs study |
|
|
|
Schemas
|
A schema a highly organized cognitive framework containing information about a person, group, or event.
Generalized conceptual knowledge used in understanding. Meaningfully organizes concepts. Tells us what to expect and also what unobserved or unstated information we can infer. Types of Schema Person schema – about particular people Group schema – about groups of people Event schemas – about particular events like eating at a restaurant Object schema – about particular objects Scene schema – about a particular scene like an office Evidence for Schemas Agree on what is in schemas. Recall steps from schemas in order. Read faster if story fits schema. Recall schema items that were not actually in story. Memory for objects in office – shown picture of an office and later asked to free recall as many items that were in the picture as possible. Easier to remember things you expect to see in an office (in office schema). Remember things consistent with schema (desk, chair). Memory not as good for objects if no expectations one way or other (bulletin board). False memory for things that were not in office but are in “office” schema (books). Brewer and Treyens (1981) Memory for graduate student's office 29/30 subjects remembered desk/chair 88 objects mentioned in recall 19 were inferred (not present) 9 people recalled books 8 people recalled skull 1 person recalled umbrella Participants recalled expected objects or highly unexpected objects Scripts – type of schema about events Structure captures general information about routine events – Eating in a restaurant, attending a movie, a visiting a doctor’s office Scripts have typical roles – (Customers, waiter, cook), (ticket vendor, patrons, refreshments), (doctor, nurse, patient) When we hear or read about a scripted event, our knowledge of the entire script is activated We can fill in or infer the scenes and actions that are not explicitly mentioned Scripts are often very similar across people. Schank and Abelman (1977) Visit a restaurant script Sit down Look at menu Order food Eat Pay Leave 73% of subjects produce the above actions 48% agreed on a further 9 actions Schemas and scripts: Can help us to understand, encode and retrieve information. Just as our perceptual systems may “fill in” missing information (paradoxical correspondence). Memories will “fill in” events if they correspond to a schema or script. Distortions from Schemas and Scripts Schema Intrusions May “remember” something because it is consistent with a schema, not because it really happened. -- “I always order extra ketchup” -- “Sure, I turned off the stove” Bower, Black, and Turner (1979) Participants read 18 stories 1, 2, or 3 stories read about each schema 1 story about going to the doctor 1 story about going to the dentist Health care schema activated for both Participants then asked if 3 particular types of events happened in the stories Events actually in stories Events consistent with schemas, but not actually in stories Novel, unrelated events Participants also rated their level of confidence about each of their answers Results Participants were confident About the actual events that they did read About schema-consistent events not actually in story The more stories read about a certain schema, the more confidence that the schema-consistent event was in a story Implications of the results Ideas contained in the schema become a part of the memory with items and events actually experienced Schemas Interference at Retrieval Told story about Carol Harris which bears some similarities to the story of Helen Keller. Story never explicitly mentioned deafness, blindness, and muteness. 1 Week later: 1/2 subjects told: really Helen Keller Test: “She is deaf, blind, and cannot speak.” Helen Keller group: 50% say yes. Carol Harris group: 5% say yes. Hellen Keller schema distorts memory of story War of Ghosts (American Indian tale) British Students read Native American myth Subsequently tried to recall Memory distorted to fit own knowledge Neglect information that doesn’t fit (ghosts) Distort facts to fit (canoe/boat, seals/fish) 'Something black came from his mouth' tended to become 'he frothed at the mouth', 'he vomited' or 'breath escaped from his mouth'. 'Hunting seals' tended to become 'fishing'. 'Canoe' tended to become 'boat' and 'paddles' to become 'oars'. Cultural schemas distorts memory |
|
|
|
Semantic vs. syntactic information: Sachs (Galileo paragraph) study.
|
Read a paragraph about Galileo with a probe sentence.
Showed participants one of 4 types of sentences and asked if it was in the story. Identical (No diff.): He sent a letter to Galileo, the great Italian scientist. Semantic difference: Galileo, the great Italian scientist, sent him a letter. Syntactic difference: A letter was sent to Galileo, the great Italian scientist. (Passive voice) Word order difference: He sent Galileo, the great Italian scientist, a letter. Results showed that memory for meaning or semantic was best. In other word we remember the gist better than we remember verbatim |
|
|
|
The role of prior knowledge
|
Prior knowledge helps us to understand and encode sequences of events.
Improves retention of information (usually). Balloons study – 14 ideas total Five Conditions: – Text alone, once 3.60 Text alone, twice 3.80 Suitable illustration after text 3.60 Partial illustration before text 4.00 Suitable illustration before text 8.00 Only the coherent illustration helped, and only when it was presented before the text. Laundry Study – Half class shown “this text is about laundry”; Other half not shown Laundry instructions shown without saying what it is Half shown “this text is about laundry” understood text better than half who didn’t see what “this text is about laundry” Giving context affects what is encoded and later recalled Number of ideas recalled: No topic Topic after Topic before 2.82 2.65 5.83 |
PET study of Explicit & Implicit Memory
|
|
|
Constructive memory
|
Memory is reconstructed, not reproduced.
"The reproduction [referring to a recall protocol] is a beautiful illustration of a strong tendency to rationalize, common to all my subjects. Whenever anything appeared incomprehensible or [odd], it was either omitted or explained. Rather rarely this rationalization was the effect of a conscious effort. More often it was effected apparently unwittingly, the subject transforming his original without suspecting what he was doing wrong." “Remembering is not the re-excitation of innumerable fixed, lifeless and fragmentary traces. It is an imaginative...construction, built out of the relation of our attitude towards a whole active mass of organized past reactions or experience. It is thus hardly ever really exact.” |
|
