lecture 15 - semantic memory - just the facts Flashcards

Question

spreading - activation theory and beyond

Answer 1

Collins and Loftus (1975) proposed a spreading-activation theory to resolve problems with Collins and Quillian’s (1969) theory. They argued (correctly!) that the notion of logically organized hierarchies was too inflexible. They assumed semantic memory is organized on the basis of semantic relatedness or semantic distance. We can assess semantic relatedness by asking people to decide how closely related pairs of words are. Alternatively, people can list as many members as possible of a particular category. Those members produced most often are regarded as most closely related to the category. You can see part of the organization of semantic memory assumed by Collins and Loftus (1975) in Figure 7.4. The length of the links between two concepts indicates the degree of semantic relatedness between them. Thus, for example, red is more closely related to orange than to sunsets. According to spreading-activation theory, the appropriate node in semantic memory is activated when we see, hear, or think about a concept. Activation then spreads rapidly to other concepts, with greater activation for concepts closely related semantically than those weakly related. Spreading-activation theory predicts the typicality effect (discussed earlier). Activation passes strongly and rapidly from robin to bird in the sentence, “A robin is a bird”—robin is a typical bird and robin and bird are closely related semantically. Less activation passes from penguin to bird in the sentence, “A penguin is a bird”—penguin is an atypical bird and penguin and bird are only weakly related.

Answer 2

indings Supporting Spreading Activation Theory Semantic priming: Faster word recognition when a word is preceded by a semantically related word (e.g., bread → butter). First shown by Meyer & Schvaneveldt (1971) using lexical decision tasks. Supported by later studies (e.g., Bauer & Just, 2017), though effects can be small/inconsistent (Heyman et al., 2018). False memory support: Schacter et al. (1996): People falsely recognized unpresented but semantically related words (e.g., doctor) after studying related words (nurse, hospital), due to high activation. Semantic distance: Kenett et al. (2017): Measured “path distance” (number of steps between words). Shorter distances led to higher perceived relatedness and better episodic memory (e.g., better recall and association tasks). Language production: Rose et al. (2019): Naming target pictures took longer when nearby distractors were semantically close (e.g., owl next to eagle)—demonstrates semantic interference.

Answer 3

The notion that activation spreads from a presented word or concept to semantically related words or concepts has been (and remains) extremely influential. The spreading activation theory has generally proved more successful than the hierarchical network theory at accounting for the various findings. One important reason is that it is much more flexible. What are the limitations with the theory? First, the notion that each concept in semantic memory is represented by a single node is oversimplified. As we will see shortly, information about most concepts is distributed in various brain regions rather than all being represented in a node. Second, the model implies that each concept has a single, fixed representation. In fact, however, our processing of any given concept is flexible (discussed further shortly). Consider the following two sentences: 1 Fred greatly enjoyed playing the piano. 2 Fred found it difficult to lift the piano. I imagine your processing of the word piano in the second sentence focused on the heaviness of pianos but did not do so when processing the first sentence. Such findings cannot easily be explained by the spreading-activation model. Third, several ways of measuring semantic distance have been proposed (see Kenett et al., 2017, for a review). There is, as yet, no consensus concerning the most appropriate measure of semantic distance.

Answer 4

Suppose you are shown a photograph of a chair and asked to identify it. You might provide various answers based on the relevant knowledge you have stored in semantic memory. For example, you might describe it as an item of furniture, a chair, or an easy chair. In fact, the great majority of people would describe it as a chair. Below we discuss why that is the case. The above example suggests concepts are organized into hierarchies. Rosch, Mervis, Gray, Johnson, and Boyes-Braem (1976) identified three levels within such hierarchies. There are superordinate categories (e.g., item of furniture) at the top, basic-level categories (e.g., chair) at the intermediate level, and subordinate categories (e.g., easy chair) at the bottom. We sometimes use superordinate categories (e.g., “That furniture is expensive”) or subordinate categories (e.g., “I love my new iPhone”). However, we generally have a strong preference for using basic-level categories. Rosch et al. (1976) asked participants to name pictured objects. Basic-level categories were used 1,595 times during the course of the experiment, subordinate names 14 times, and superordinate names only once. Why do we make such extensive use of basic-level categories? Most of the time, the basic level provides the best balance between informativeness and distinctiveness. Informativeness is lacking at the superordinate level (e.g., simply knowing an object is an item of furniture tells you little). Distinctiveness is lacking at the lowest level (e.g., most types of chairs possess very similar attributes or features). Rigoli, Pezzulo, Dolan, and Friston (2017) developed the above ideas. They argued that, “Categorization requires computations that have benefits in terms of goal achievement [e.g., selecting an appropriate action] but also costs (e.g., metabolic, opportunity costs, etc.) that need to be balanced against the benefits” (p. 2). Categorizing objects at the basic level generally permits selecting the most appropriate action while incurring relatively modest costs.

Answer 5

Categorization Levels & Brain Activation Basic-level advantage: Bauer & Just (2017): More brain areas (sensori-motor + language) are activated for basic-level concepts than for subordinate ones (mostly perceptual). Expertise effect: Tanaka & Taylor (1991): Experts (e.g., birdwatchers, dog experts) use subordinate categories more in their field—familiarity influences category use. Familiarity effect: Anaki & Bentin (2009): People categorize familiar subordinate items (e.g., Eiffel Tower) faster than basic-level ones (e.g., tower). Superordinate speed advantage: Prass et al. (2013): Categorization was fastest and most accurate at the superordinate level (e.g., animal vs. dog vs. beagle). Besson et al. (2017): Less information needed at the superordinate level explains this speed. Semantic dementia evidence: Rogers & Patterson (2007): Patients with severe semantic dementia performed better at the superordinate level—supports idea that it requires less processing.

Answer 6

As we have seen, numerous concepts are represented in semantic memory. What do these representations look like? This question is associated with considerable theoretical controversy (Mahon & Hickok, 2016). We start by considering the “traditional” viewpoint, according to which concept representations have the following characteristics: 1 They are abstract in nature and are thus detached from input (sensory) and output (motor) processes. 2 They are stable in that any given individual uses the same representation of a concept on different occasions. 3 Different people generally have fairly similar representations of any given concept. At the risk of oversimplification, traditional theories assume that concept representations “have the flavor of detached encyclopedia descriptions in a database of categorical knowledge about the world” (Barsalou, 2012, p. 247). This approach forms part of what Barsalou (2016) described as the sandwich model: cognition (including concept processing) is “sandwiched” between perception and action but is regarded as being almost totally separate from them. This model seems problematical because it is unclear how we could use such concept representations to perceive the visual world or decide what actions are appropriate in a given situation.

Answer 7

Barsalou (2012) argued in his situated simulation theory that all the theoretical assumptions of the traditional approach discussed above are incorrect. He argued we rarely process concepts in isolation. Instead, we process them in various settings with that processing being influenced by the current context or setting. More generally, our concept processing is influenced by our current goals and the major features of the situation. Barsalou (2009) illustrated the limitations with many previous theories by considering the concept of a bicycle. Traditionally, it was assumed a fairly complete abstract representation of the concept would be activated in all situations. This representation would resemble the Chambers Dictionary definition: “vehicle with two wheels one directly in front of the other, driven by pedals.” According to Barsalou (2009), those aspects of the bicycle concept activated depend on your current goals. For example, information about the tires will be activated if you need to repair your bicycle, whereas the height of the saddle will be activated if you want to ride it. In sum, Barsalou’s situated simulation theory makes various predictions. Of particular importance, it predicts that conceptual processing involves extensive use of the perceptual system and the motor or action system. Finally, Barsalou’s theoretical approach differs substantially from the traditional approach with respect to the conduct of experimentation on concepts. Most concept research has involved presenting words referring to concepts in isolation (in the absence of relevant context). This is appropriate if concepts are detached from perception and action. In contrast, it follows from Barsalou’s situated simulation theory that information acquired from studying concepts in isolation will often be limited and misleading (Barsalou, Dutriaux, & Scheepers, 2018, p. 1).

Answer 8

Perceptual Aspects of Concept Processing Wu & Barsalou (2009): People list different properties for same object in different contexts (e.g., lawn = blades, rolled-up lawn = soil). Shows concept processing is influenced by perceptual features and context. Participants often listed background-related properties (e.g., picnic for lawn)—supporting situated simulation theory. Abstract concepts: Still involve perceptual/contextual features (e.g., invention linked to laboratory). Barsalou & Wiemer-Hastings (2005): Abstract concepts often tied to concrete settings. Neuroimaging evidence: Wang et al. (2010): Perceptual brain regions more active for concrete than abstract concepts. Borghi et al. (2017): Abstract concepts sometimes involve perceptual processing. Limitations and Findings in Neuroimaging Studies Key Limitation: Neuroimaging studies often provide correlational, not causal, evidence about brain regions involved in conceptual processing (Mkrtychian et al., 2019). Motor System Activation: Hauk et al. (2004): Reading action verbs (e.g., lick, kick) activated motor cortex areas tied to those body parts. Supports motor involvement, but may reflect post-concept processing. Miller et al. (2018): Faster hand/foot responses when words matched the responding limb (e.g., kick → foot). Suggests motor system can influence word processing but… EEG showed no early motor activation → suggests effect was semantic, not motor. Task Design Matters: Differences in findings may be due to task timing: Hauk et al. used slower tasks (allowing motor imagery), Miller et al. used speeded tasks (limited time for motor involvement). Neuropsychological Evidence: Vannuscorps et al. (2016): Patients with motor-system damage (e.g., patient JR) showed intact understanding of action-related concepts. Challenges the idea that motor areas are necessary for conceptual processing.

Answer 9

There is much support for the theoretical assumption that conceptual processing in everyday life Memory otten involves the perceptual and motor systems. This assumption helps to explain why concept processing varies across situations depending on the individual's goals. In other words, the precise way we process a concept depends on the situation and the perceptual and motor processes engaged by the current task. In essence, Barsalou's approach explains much of the flexibility that characterizes conceptual processing. What are the main limitations of Barsalou's theoretical approach? First, he exaggerates the extent to which concept processing varies across time and across situations or contexts. The traditional view that concepts possess a stable, abstract core has not been disproved by Barsalou (Borghesani & Piazza, 2017). As we will see below, both theoretical approaches are partially correct-concepts have a stable core and concept processing is context-dependent. Second, much of our concept knowledge does not consist simply of perceptual and motor features. Borghesani and Piazza (2017, p. 8) give the following example: "Tomatoes are native to South and Central America." Third, we can recognize the similarities between concepts not sharing perceptual or motor features. For example, we categorize watermelon and blackberry as fruit even though they are very different visually and we do not eat them using the same (or similar) motor actions. Fourth, the finding that concept processing often includes perceptual and/or motor features does not mean it is generally necessary to use perceptual and/or motor processes to understand concepts. Alternatively, perceptual and motor processes may not be necessary and may even occur after concept meaning has been accessed (Mahon & Hickok, 2016). The finding that some patients with damage to their motor system can nevertheless understand action-related words (Vannuscorps et al., 2016) is more consistent with the latter viewpoint as are the findings of Miller et al. (2018).

Answer 10

How Semantic Information Is Stored Not stored in one place: Knowledge about a concept (e.g., your cat) is distributed across different brain areas (e.g., visual, auditory, motor features). Supports a feature-based model, aligning with Barsalou’s theory of perceptual and motor involvement. Evidence from Brain-Damaged Patients: Studying patients helps reveal how semantic memory is organized. If features are stored in distinct brain regions, damage should cause category-specific deficits. Category-Specific Deficits: Herpes simplex encephalitis (damaging antero-medial temporal lobes) often leads to selective problems recognizing biological entities (Gainotti, 2018). Chen et al. (2017): Confirmed various category-specific deficits exist. Interpretation Challenges: Hard to isolate why certain categories (e.g., living things) are harder to recognize. Possible reasons: higher visual similarity, structural complexity, and less motor association (Marques et al., 2013).

Answer 11

We saw earlier that concept processing often (but not always) involves the perceptual and motor systems. However, there are several reasons for assuming there is more than that to concept processing. First, we would not have coherent concepts if our processing of any given concept varied considerably across situations. Second, we can detect similarities in concepts that are very different perceptually. For example we know scallops and prawns are both shellfish even though they differ in shape, color, and form of movement (Patterson, Nestor, & Rogers, 2007). Patterson et al. (2007) put forward a hub-and-spoke model (developed by Lambon Ralph, Jefferies, Patterson, and Rogers, 2017) combining several ideas discussed earlier. You can see key features of this model in Figure 7.6. The spokes in the model consist of several modality-specific brain areas where sensory and motor processing occur. The six spokes shown in Figure 7.6 relate to visual features, verbal descriptors, olfaction (smell), sounds, praxis (motor information), and somatosensory information (sensations from the skin and internal organs). Each concept also has a "hub"—a general, modality-independent unified conceptual representation that provides an efficient way of integrating our knowledge of any given concept. It is assumed within the theory that hubs are located within the anterior temporal lobes.

Answer 12

Hub-and-Spoke Model of Semantic Memory Hub = Anterior Temporal Lobes (ATLs) Binder et al. (2009): Meta-analysis of 120 neuroimaging studies—ATLs consistently activated during semantic tasks. Murphy et al. (2017): Ventral ATL = semantic hub (general meaning). Anterior ATL = responds to modality differences (visual vs. auditory), not truly hub-like. Evidence from Semantic Dementia Mayberry et al. (2011): Semantic dementia causes loss of core/hub info, leading to blurring of category boundaries. Patients struggled with: Atypical members (e.g., emu as a bird), Similar non-members (e.g., butterfly as bird-like). Spokes = Modality-Specific Areas Cree & McRae (2003): Identified 7 category-specific deficit patterns across brain-damaged patients. Most commonly impaired features: color, taste, smell, visual motion, function. Supports idea that different properties of concepts rely on distinct brain regions. Experimental Support: tDCS Stimulation Ishibashi et al. (2018): Used anodal tDCS (to enhance brain activity) during tool tasks: Anterior Temporal Lobe (ATL): improved both function and manipulation tasks—supports ATL as a hub. Inferior Parietal Lobule: improved only manipulation, consistent with its role in action-related processing

Answer 13

Memory hub-and-spoke model provides a more comprehensive account of semantic memory than previous theoretical approaches. There is considerable support for the notion that concepts are represented in semantic memory by a combination of abstract core (hub) and modality-specific information (spokes). There has been good progress in identifying the brain areas associated with hubs and the various types of spokes. Memory What are the model's main limitations? First, the role of the anterior temporal lobes in concept processing is more complex than assumed theoretically (e.g., Murphy et al., 2017). Second, more remains to be discovered about the information contained within concept hubs. For example, is more information stored in the hubs of very familiar concepts than less familiar ones? Third, how is modality-specific "spoke" information integrated with modality-independent "hub" information? Fourth, there is still no consensus concerning the number and nature of concept "spokes."

Answer 14

The original hub-and-spoke model focused on the semantic representations of concepts. However, Memory " vro are to use our semantic knowledge effectively, we need to control our processing to emphasize those aspects of our semantic knowledge of most current relevance in the present context. For example, consider the piano concept. As Hoffman, McClelland, and Lambon Ralph (2018) pointed out, if we want to play a piano, we need to focus on its keys and pedals. In contrast, if we want to move a piano, this information is irrelevant and our focus should be on features of a piano such as its weight and whether or not it has wheels. The notion that current context strongly influences which aspects of any given concept are relevant has led theorists (Lambon Ralph et al., 2017) to develop the hub-and-spoke model. Hoffman et al. (2018) produced a detailed model combining a hub-and-spoke architecture with a mechanism to take account of the current context that accurately predicted concept processing in brain-damaged and healthy individuals.

Answer 15

Our discussion so far may have created the false impression that nearly all the information in Memory semantic memory is in the form of simple concepts. In fact, however, much of the knowledge we have stored in semantic memory consists of larger structures of information. What do these larger knowledge structures look like? Frederic Bartlett (1932) provided an extremely influential answer to that question. He argued for the importance of schemas, which are "superordinate knowledge structures that reflect abstracted commonalities across multiple experiences" (Gilboa & Marlatte, 2017, p. 618). Bartlett's key insight was that what we remember (including our errors when remembering) is strongly affected by our schematic knowledge. Ghosh and Gilboa (2014) provided a more detailed definition of schemas. They argued schemas possess four necessary and sufficient features: Associative structure: schemas consist of interconnected units. Basis in multiple episodes: schemas consist of integrated information based on several similar events. Lack of unit detail: this follows from the variability of events from which any given schema is formed. 4 Adaptability: schemas change and adapt as they are updated in the light of new information. There are various kinds of schemas. Scripts contain information about sequences of events. For example, Bower, Black, and Turner (1979) asked people to list actions typically occurring during a restaurant meal. At least 73% mentioned the following: being given a menu; ordering; eating; and paying the bill. Frames are knowledge structures referring to some aspect of the world (e.g., building) containing fixed structural information (e.g., has floors; has walls) and slots for variable information (e.g., materials from which the building is constructed).

Answer 16

Concept vs. Schema Processing Prediction: Different brain areas should activate for concepts (individual words) and schemas (broader organizational structures). Patients might show double dissociation: difficulties accessing concept-based info vs. schema-based info, suggesting distinct brain mechanisms. Brain Activation Concept Processing: Anterior Temporal Lobes (ATLs) activated (Binder et al., 2009; Murphy et al., 2017). Schema Processing: Ventral Prefrontal Cortex (PFC) especially important (Gilboa & Marlatte, 2017). Overlap: ATLs involved in both concept and schema processing. Evidence from Brain-Damaged Patients Semantic Dementia (SD): Patients with damage to ATLs show impaired concept processing but retain some ability to use schema-based info (Bier et al., 2013). Fronto-Temporal Dementia (FTD): Damage to prefrontal cortex impairs script memory, especially sequencing (Cosentino et al., 2006). Damage to ventromedial prefrontal cortex affects schema-related tasks (Ghosh et al., 2014). Double Dissociation in Schema/Concept Processing Ghosh et al. (2014): Ventromedial PFC damage impairs schema-related processing. Zahn et al. (2017): Fronto-polar cortex damage = poorer script knowledge than social concept knowledge. Summary: Concept Memory = primarily involves anterior temporal lobes. Schema/Script Memory = involves prefrontal cortex (especially ventromedial). Overlap: Concept knowledge is needed to use schemas effectively (e.g., preparing a meal).

Answer 17

We have seen that schematic knowledge in the form of scripts is useful because it allows us to form Memory realistic expectations about the immediate future. Schemas (including scripts) make the world more predictable than would otherwise be the case because our expectations are generally confirmed. If our script-based expectations are disconfirmed, we usually take action. For example, if no menu is produced in a restaurant, we try to catch the eye of the waiter or waitress. There are other reasons why schematic knowledge is useful. First, schemas are important in reading and listening because they allow us to fill in the gaps in what we are reading or listening to and so enhance our understanding. More specifically, they enable us to draw inferences as we read or listen (see Box 7.2). Second, schemas help to prevent cognitive overload. Consider stereotypes (schemas involving simplified generalizations about various groups). When meeting someone for the first time, we often use stereotypical information (e.g., about their sex, age, and ethnicity) to help form an impression of that person. It is simpler and less demanding (but potentially very misleading) to use such information rather than engage in detailed cognitive processing of his/her behavior (Macrae & Bodenhausen, 2000). Potential disadvantages of relying on stereotypical information were shown by Reynolds, Garnham, and Oakhill (2006). Read the following passage they used in their study and then answer Memory the yuesuon: A man and his son were away for a trip. They were driving along the highway when they had a terrible accident. The man was killed outright but his son was alive, although badly injured. The son was rushed to the hospital and was to have an emergency operation. On entering the operating theater, the surgeon looked at the boy, and said, "I can't do this operation. This is my son." How can this be? If you found the problem difficult, you are in good company. We tend to have a stereotypical view that surgeons are men. However, some surgeons are female and the surgeon in the passage above was the boy's mother. Thus, schemas in the form of stereotypical information can interfere with problem solving. Third, schematic information can assist us when we are trying to recognize an object. For example, Auckland, Cave, and Donnelly (2007) presented observers briefly with a target object (e.g., playing cards) surrounded by four context objects. Sometimes the context objects were semantically related to the target object (e.g., dice; chess pieces; plastic chips; dominoes) and so provided information relevant to the game schema. The target was recognized more often in this condition than when the context objects were semantically unrelated. Lupyan (2017) reviewed research showing how top-down processes triggered by contextual or schematic information facilitate object recognition. Fourth, as mentioned earlier, Ghosh and Gilboa (2014) identified adaptability as an important lemory uspect of schemas. As Richter, Bays, Jeyarathnarajah, and Simons (2019) pointed out, adaptability is very useful. It means we can adapt to changing environmental conditions by flexibly making changes to incorporate additional information to a pre-existing schema structure or by modifying the existing structure itself. Richter et al. showed experimentally how schemas are modified and updated when the knowledge within them no longer reflects current environmental conditions.

Answer 18

So far we have emphasized the value of schematic knowledge-it makes the world a more Memory predictable place, enhances our understanding of what we read and other people say, and it facilitates visual perception of the world around us. However, Bartlett (1932) argued that schematic knowledge can cause significant memory costs. He argued our memory for stories is affected not only by the presented story itself but also by the participant's store of relevant schematic knowledge. Bartlett tested the above notions by presenting people with stories producing a conflict between what was presented and their prior knowledge. Suppose people read a story taken from a different culture. Their prior knowledge might produce distortions in the remembered version of the story, making it more conventional and acceptable from their own cultural background. Bartlett (1932) carried out several studies in which English students read and recalled stories taken from the North American Indian culture. One such story was The War of the Ghosts (reproduced on p. 165). As predicted, participants' schematic knowledge in the form of cultural expectations led to numerous recall errors conforming to that knowledge. Bartlett used the term rationalization for this type of error. According to Bartlett (1932), memory for the precise information presented is forgotten over time whereas memory for the underlying schemas is not. Thus, there should be more rationalization errors (which depend on schematic knowledge) at longer retention intervals. In the interests of historical accuracy, it should be noted that Bartlett's (1932) approach was less original than typically assumed (Davis, 2018). Henderson (1903) had previously used experimental paradigm very similar to Bartlett's, and had anticipated many of Bartlett's theoretical ideas.

Answer 19

Numerous experimental studies have supported Bartlett's general approach (see Chapter 6 for a detailed account). However, it is arguable that most of these studies lack ecological validity (applicability to everyday life). For example, many studies involved participants reading artificially constructed texts knowing their memory for these texts would be assessed. In contrast, Brewer and Treyens (1981) argued that most information we remember during our everyday lives is acquired incidentally rather than deliberately. Brewer and Treyens In their own research, Brewer and Treyens (1981) used a naturalistic learning situation. Participants spent about 35 seconds in a room designed to look like a graduate student's office (see photograph). The room contained a mixture of schema-consistent objects you would expect to find in a graduate student's office (e.g., desk, calendar, eraser, pencils) and schema-inconsistent objects (e.g., skull; toy top). Some schema-consistent objects (e.g., books) were omitted. Finally, participants received unexpected recall and recognition tests. Schemas and Memory: Effects on Recall and Recognition 🧠 Brewer & Treyens (1981) Participants waited in an office and were later asked to recall its contents. False memories: High-confidence “recall” of objects that were not present but were schema-consistent (e.g., books, filing cabinet). Shows schemas can distort memory. True recall: More schema-consistent than inconsistent items remembered—positive effect of schemas. 🧠 Webb, Turney & Dennis (2016) Participants viewed scenes (e.g., bathroom) with both schema-consistent and inconsistent objects. Later did a recognition-memory test including: Seen consistent/inconsistent objects Unseen schema-consistent lures (e.g., toilet paper) Neuroimaging findings: Schema-inconsistent retrieval required more prefrontal cortex activity → more effortful cognitive control. False memories for nonpresented schema-consistent objects involved greater lateral temporal activation, suggesting reliance on gist-based (schematic) retrieval. Schema-consistent retrieval is easier but more prone to false memories. 🧠 Steyvers & Hemmer (2012) – A More Balanced View Argued previous studies exaggerate memory errors. In real-life, schema-driven guesses are usually accurate. Experiments: Naming objects expected in various real-world scenes: False recall was lower for high schema-relevance objects (9%) vs. low schema-relevance (18%). Correct guesses more likely for high-relevance objects. Photograph-based recall: Best recall for high schema-consistency objects. Also, very schema-inconsistent objects were remembered well — von Restorff effect (distinctive items stand out). 📌 Key Takeaways Schemas improve recall of consistent info but can also cause false memories. Inconsistent info requires more effort to retrieve but may be well remembered if distinctive. Memory is adaptively efficient, not simply error-prone—schemas often help in realistic settings.

Answer 20

Schemas theories have proved generally successful. There is compelling evidence that learning and Memory merory often involve top-down processes triggered by schematic knowledge. More generally, schemas allow us to form expectations that are often confirmed subsequently. Schemas are adaptable and can be altered in response to changing environmental conditions. Schemas often enhance long-term memory for both schema-consistent and schema-inconsistent information (Greve, Cooper, Tibon, & Henson, 2019; Steyvers & Hemmer, 2012) with the latter occurring because schema-inconsistent information conflicts with learners' expectations and leads to more thorough encoding. However, use of schematic knowledge can lead to various memory distortions and errors. What are the limitations of schema theories? First, they are typically vague with their precise scope and nature remaining unclear. In addition, much remains to be discovered about how episodic and semantic memory processes interact. Second, our memory representations are often more complex than implied by schema theories. For example, we do not just have a basic restaurant script. We also know you do not sit down before ordering your food at fast-food restaurants, expensive restaurants often have wine waiters, you need to book at some restaurants but not others, and so on. Most schema-based theories have not focused on these complexities. Third, most schema theories exaggerate the number of schema-driven memory errors occurring in everyday life. As Steyvers and Hemmer (2012, p. 140) argued, "In a naturalistic environment, the prior knowledge of the occurrence of objects in a given scene type can lead to effective guesses ... Such guessing with prior [schematic] knowledge can result in high accuracy and a low number of intrusions."

Answer 21

There is an important distinction between semantic and episodic memory: the latter involves more conscious recollection of the past and is more "personal." The distinction between semantic and episodic memory is supported by research on brain-damaged patients: amnesic patients typically have greater problems with episodic than semantic memory whereas patients with semantic dementia show the opposite pattern. In spite of the differences between semantic and episodic memory, many long-term memories combine episodic and semantic information. In addition, memories that are initially episodic can become transformed into semantic memories over time: semanticization. According to Collins and Quillian's (1969) hierarchical network model, concepts are represented by nodes within hierarchical networks; concept properties or features are stored as far up the hierarchy as possible. The hierarchical network theory is based on the erroneous assumption that concepts are stored in semantic memory much more neatly than is actually the case. The theory also fails to acknowledge that many concepts are fuzzy or imprecise. According to Collins and Loftus's (1975) spreading-activation theory, semantic memory is organized by semantic distance. Activation of any given concept causes activation to spread to all other related concepts. The spreading-activation theory assumes all information about a given concept is stored at a single node, which is a substantial oversimplification. Many concepts within semantic memory are organized into hierarches consisting of superordinate, basic, and subordinate levels. Individuals generally prefer the basic level because it combines informativeness and distinctiveness. Memory nuwever, experts often prefer the subordinate level because it is more informative than the basic level. Categorization typically occurs faster at the superordinate level than the basic level because less information is required. Barsalou claimed in his situated simulation theory that concept processing (even with abstract concepts) involves the perceptual and motor systems and depends very much on the current context. It is the case that concept processing often includes perceptual and/or motor features, but this does not mean it is necessary to use perceptual and/or motor processes to understand concepts. Barsalou's theoretical approach de-emphasizes that evidence that most concepts have a stable, central core of meaning unaffected by context. According to the hub-and-spoke model, concepts consist of hubs (unified abstract representations) and spokes (modality-specific information). Evidence from patients with semantic dementia indicates that hubs are stored in the anterior temporal lobes. In contrast, spokes are stored in several different brain areas, as is indicated by the existence of category-specific deficits and brain-stimulation studies. It is not clear within the hub-and-spoke model how modality-specific "spoke" information integrated with modality-independent "hub" information. Schemas are well-integrated chunks of knowledge about the world, events, people, and actions. As such, they are broader in scope than concepts. Research on concepts and schemas has provided some evidence of a double dissociation: patients with semantic dementia generally have greater problems with accessing concepts than schematic information, whereas those with fronto-temporal dementia show the opposite pattern. Schamas are useful because they allow us to make predictions about the immediate future, to make Memory inferences during reading, and they adapt to take account of changing environmental conditions. Schemas are also useful because they often enhance long-term memory for both schema-consistent and schema-inconsistent information. Schematic knowledge can cause distortions in long-term memory when what we read or hear is inconsistent with that knowledge. However, such distortions are relatively infrequent when we are exposed to natural scenes containing mostly objects that are highly probable in the particular context. In contrast, memory distortions are much more common when we are exposed to manipulated scenes in the Laboratory where high-probability objects are often replaced by low-probability ones.

lecture 15 - semantic memory - just the facts Flashcards

(45 cards)