Foundations of Cognition I - Week 08 Notes (PDF)

2024-11-14 Foundations of cognition I PSYC-2014 FAO – Week 08 Last Class: Test #2 – grades to be returned soon Last time we discussed memory codes The formats of information stored in memory Factors affecting transfer of information into memory Today: Two short chapters: Visual...

2024-11-14 Foundations of cognition I PSYC-2014 FAO – Week 08 Last Class: Test #2 – grades to be returned soon Last time we discussed memory codes The formats of information stored in memory Factors affecting transfer of information into memory Today: Two short chapters: Visual Imagery (Visual images) Categorization 1  2024-11-14 Verbal vs. Spatial Knowledge We can distinguish between different types of knowledge: Verbal Knowledge: Knowledge expressed in language (i.e., words, phrases, rehearsed numbers, etc.) Spatial Knowledge: Knowledge of spatial relations, that may be stored as imagery Visual Imagery Earlier, we discussed how some memory types are modal I.e., Memories are stored in a form equivalent to how they are experienced The same is thought to apply to visual information both in memory and during mental operations These are referred to as visual imagery (or visual images) Visual Images: refers to mental representations or perceptions of visual information formed within the brain Do Visual Images Exist in Memory? Historically, some researchers questioned whether visual images exist or not I.e., Visual images may be amodal (stored as knowledge/information that could be semantic in nature – rather than stored as “pictures” in the mind) However: we now know from cognitive neuroscience that visual experience is re- capitulated in the brain during imagining 2  2024-11-14 Activation of the visual cortex (V1) occurs when people either (a) view a flashing light pattern, (b) imagine the flashing light pattern, or (c) imagine walking in their hometown. Higher levels of activation are shown in red. Memory for Pictures Most of the studies on memory codes that we discussed earlier rely on verbal knowledge However, people are generally very good at recognition for images Recognition for images is typically much better than it is for words In fact, humans have remarkably accurate recognition of images Konkle (2010): Memory for Pictures Observers viewed thousands of scenes over 5.5 Hours Then, completed forced-choice tests Over 2900 images from 128 categories 3  2024-11-14 Konkle (2010): Scene category example (golf course) Konkle (2010): Memory for Pictures Observers viewed thousands of scenes over 5.5 Hours Then, completed forced-choice tests Over 2900 images from 128 categories Participants were asked: Which of these two images did you observe earlier? When one image was from a category they saw earlier and one was from a new previously unseen category: 96% guesses were correct Recognition vs. Recall This reflects a difference between Recognition and Recall Recognition: Identifying previously encountered information when presented with cues or options Recall: Retrieving information from memory without explicit cues, relying solely on mental effort People tend to be much better at recognition that recall 4  2024-11-14 Konkle (2010): Memory for Pictures When they were shown a pairing of two similar images from the same category, accuracy was still very high 76-84% of guesses were correct (Exemplar Condition) This is a much more difficult appearing task Demonstrates remarkable memory for scenes Can you see a pickle? Can you see ‘pickling’? 5  2024-11-14 Can you see ‘pickled’? Which of these feels like it would be the hardest to see? Oath Anvil Hallway Honor Goat Dual Coding Theory (Paivio) Some words are more directly relate to something you can see Allan Paivio: Dual Coding Theory (University of Western Ontario) Proposed that there are two main ways of making associations in memory: (1) Verbal associations (2) Visual imagery These two are not equal in terms of their potential to be stored in memory 6  2024-11-14 Paivio (1968): Imagery Potential Paivio proposed that some words have more imagery potential I.e., They more strongly relate to things you can see E.g., The words “Effort”, “Poetry”, “Necessity”, “Preponderance”, etc. are all more verbal in nature than visual They have relatively little relation to a concrete object Paivio (1968): Imagery Potential On the other hand – words like “Apple”, “Anvil”, “Cow”, “Airplane”, and “Hospital” are very easy to visualize They have high linkage to concrete objects that can be seen/observed Paivio referred to this as Imagery Potential Imagery Potential: The ease with which a concept can be imagined Paivio (1968): Imagery Potential Easily-imaged words have more imagery potential, and words which are difficult to image have less imagery potential Since images are remembered better than words, memory should be better for words with high imagery potential 7  2024-11-14 Paivio (1968): Imagery Potential Paivio also related imagery potential to whether or not words refer to objects that are tangible (something perceptible or concrete) Paivio referred to the scale of imagery potential as the concrete-abstract dimension: The extent to which a concept can be represented by a picture. Ranges from very concrete to very abstract The Concrete-Abstract Dimension “Chair” “Pickle” “Substantiated” “Officer” “Dutiful” “Hotel” “Transcendence” “Oath” CONCRETE ABSTRACT Paivio (1968): Imagery Potential 1968 Study: How does imagery potential affect recall? Task: Presented word-pairs to participants Pairs consisted of high- and low-imagery potential words High-High (Imagery Potential), High-Low, Low-High, and Low-Low E.g. “Chair” + “Suitcase” would be a ‘High-High’ pair “Quality” + “Necessity” would be a ‘low-low’ pair 8  2024-11-14 Paivio (1968): Imagery Potential Study findings: memory is much better for words with high imagery-potential than low imagery-potential Recall was highest for H-H Pairs Recall was lowest for L-L pairs Conclusion: Memory is better for words that refer to concrete objects vs. abstract concepts that are not easily visualized Mnemonic Strategies Some mnemonic strategies can exploit this fact to improve memory for abstract words One approach: use a Keyword Keyword: A concrete word that sounds like an abstract word so that it can be substituted for the abstract word in an interactive image. Mnemonic Strategies Keyword Method: using keywords to improve paired-associated learning. E.g., Allegory sounds like Story By remembering that the two are related, we can remember the more abstract one (allegory) more easily This also works for remembering non-native language words (e.g., An English-speaker remembering Russian words because they sound similar to something in English that is concrete) 9  2024-11-14 Spontaneous Imagery We sometimes produces spontaneous visual imagery without input from the senses Two examples are Dreams and Daydreams Dreams: immersive and often vivid mental experiences that occur during sleep Daydreams: are spontaneous, internally generated fantasies or thoughts that occur while awake Spontaneous Imagery It is unclear whether dreams and daydreams serve a biological function One hypothesis (we already touched on) is that dreams enable the rehearsal of Embodied Cognition I.e., Memories of movement (procedural memories) Work of J. Allan Hobson & Others There are two main lines of evidence for this Sleep & Memory (1) The first line of evidence is that sleep improves memory, but Learning also affects sleep (suggesting rehearsal is happening during the night) 10  2024-11-14 Sleep & Memory (1) The first line of evidence is that sleep affects learning, but Learning also affects sleep (suggesting rehearsal is happening during the night) After learning, REM-sleep Intensity increases The amount of this increase is correlated with gains in performance across the night Sleep & Memory (1) The first line of evidence is that sleep improves memory, suggesting rehearsal is happening during the night (2) The second line of evidence is that those with REM Behaviour Disorder (a sleep disorder) appear to act out procedural memories (movements) during the night This results from damage to the pontine inhibitory area, which prevents movements during REM sleep 11  2024-11-14 Threat Simulation Theory Many of these dreams appear to re-enact conflict, not just memories from throughout the day 12  2024-11-14 Threat Simulation Theory As a result, some have suggested that embodied cognition during the night also serves an evolutionary purpose: visualizing & preparing to deal with hypothetical conflicts or threats Threat Simulation Theory: the function of dreams is to rehearse threatening situations Proposed by Antti Revonsuo (Finnish Psychologist) Categorization 13  2024-11-14 Organizing Knowledge The next two topics for the course will cover how knowledge is organized Humans are very adept at organizing knowledge We have several strategies for organizing knowledge One such strategy is categorization Categorization: the mental process of organizing and classifying information into distinct groups based on shared characteristics or features. Organizing Knowledge Organizing knowledge has also been proposed to offer several specific adaptive benefits: (1) Reducing the complexity of out environment (2) Assisting with object identification (3) Reducing the need for constant learning (4) Assists in decision making (5) Allows us to order and relate classes of objects and events Categorization Categories are concepts which are the ‘building blocks’ of knowledge Involves classification (i.e., assigning chunks of knowledge to a particular group or category) E.g., Crows are birds, apples are fruit, etc. Helps us to efficiently organize mental information 14  2024-11-14 Categorization: Concept Identification Researchers initially studied categorization using a task called concept identification Concept Identification: A task that requires deciding whether an item is an example of a concept or not Concepts are defined by some logical rule Logical Rule: A rule based on logical relations Logical Rules Logical rules can be used to discern or define categories Several types: Disjunctive: Patterns that use or (e.g., Anything large OR round are part of the category) Conjunctive: Patterns that use and (e.g., Anything with wheels AND a motor s a vehicle) Research tasks involve determining which objects belong to which category. The relevant attributes are size and shape. What is the rule for which objects are part of this category? (+ and – indicate whether an item is part of the category or not) 15  2024-11-14 Rule Learning The rule is: Large OR Circular This is a rule learning task Two possible approaches to this task: Rule Learning: When participants are told the relative attributes, and must discover the rule. Attribute Learning: When participants are told the type of rule, but must discover the relevant attributes for inclusion (e.g., size, shape, etc.). The problem with Concept Identification Fails to address some observations (1) Hierarchical organization: An organization strategy in which larger categories are partitioned into smaller ones. We observe this is human thought organization Parka is a type of coat; Coat is a type of clothing; clothing is a type of object Kiwi is a type of fruit; Fruit is a type of food; Food is a type of object The problem with Concept Identification Fails to address some observations (2) Continuous Dimensions: Many items that we categorize are not naturally separated E.g., Red, yellow, blue, green…..why do we categorize colors this way? The visual spectrum is continuous, and often colors bleed into one another I.e., “Pink-red” is different than “Orange-red” Different cultures categorize colors differently (e.g., Russian has more words for Blue than English does) 16  2024-11-14 The problem with Concept Identification Fails to address some observations (3) Categorized objects differ in their typicality (how well a category memory represents its category) E.g., Many say that Blood red is more red than pink-red. The problem with Concept Identification The scientific definition of water is H2O If Logical Rules are the correct way we categorize things, something that is mostly water should be categorized as water However: People rate some types as “more water” than others E.g., Pure, Ocean, & Tap water were rated as “4” on typicality However, they have varying amounts of H2O Natural Categories Members of mental categories are not all equally good members Rosch et al. (1976): Put forth the idea of category levels Superordinate category: A large category at the top of a hierarchy (e.g., Tools, vehicles) Basic-level categories: An intermediate category in the middle (e.g., Saw, truck) Subordinate categories: Small categories at the bottom of a hierarchy (e.g., Jigsaw, Table saw, Hand saw) 17  2024-11-14 Rosch: Processing Speed of Natural Categories Rosch asked participants to identify objects at different levels Found that people identify objects the fastest at the Basic Level Volunteer: Identify the Object Is this a Table or a Car? 18  2024-11-14 Is this a Coffee Table or a Side Table? Did one feel easier? Did one feel easier or faster to identify? Did one feel easier? Did one feel easier? After many trials, Rosch concludes the basic-level is faster/easier May be due to interference: More similar objects have more conflicting attributes Basic-level objects are maximally different from one another (e.g., Table vs. Car) 19  2024-11-14 On a scale of 1-10, how different is a chicken from a duck? On a scale of 1-10, how different is a chicken from an ostrich? Multidimensional Scaling These are both birds However, one pair feels more different than the other pair The two have more dissimilar attributes This is what is addressed by multidimensional scaling 20  2024-11-14 Multidimensional Scaling Some research has focused on categorization of non-discrete objects Task: Participants rate how similar two objects are After many successive trials, you can plot the results based on distances from one another This reveals how similar participants thought two objects within a category are Multidimensional Scaling E.g., A hawk seems more similar to an eagle than it does do a goose These are very similar Which type of bird is the most bird? 21  2024-11-14 Average Distance Rule The average of these will reveal the “most average” item for that category This is a Prototype Prototype: A category example that has average attributes Multidimensional Scaling In this example, notice that the word “Bird” is very close to the average (the middle of the plot) of all birds This is the Prototype – the mental representation of the most average bird This is a mental “average” of all members of the category Prototypes vs. Exemplars Some research has focused on whether people use prototypes for categorization, or other approaches Prototype Rule: When seeing new objects, you mentally categorize is according to the existing prototype you know that is is most similar to the new item. Exemplar Rule: Patterns are categorized by comparing their similarity to existing category members (not an average of them all) 22  2024-11-14 Prototypes vs. Exemplars There is some support for both theories People seem to use both approaches In particular: When categories are small (have few items) people tend to remember and compare to the individual items (exemplars) in the existing category This is also true when a category is poorly structured 23