Class Notes (838,030)
Canada (510,640)
COG250Y1 (127)

Cognitive Science Notes.docx

16 Pages
Unlock Document

Cognitive Science
John Vervaeke

Cognitive Science Notes LECTURE NOTES Generic Nominalism (weakest approach): -All separate sciences contributing -Cognition: set of processes by which human beings take in, store and transform & then apply information in behaviour -Problematic: because it suggests each contributing subject doesn‟t communicate with the others  cognition must be studied at various levels Interdisciplinary Ecclecticism (stronger approach): -Disciplines communicate, but no significant integration of them occurs ex. Interfaith Dialogue -Conferences 1 or 2 times a year -Problematic: unstable resulting in: a) loss of interest and return to GN or b) too weak to manage and becomes SI Synoptic Integration (strongest approach): -Integration of various levels of discourse and theory with others -“From Philosophy to the Brain” 1. Plausible Theoretical Constructs -Convergents increase in trustworthiness of construct (ex. Numbers) 2. Usefulness (Integration, Kitcher) Ex. Hands= multi-apt (effectively function in many contexts) Ex. Metaphor- consider processing, not product -Aptness: correct description for transfer a) Need a difference to make one (ex. bees are hornets) b) Reasonable amount of correlation (ex. arguments are chairs) -Multi-Apt: ex. F=ma unification of formulae elegance -There should be a high level of profundity: depth of balanced convergence and elegance -Avoid biased „weirdness‟ (lack of trustworthiness) -Science explains the familiar with the unfamiliar, but this is different than theoretical weirdness -Get various disciplines to converge to explain many different levels -Cognitive Science is governed by an imperative -Naturalistic Explanations Three Components of Naturalistic Imperative of Cognitive Science (Revolutions) 1. Thales “All is moist. The lodestone is psyche” Everything is made of water. The lodestone is the life force of living things. -Explaining the familiar with the unfamiliar -Ontological Analysis: study of content of peoples‟ analyses separated from their representation -Cognitive Science analyses goes from complex to basic 2. Scientific Revolution -Rene Descartes formalized (explanation in terms of mathematical terminology) *Avoid homuncular fallacy (presupposition of phenomena you‟re trying to explain) -We must attempt to explain in terms not of the mind 3. Turing -Mechanization of formalisms empirical testing (computer as metaphor teaches us about method) -To make mind artificial intelligence -Naturalistic Imperative in Cognition TEXTBOOK Chapter 1: Categorization -Category: usually refers to a group of objects in the world whereas Concept: refers to a mental representation of such a group Functions of Categorization -Primary means for coding experience (Working Memory cannot process too much information  Categorization reduces the demands of cognitive processes) -Concepts may vary in the extent to which they‟re used as codes (Ex. Taxonomy) Inductive Inferences -Inductive: inference considered improbable that new belief is false if old one is true -Categorization infers invisible properties from visible ones, mentally -Membership is often used to justify inference of invisible properties when in same group Ex. Flamingo vs. Bat Aortas -Most support for inductive inferences comes from basic & subordinate categories -People are more likely to make I.Is about natural-kind categories, with regards to invisible properties, than artefacts -Inductive potential differs between classes of categories Similarity Category Members -Categorization is accomplished by: those aspects that are relatively common among members of a category are incorporated into the concept representing that category People decide whether something belongs to a category based in whether or not its representation is sufficiently similar to the concept of interest Geometric Approach 1. Minimality: same distance for all items 2. Symmetry: distance between items is same no matter what 3. Triangular inequality: shortest distance between two points is a straight line -Geometrical approach is successful at representing perceptual objects, but less so with conceptual items Problems with GA: -Minimality: the more we know about an item, the more similar it is judged to itself -Symmetry: an unfamiliar category is judged more similar to a familiar prominent category, but not the other way around -Triangular: implies A&C cannot be very dissimilar (Ex. Countries) -“Nearest neighbour problem”: use of low dimentionality to represent categories metrically Featural Approach -Item represented as a set of discrete features and the similarity between two items is assumed to be an increasing function of the features they have in common and a decreasing function of the features on which they differ -Tversky‟s model: -“Product-rule” model -Contrast model does not tell us features of objects (must be determined) -Contrast model does not offer theory of function „f‟ that measures the salience of each set of features -CM says little about algorithms used to implement computation LECTURE NOTES Analysis of Everyday Cognition Categorization -Analyze, formalize, mechanize -Categorization: a class of objects that we sense belong together -Categorization affords the coding of experience -Optimizes ability to deal with the world -Bad for noticing individuals (stereotyping) -Allows abstracting useful informationinductive inferences -Inductive generalizing allows ability to learn about things in an indirect manner -Improves communication (with self and other people) Allows use of common nouns (Ex. yelling the word “fire!” -Categorization seems like a basic process grouping by similarity and dissimilarity -The obvious common sense needs explanation “One of these things is not like the other, but all of them are kinda the same”- Sesame Street Similarity -Nelson Goodman (1972): similarity isn‟t property of the world -There are indefinitely large similarities and dissimilarities -Note relevance (cognition isn‟t only concerned with the truth) Barsalou -Categorization according to context: blend of categorization and perception -Why selective attention? We select or ignore/override certain features -Categorization by concept of object to us -If concepts are used to explain categorization, concepts can‟t be explained by categorization The Resemblance Theory Defending it: Smith (author) Criticizer: Lance Rips (named the theory) Main Idea: -Somehow, similaritycategorization 1. Theory must explain similarity 2. Theory must prove causality proposed -Seems that the opposite is the case (similaritycategorization) Ex. Snakes as dangerous -Think about cultural effects Tversky‟s formula: -An external homunculus can occur with contradictions to the formula -Taking the design stance: give machine abilities to perform calculation of formula (complex cognition) -Presupposition of phenomena formula is trying to explain Lance Rips‟ Counter-Argument: -Showed double-dissociation A B -Alteration of A with no significant difference shown in B and vice versa -So, two things are probably not significantly causally related If shown,  A cannot serve as an explanation to B Smith Responds by making a distinction: *Note that distinction must be believable beyond defending theory 1. Similarity-Based Categorization 2. Knowledge-Based Categorization -Features -Reasoning/inference (lay theories: implicit, non academic theories -Visible features (perceptive) -Invisible features -Direct -Indirect -Start with SBC -Built out of SBC -Must distinguish between types of features for both categorizations (visible vs. non) -Concrete example is perceptual features for bird (same aorta as other bird and not bat) -We think we see things, but are often concluding them -Distinction without significant difference -Require independent difference -Primitive features assembled to create definition (moved into concepts) -We make definitions (move away from Resemblance Theory) Classical Theory of Concepts -Essence: Individually necessary and jointly sufficient features (Ex. Triangles) 1. Tells you why you make transitive inferences (properties we‟re supposed to pay attention to 2. Explains how we generate concepts with presumably finite resources Ex. Songs compositionality (humans‟ concepts within concepts) 3. How we can have so many thoughts without pre-wiring then into the brain -Unfortunately, it‟s false -How people think concepts work? TEXTBOOK CONTD… Similarity and Verbal Categorization Typicality Effects: -Ratings predict how efficiently people can categorize various instances in a verbal- categorization task -Speed of response correlated with typicality 1. The representations of the target category and test item are permanently stored in LTM, but once activated they also become part of WM, where all subsequent processing ensues 2. An item will be categorized as an instance of a category if, and only if, its representation exceeds some critical level of similarity to the concepts associated with that category 3. The time needed to determine that and item exceeds this critical level of similarity is shorter; the more the item is similar to the concept.  Spreading activation: When an item representation and concept are activated, activation spreads to features associated with them, the activation being subdivided among features of the source. Once intersection occurs in activation (common features appearing), the process terminates when the number of intersections is high enough -In a case where several concepts are considered, the target one is contrasted with all the others Similarity and Visual Categorization -Study done yielded similar results to that of VC (typicality and speed correlation) Importance of Shape: -Featural-Similarity Model cannot be used because prototypical features are most often not represented in images Instead, we rely on shape to categorize -Study done by Biederman and Ju (1988): detail of image did not alter speed of categorization Typicality as Shape Similarity Adapted Model of Categorization: 1. People form a mental representation of the shape of the test object, and determine its similarity to the shape information in the target object 2. Only if the similarity in shape exceeds some criterion, do people decide that the test item is a member of the target category 3. The more typical of its category an object is rated, the more similar its shape is to those of other category members, and to the shape information of any prototype of the category 4. There more similar an instance is to a category, the faster and more accurately it can be categorized A concept of a category must include: prototypical features in connection with verbal categorization and detailed shape representation. Breakdowns of Verbal and Visual Categorization Agnosias: brain damage causing breakdown of categorization (often damage to temporal lobe) Category-Specific Deficits -Selective deficit ( Ex. Prosopagnosia: inability to categorize faces) Verbal categorization: artefact description was fine, but that of animals was not clear deficit in visual categorization (same results in that test) -Patients showed deficits in both areas & either  Is there a correlation between deficits? 2 Hypotheses about Category-Specific Deficits: Structural-Similarity Hypothesis: instances of superordinate, natural-kind categories seem far more similar to each other than instances of superordinate artefact categories. 1. Categorization considers first shape 2. If shape representation is associated with more than one object (as is likely with natural kind), shape won‟t provide unique categorization of test object 3. Additional processing needed (which is vulnerable to effects of brain damage)  Categories with similar memories are associated with larger deficits in brain- damaged patients Perceptual Functional Hypothesis: deficit at prototypical features 1. Categorizing objects, particularly verbal ones, depends on prototypical features- some of which refer to perceptual aspects and some to functional objects 2. Natural kinds: mainly perceptual, whereas artefacts: functional and perceptual 3. Perceptual features are particularly vulnerable to the effects of brain damage  Significance of distinguishing between two types of prototypical features -Neurobiological basis of categorization Visual Categorization: lesions in inferior part of temporal lobes Verbal Categorization: lesions in inferior part of temporal lobes and in frontal lobes Beyond Similarity -Theoretical features: constitute something like a lay theory of the object in question -Similarity in categorization can go opposite directions (bird that looks like an insect)  Only appears to influence categorization in special circumstances ^ Other Issues -Exemplar representation and right similarity algorithm works better to categorize than prototype We relate experiences and encounters to things in our past experiences and encounters as opposed to prototypical situations (examples) Does a good job of predicting subjects‟ categorization of novel items LECTURE NOTES CONT… -Concept: a set of necessary conditions for falling under the rule of the concept Classical Theory (not disputed until 1980s) -Concepts as mental definitions 1. Meaning of a concept can be captured by conjunctive list of features 2. Automatic or primitive features  Primitive building blocks compositionalityproductivity 3. Each feature is individually necessary and the set is jointly sufficient 4. What is a member of the category for the concept is clearly
More Less

Related notes for COG250Y1

Log In


Join OneClass

Access over 10 million pages of study
documents for 1.3 million courses.

Sign up

Join to view


By registering, I agree to the Terms and Privacy Policies
Already have an account?
Just a few more details

So we can recommend you notes for your school.

Reset Password

Please enter below the email address you registered with and we will send you a link to reset your password.

Add your courses

Get notes from the top students in your class.