Chapter 10 Concepts and General Knowledge - PowerPoint Presentation

Chapter 10 Concepts and General KnowledgeSemantic Memory Stores Chapter 10 Knowledge

Concepts provide a kind of mental shorthand, economizing cognitive efforts. Building Blocks of Thought Concepts— a mental representation representing a category of objects: stored in long-Term MemoryCategory – a set or class of objects that belong together (e.g., groupings of similar objects, events, ideas, or people) Chapter 7 General Knowledge

Classical Approach: Concepts Concepts viewed as definitions with lists of necessary (properties that instances of the concept MUST have) and sufficient (a set of properties that fully define all members of the concept. Chapter 7 General Knowledge

Chapter 7 General Knowledge Defining the properties or features on a Concept Formal Concept – clear cut defining rules

Chapter 7 General Knowledge “Real-World” or Natural Concepts– Fuzzy defining rules

Classical Approach Assumes: Sharp category boundary (in or out) Equality of members Representation of category is list of necessary and sufficient featuresIf an entity meets the conditions it is a member If an entity is a member we know it meets the conditions

Ludwig Wittgenstein (1953) Used examples such as triangle and square (text example) and friends and games(text example) to demonstrate the amazing ability of humans to conceptualize thoughts that they could not define. Chapter 7 General Knowledge

While there are a small set of formals concepts such as scientific categories and geometric shapes, there are vastly more concepts that humans are able to understand that are “real-world”. There is no set of necessary and sufficing properties that define them. Instead they are defined by a set of properties that are likely to be true, but not all members of the concept need to have each property, nor is there a set of properties that fully define the concept. Chapter 7 General Knowledge

Family Resemblance ( Wittgenstein) No one feature that all have in common…. … Yet all are in some way similar to the other category members Variation within categories allowed

Family Resemblance Assumes: No strict “definition” of what’s in/out based on individual features Membership based on similarity Some members can be “better” examples than others Problem: If category membership is based on similarity, the question remains; similar to what?

McCloskey & Glucksberg (1978) McCloskey & Glucksberg – subjects disagree about whether atypical items belong in a category: 30/30 apple is a fruit, chicken is not a fruit 16/30 pumpkin is a fruit Subjects change their minds when tested later.Conclusion: Category Boundaries are not always clear-cut.

Chapter 7 General Knowledge Is X a member of the category Y? Not all members of a category are equally as good members of those categories. Some members are more typical than others. We are able to more quickly respond to typical than to atypical instances. This is called the typicality effect. Therefore, membership is not equal.

Rosch and Mervis (1975) Participants got category names ( bird ) and lists of 50 members ( robin, canary, ostrich, penguin, sparrow…)Provided rating on how well the item represented the category (typicality ratings) Results: Much agreement on ratings between participants

Rosch (1975) Very Good Poor CATEGORY: BIRDS Bat Penguin Owl Sparrow Very Good Poor CATEGORY: FURNITURE Telephone Mirror China Closet Chair, Sofa

Rosch & Mervis (1975) continuedNew participants list attributes (properties) of each of the instances. Some attributes were listed more frequently than others. e.g., can fly vs. lives on a farm Results: Typical instances had more frequently listed features than did less typical instances. Chapter 7 General Knowledge

Press a button to answer TRUE or FALSE to the following statements Question Response A canary is a bird A ostrich is a bird TRUE TRUE fast slow Typicality effect Rips, Shoben & Smith, 1973 Verification Task

Other Support Mervis, Catlin & Rosch, (1976)When asked to list instances of a category, typical instances are generally listed first, indicating that they come more quickly to mind. Chapter 7 General Knowledge

Typicality and Language Typical instances are acquired earlier in life than atypical instances. When mentioning two category members together, the more typical member generally is mentioned first. Chapter 7 General Knowledge

Semantic Dementia (Fronto-Temporal Degeneration) A syndrome of progressive deterioration in semantic memory (knowledge of objects, people, concepts and words).  can still speak fluently (good grammar and pronunciation) Lose of words for certain items and the knowledge of the meaning of the word. For example, someone may not only forget the word hippopotamus when shown a picture, but also loses all the knowledge they once had about this (e.g. that it is an African animal that lives in rivers). Chapter 7 General Knowledge

may also have difficulty recognizing what things are. Effects all modalities (sound, taste, empathy)Unlike Alzheimer's disease, memory for day-to-day events may be good. Visuo-spatial skills intact. At later stages, personality is often affected. Deterioration of left anterior temporal lobes. Chapter 7 General Knowledge

Chapter 7 General Knowledge Animal copies by patient B.F. (shown on right) demonstrate retention of visuo-perceptual ability in SD patients. Naming tests highlight their semantic deficits. When asked to name these animals, B.F. called the cow a ‘dog’ and the elephant a ‘cow’.

Chapter 7 General Knowledge Drawings by patient G.W. show semantic deficits in spontaneous recall. Distinguishing features are absent: the fish lacks fins, the first bird lacks wings and the elephant lacks a trunk.

Alternative Approaches to Concepts Prototypes Exemplars World Knowledge Chapter 7 General Knowledge

Prototype Approach a.k.a. Family Resemblance or Probabilistic approach)Eleanor Rosch A Prototype is an abstract, typical representation of the category members. You can think of it as the weighted average of the important features of members of the category. Chapter 7 General Knowledge

Prototype Approach The guy in the center is closest to the prototype ( highest prototypically ) but he isn’t the prototype Prototype isn’t here Prototype is abstract

Prototype Approach Features: Glasses (yes/no) Hair (dark/light) Nose (big/small) Ears (big/small) Mustache (yes/no)Prototype 2/3 glasses, 7/9 light hair, 7/9 big nose, 7/9 big ears, 5/9 mustache Center guy has the highest prototypicality

Chapter 7 General Knowledge Probabilistic (not all or none). Prototypes include characteristic features that are usually present, not only necessary or sufficient features. Objects that do not share all the characteristics of the prototype are still members of the category but not prototypical ones.

Prototype Theory Rosch (1970s) suggested that when people categorize items they match them against a "prototype" which contains the most representative features inside the category. Chapter 7 General Knowledge .

Exemplar ApproachMedin & Schaffer Exemplars are specific examples Examples of category members are saved in memory (typical as well as atypical) Potential members compared to all exemplarsThose with high family resemblance are like more of the exemplars May be more useful for smaller categories (US presidents, very tall mountains)

Novick (2003) – Exemplars Change with ExperienceExtensive media coverage following 9/11 effected the typicality of airplane as a response to the category of vehicle. Airplane was judged to be a more typical vehicle for 1 month following the attacks, relative to a baseline calculated from data collected yearly for 5 years preceding the attacks. By 4.5 months, however, typicality was back to baseline. Chapter 7 General Knowledge

Evidence supporting Exemplars Much of the evidence that supports the prototype model also predicts the exemplar model. Chapter 7 General Knowledge

Exemplars or Prototypes Brooks & Allen (1991) Asked subjects to discriminate between two animals ( Diggers and Builders)Two kinds of animals could be distinguished in two different ways (“two-out-of-three” rule). 3 Relevant Features (leg length, angularity of body, spotted/not spotted).2 non-relevant features (number of feet and length of neck).Chapter 7 General Knowledge

Half of the participants explicitly told rule, others not told. Classification rule: 2 of (long legs, angular body, spots) => Builder 2 of (short legs, curved body, not spots) => Digger Example of Builder Example of Digger Chapter 7 General Knowledge

Learning Phase :Rule (Prototype) Condition: Participants explicitly told the rule and used it to categorized 8 exemplars. Memory ( Exemplar ) Condition: Shown 8 exemplars and told which were diggers and which builders without being told the rule. Chapter 7 General Knowledge

Generation of new items for Test phase. New items that varied from the learning phase exemplars were generated by manipulating relevant or non-relevant features. Chapter 7 General Knowledge

Good Matches : Changed a feature that did not change the category. In the example below shortening the neck does not change the category. Chapter 7 General Knowledge Original New

Bad Matches : change a relevant feature so that the new item is now a member of the other category. Adding spots Chapter 7 General Knowledge

Results Participants made errors that indicated they were relying on similarity to known exemplars rather than on the rule (prototypes). Even when a rule is known and easy to explicitly state, past examples can override application of the rule!Chapter 7 General Knowledge

A Caveat Evidence shows that we use both prototypes and exemplars depending on context. Natural categories – we may have richer prototypes and more complex rules, therefore we rely on prototypes. For Artificial categories we rely on exemplars.Chapter 7 General Knowledge

World Knowledge Approach When we do “real-world” categorization tasks we don’t just rely on prototypes or remembered exemplars, we also call on our world knowledge which included richer conceptual representations (uses for, causal and contextual information). Chapter 7 General Knowledge

Example: Goal-Derived Categories Goal-derived categories are those that satisfy some goal (e.g., birthday gifts that make the recipient happy or what you would take with you if your fire alarm went off). Family resemblance scores and prototypes are of little value for these categories since membership is based on goal satisfaction rather than sharing some particular attribute or set of attributes. Chapter 7 General Knowledge

The Role of Knowledge in Concept Learning Lin and Murphy (1997) People learned about objects used in foreign countries. One group learned that the tuk is used for hunting. The hunter holds the handle (3) with one hand behind the hand guard (2) and pulls the end of the rope (4) while slipping the noose (1) around the animal’s neck.

Lin and Murphy (1997) The other group was told that the tuk was a fertilizing tool. Liquid fertilizer is held in the tank(2), the knob (3) is turned to allow it to come through the outlet pipe (4). The loop is used to hang the tuk up .Chapter 7 General Knowledge

After learning a variety of concepts, items were then presenting missing one or more of the parts in the original object. The missing part really had a big effect on the participants’ classification of the object as a tuk. Those learning that Part 1 was a noose were far less likely than those learning that Part 1 was a loop hanger to classify the object missing Part 1 as a tuk, presumably because the missing part was more central to its function.Chapter 10 – 1 Concepts and Knowledge

Tomato: Fruit of Vegetable? Depends on your reason for categorizing! Definitions If you are a gardener, it is a fruit. In the kitchen, it is a vegetable. Chapter 7 General Knowledge