Chapter 9 Conceptual Knowledge PDF

Lecture 9 Chapter 9 Conceptual knowledge key word: Conceptual knowledge • knowledge that enable us to recognise objects and events and to make inferences about their properties key word: Concept - Mental representation: the knowledge exist in the form of concepts 1. Meaning of objects, events and...

Lecture 9 Chapter 9 Conceptual knowledge key word: Conceptual knowledge • knowledge that enable us to recognise objects and events and to make inferences about their properties key word: Concept - Mental representation: the knowledge exist in the form of concepts 1. Meaning of objects, events and abstract ideas 2. Mental representation of a class or individual include visual features, references to touch, sound, movement and habits e key word: categorisation • The process by which things are placed in groups • A category includes all possible examples of a particular concept • “pointer to knowledge” About meaning (semantic) • provide a wealth of general information of an item • identify the special characteristic of a particular item 9 . freent ↳egort Banana ↓ Example e apple e.g. an apple is an example of ‘banana’ category although an apple is round, it won’t be in the group of a ball Why categories are useful provide information about its expected behaviour and how to respond to it Concept: provide the rules for creating categories . Why definitions don’t work for categories? key word: Definitional approach to categorisation • decide whether something is a member of a category by determining whether a particular object meets the definition of that category • category works well on simple object e.g. geometric objects • category don’t work well on natural objects e.g. bird, trees, vehicles and bottles = membership of a category seems to be gradual, in terms of more-and-less , rather than allor- nothing *E Y key words: Family resemblance Wittgenstein • Deal with the problem that definitions often do not include all members of a category • ideas that things in a particular category resemble one another in a number of ways • no definite criteria that every member of a category must meet • allow some variation within a category e.g. Chair may be in different sizes and made of different materials . Every chair does seem resemble in some way = categorisation is based on determining how similar an object is to some standard representation of a category The prototype approach: finding the average case key word: prototype approach • membership in a category is derminated by comparing the object to a prototype Key word: prototype • ‘typical’ member of the category • represent the category • an average of members of a category that are commonly experienced e.g. a prototype for the category ‘bird’ might be based on some of the birds you frequently see, but doesn’t look exactly like any of them = prototype is NOT an actual member of the category but is an average representation of the most common members of a category Variation in typically key word: typicality • ** E #A High typicality: a category member closely resembles the category prototype e.g. sparrows, chair (popular examples) • Low typically: a category member does not closely resemble the category prototype e.g. penquine, owls = these variation represent differences in typicality Rosch 1975 extend to purpose: to quantify the level of typicality for a high number of category members, spanning several categories through subjective rating Procedure: 1. Ppt were asked to rate the extent to each member represented the category title Prototypical objects have high family resemblance the relationship between Prototype and family resemblance Rosch and Mervis 1975 hypothesis: high family resemblance overlapping attributes as measured with the method would correlate with the proto-typicality of the objects Purpose: to examine whether more typical members of a category share more attributes with each other compared to less typical members procedure: 1. design many of the same characteristics to chair and sofa E.g. share the fact that they have legs, arm and backs you can sit on • when item’s characteristics have a large amount of overlap with the characteristics of many other items in a category = this mean family resemblance of these items is high e.g. mirror and telephone • little overlap with other members of a category means that the family resemblance is low result: - correlation is found between family resemblance and typicality varied between 0.84 and 0.94 for six different categories. = strongly confirmed that the higher the family resemblance the more prototypical an object is for a certain category. The Prototype Approach A key word: Typicality effect typical • the ability to judge highly prototypical objects more rapidly • emphasises the idea that not all members of a category have equal status ability judge Sentence verification technique Smith 1974 purpose: use the sentence verification technique to determine how rapidly people could answer questions about an object’s category procedure: 1. An apple is a fruit (yes/ no) 2. A pomegranate is a fruit (yes/ no) - reaction time is the measure of interest which in this context can tell us about something how easily certain objects, categories and relation between the two are accessed in our mind objects --> relation categories to The prototype approach Prototypical objects are named first • when ppt are asked to list as many objects in a category as possible, ppt tend to list the most prototypical members of the category first • vary across culture and age groups Prototypical objects are affected more by priming • priming occurs when presentation of one stimulus facilitates the response to another stimulus that usually follows closely in time Rosh 1975 purpose: demonstrate the prototypical members of a category are more affected by a priming stimulus than are non-prototypical members procedure: 1. ppt first heard then prime, which was the name of a colour ‘green’ 2. two seconds later: ppt saw a pair of colours side by side and incited by pressing a key asap -> whether the two colours were the same or different result: • ppt responded fastest for ‘good’ green = the principle of priming is that the prime will facilitate thee ppt’s response to a stimulus if it contains some of the information needed to respond to the stimulus = ppt created the images in their mind of colour prototypes in response to the colour names The exemplar approach key word: exemplar approach to categorisation • determine whether an object is similar to other objects • the standard for the exemplar approach involves many examples key word: exemplars • actual members of the category that a person has encountered in the past • if the person encountered sparrows in the past, each of these would be an exemplar of the category ‘bird’ • Rosh experiment’s result support prototype and exemplar approach, can explain the typicality effect = the reaction times on the sentence verification task are faster for good examples of a category than the poor examples e.g. sparrow is similar to many birds, so ppt classified faster than a penguin Which approach do we use? • exemplars may work best for small categories • prototypes may work best for larger categories Hierarchical organisation key words: Hierarchical organisation • a large, more general categories are divided into smaller, more specific categories, creating a number of levels of categories What’s special about basic level categories? three levels of categories: 1. Superordinate level = global level e.g. furniture ↓ loss feature 2. Basic level e.g. table 3. subordinate level = specific level e.g. kitchen table Again features Rosh’s experiment procedure: 1. ppt were given the same task, list a few features that were common to all furniture, but many features that were shared by all tables and by all kitchen tables 2. Ppt listed an average of 3 common features for the global level category ‘furniture’, 9 for the basic level of categories such as ‘table’ , for the specific level such as ‘kitchen table’ result: • going above basic level results in a large loss of information • going below basic level results in a little gain of information = Basic level is psychologically special Ve es Evidence that basic-level is special procedure: 1. ppt were asked to name these object, they named by basic level name e.g. guitar rather than electrical guitar (specific), musical instrument (global) its = Basic level categories were the most inclusive category for which a concrete image of the category as a whole can be formed to be the first categorisation made during perception of the environment Culture and categorisation how knowledge can affect categorisation Coley medin atran 1997 procedure: 1. ask a group of university undergraduates to name as specifically as possible 2. 44 different plants on a walk around campus. result: • 75% of the ppt’s responses contained label like ‘tree’ rather than ‘oak’ • People who grow plants use ‘specific' category instant ‘oak’ Tanaka and Taylor 1991 procedure: 1. ask bird experts and non- experts to name pictures of objects from many different categories (tools, clothing...) result: • experts responded to the pictures by specifying the birds species robin, sparrow, while non- experts responded to the pictures by saying birds. • people knowledge and properties of objects are both importance in categorisation Cognitive economy. Semantic networks Collins and Quillian 1969 procedure: ppt responded much quicker to sentences such as ‘a shark can move’ (highfrequent attribute) than to ‘a salmon has a mouth’ (low frequent attribute) result: • this reaction differences cannot be explained by the Rosh’s model since equal distances Ent #t category general ct specific need to be trialled from specific to global • category and concept is represented a node • concepts are linked • models for how concepts and properties are associated in the mind • Hierarchical model, more specific to more general ~ - concept E 7 "r ⑭ 2 3 slowest fastest TBP. P.275 key words: spreading activation - activity that spreads out along any link that is connected to an activated node e.g. moving through the network from ‘robin’ to ‘bird’ activates the node at ‘bird’ and the link we use to get from robin tabard -> indicate by the coloured arrow and spread to other nodes in the network Lexical Decision task semantic priming procedure: 1. the ppt’s task was to press asap the ‘test’ key when both items were words and ‘no’ key when at least one item in the pair was a non-word result: • pair 1 and 2 would require a no response • pair 3 and 4 would require a yes response (semantic related) key variable of this experiment: some trials the words were closely associated e.g. bread and wheat some trials were not associated e.g. chair and money = response faster when two words were associated = when retrieving one word from memory triggered a spread of activation to nearby locations in a network = more activation would spread to words that were related, the response to the related words was faster than the response to unrelated words Criticism of Collins and Quillian • cannot explain the typicality effect, reaction times for statements about an object are faster for more typical members of a category than for less typical members e.g. the statement ‘a canary is a bird’ is generally certified more quickly than ‘an ostrich is a bird’ • Cognitive economy e.g. people store specific properties of concepts rights at the node for the concept • some sentence-verification results are problematic for the model e.g a pic is a mammal, RT = 1,476 ms e.g. a pig is an animal, RT = 1,268ms object - - catagory pi The connectionist approach & Key word: Connectionism • an approach to creating computer models for representing concepts and their properties based on characteristics of the brain key word: Connectionist network • the circles are units • inspires by the neuron in brain key word: Input units • units activated by stimuli presented from the environment key word: hidden units Input -> axon -> Output - input units send signals to hidden units which send signals to output units key word: Connection weights - determine how signals sent from one unit either increase/ decrease the activity of the next unit - these weight correspond to what happen at a synapse that transmits signals from one neuron to another - some synapse transmit signal strongly than others-> high firing rate in the next neuron • lower connection weight: cause less excitation ① environment Stimuli Basic assumption: ② Strong/weak Signal firing How are concepts represented in a connectionist Old TBP. 258 • how different concepts and their properties can be represented in connectionist network Connectionism key word: back propagation 繁殖 • the process which error signals are sent back to the hidden and representation units to provide information about the connection weights should be change so that the correct property units will be activated • error signal: difference between the activity signal of each output unit and the correct activity Learning in a connectionist network • indicate the activation of the eight representation units in response to different inputs Procedure: 1. Set the connection weight so the activity was about the same in each unit 2. As learning progressed, each concept being presented one after one and the computer changing weight slightly after each trial in response to error signal (patterns are adjusted) Result: • trial 250 ‘Salmon’ and ‘canary’ begin to look different • trial 2500 easy to tell the difference between the patterns for ‘Salmon’ and ‘canary’ = connectionist networks are created by a learning process that shapes the networks so they are capable of handling a wide range of inputs The Connectionist approach characteristic: • information about each concept in the network is contained in the distributed pattern of activity across number of units 1. the operation of connectionist network is not totally disrupted by damage - key word: graceful degradation: disruption of performance occurs only gradually as parts of the system are damages e.g brain damage: damage to the brain causes only a particular loss of functioning 2. Connectionist networks can explain generalisation of learning - similar concepts have similar patterns, training a system to recognise the properties of one concept (‘canary’) - provide information about other, related concepts (‘robin’ and ’sparrow’) e.g. this is similar to the way we actually learn about concepts because learning about canaries enable us to predict properties of different types of birds we have never seen. Google and connections Categories information in single neuron; the brain the representation of concepts in the brain • recording the neurons ↓ ↑ Neuropsychology of categories key word: Category-specific knowledge impairment • an impairment in which they had lost the ability to identify one type of object but retained the ability to identify other types of objects • patient has trouble recognising object in one category • patient K.C and E.W. have difficulty naming animals • cannot name nonliving things, fruits and vegetables • cannot tell the difference between different animals, knowing the properties of animals I ! It = K.C and E.W. • had trouble naming different kinds of animals, answering questions about animals such as ‘Does a whale have legs? • they were able to answer similar questions about NON animals Sensory functional hypothesis Brain area specialised for specific concepts x EEY8Y 7574x semantic sensory ability , ** 24 d - • also impaired functional abilities IE # S-F hypothesis predict only impaired sensory ability things (Animals) Living Artifact -> using with -> or an motion , colour interacting object Ey x functional ability x sensory Sensory-functional hypothesis Multiple factors approach * ↓ e.g. musical instrument overlapped with both artefacts (involving performed actions) and animals (involved sound and motion) The case of artifacts: • musical instrument associated with specific actions (how to play them) • associated with sensory properties (their visual form and the sounds they create) -E ** 84 e.g. patients who have poor comprehension of smaller objects but better knowledge of larger artefacts such as vehicle Semantic I living things A (share many properties) X W artifact ↓ v mechanical Key word: crowding ↓ Key word: Semantic category approach occipital temporal ventral facial recognition C & domain : specific - same ⑲ share Similar features The embodied approach - Mirror neurons + 46 • premotor cortex: perception and action occurs =same brain area 1. When monkey observed experimenter grasping the food 2. When monkey grasped the food = neuron in premotor cortex fired Mirror neuron: the neuron responses to washing the experimenter is similar to the response when the money is performing the action itself