2. Beyond the Classic Approach copy.docx

CRITICAL READING: CORNELL NOTES Beyond the Classic Approach Name: Date: 16 August 2023 Section: Lecture 2 Period: Questions/Main Ideas/Vocabulary Notes/Answers/Definitions/Examples/Sentences Concepts & Categories Categories are classes of objects in the world and concepts are ment...

CRITICAL READING: CORNELL NOTES Beyond the Classic Approach Name: Date: 16 August 2023 Section: Lecture 2 Period: Questions/Main Ideas/Vocabulary Notes/Answers/Definitions/Examples/Sentences Concepts & Categories Categories are classes of objects in the world and concepts are mental representations of the categories. Graded Category Structure Typicality ratings suggest that categories have graded structure. Generation frequency data, categorisation decisions and latencies also suggest that categories have graded structure. This graded structure is also well described by the Family Resemblance measure: a weighted measure of featural commonality amongst category members. The more similar an exemplar is to other category members, the higher its family resemblance, and the more typical it is of its category. Family Resemblance Model These are Blargs. They can be described by the following features: Body shape, body colour, antennae, number of legs. We can use this information to calculate their family resemblance. Rosch et. al Found a strong correlation between typicality and family resemblance for natural categories such as fruits, vegetables, mammals, etc. In other words, the graded typicality structure that we see in the categories is well described by the featural overlap of the category members. Category members that share a high number of features with other members of the category tend to be rated as being highly typical. Hampton (1979) Implemented a similar model: The Polymorphus Concept Model. In this case, the representativeness of an exemplar is a function of the degree of overlap between the features associated with the exemplar and the features associated with the category. These ‘representativeness’ ratings were found to be good predictors of the typicality ratings and generation frequencies of the category members. Graded Category Structure The family resemblance model and polymorphous concept model both use featural overlap to measure graded category structure. Typicality, generations frequency, these also seem to reflect this graded structure. Familiarity? Typicality? Ashcraft (1978): The more familiar a participant was with a given category member, the more information that they would have access to regarding that member. McClosky (1980) suggested that familiarity is a confounding variable. The correlation between typicality and categorisation response time dropped from r = -0.46 to -0.36 when familiarity was partialled out. Hampton (1983) Collected data for 12 categories. Two measures associated with graded category structure: (typicality and generation/associative frequency), plus familiarity. There were high inter-correlations between these variables. When Hampton ran partial correlations holding familiarity constant, the typicality/generation frequency relationship held. In other words, there appears to be shared variance between typicality and familiarity, but familiarity didn’t contribute to graded category structure. Other related variables: Age of acquisition (the estimated age at which a word is learned). Word frequency (the frequency with which a word is found in text corpora). Imageability (the ease with which an object can be pictured in your mind). Evidence suggests that none of these variables contribute to category structure. Similarity & Graded Category Structure Grade category structure is due to the similarity relations between category members. Typicality, generation frequency, response time, family resemblance. Each of these variables reflect the similarity relations between category members. Family resemblance is another way of saying similarity. Categories are bound together by the degree to which they are similar to each other, and dissimilar to members of other categories. Similarity The more similar an exemplar is to other category members, the higher its family resemblance, and the more typical it is of its category. Similarity data can be collected via a number of different procedures: Pair-wise or triad similarity ratings. Card-sorting tasks. Feature by exemplar matrices: vector correlations, feature distances and contrast model. It is important to note that there is nothing special about the modality of the similarity representation. MDS representation are readily interpretable in a visual space. They are also quite useful in a mathematical sense – many cognitive models rely upon spatial representations. There are other algorithms that have sometime been used to represent the similarity relations of conceptual data. Additive trees represent category members as terminal nodes in a tree, such that the similarity of two group members is the length of the branches separating them in the tree. Additive clustering represents concepts as weighted clusters, such that the similarity of two category members is modelled as the sum of the weights of their common clusters. Similarity is the glue that holds concepts together. The Contrast Model According to the contrast model, the similarity between the exemplars i and j can be calculated from the features common to the two exemplars, the features of i that aren’t present in j, and the features of j that aren’t present in i. The family resemblance model and the polymorphous concept model are both special cases of the contrast model. Using Similarity to Visualise Semantic Structure Multi-dimensional scaling (MDS) provides a visual representation of the similarity relations between different objects or entities. Converts similarities into distances: the more similar two items are, the closer they will be located. The more dissimilar two items are, the further away they will be located. This is actually a non-trivial problem (N x N-1 distances to fit). MDS The spatial representations generated using MDS appear to reflect the structure of the categories that are being represented. We can see this in perceptual data. We can also see it in conceptual data. In many cases, the spatial structure of the categories is readily interpretable. Sometimes the structure isn’t so easily interpretable, particularly when the optimal dimensionality of the representation is greater than 3. In perceptual data, we can often see that the spatial representations reflect perceptual qualities that weren’t explicitly referred to in the similarity data collection process. In conceptual data, we often find that the spatial representations reflect measures of graded category structure such as typicality, generation frequency, etc. Category Structure Typicality = distance from category centroid. Same for other measures of graded category structure (generation frequency, categorisation response time). But not for non-structural measures such as familiarity, age of acquisition, word frequency, etc. Typicality reflects the similarity structure of categories. MDS of Similarity Data MDS of similarity data can also tell us about differences in the way that different groups of people mentally represent different categories. For example, we might imagine that the internal representations of experts and novices might be different. Criticisms of Similarity Similarity isn’t flexible enough. What about theories/rules? Surely our semantic representations are based upon data that is far richer than simple similarity? Similarity is too flexible. By explaining everything, it explains nothing. These are both good points, but they don’t negate the fact that the vast majority of the variance in empirical conceptual/categorisation data can be accounted for by similarity.