Linear separability in classification learning.

Medin, Douglas L.; Schwanenflugel, Paula J.

doi:10.1037/0278-7393.7.5.355

Cited by 242 publications

(281 citation statements)

References 29 publications

Supporting

Mentioning

275

Contrasting

Unclassified

Order By: Relevance

“…In fact, a demonstration that natural concept categories are commonly not linearly separable would be excellent evidence against the similarity view. Although it has been shown that certain non-linearly separable categories are as easy (or difficult) to learn as linearly separable ones (Medin and Schwanenflugel, 1981), I am aware of no direct evidence of this kind.…”

Section: Discussionmentioning

confidence: 85%

Similarity-based categorization and fuzziness of natural categories

1998

View full text Add to dashboard Cite

The adequacy of similarity to prototype as an account of categorization in natural concepts was assessed by analyzing the monotonicity of the relation between typicality of an item in a category and the probability of a positive categorization response using data from McCloskey and Glucksberg (1978). The analysis revealed a strong underlying similarity-based threshold curve, with systematic deviations. Further data collection showed that deviations from the curve could be attributed to the effects of unfamiliarity and non-categorial associations on typicality judgments, as well as differences between the perceptual appearance of an item (which tended to boost typicality) and its underlying nature (which tended to boost categorization). The results are discussed in terms of the different presuppositions and task constraints involved in rating typicality as opposed to performing a categorization.

show abstract

Section: Discussionmentioning

confidence: 85%

Similarity-based categorization and fuzziness of natural categories

1998

View full text Add to dashboard Cite

show abstract

“…Although a lot of recent categorization research has focused on the need for models with more elaborate, mixed representations (Ashby, Alfonso-Reese, Turken, & Waldron, 1998;Erickson & Kruschke, 1998;Love, Medin, & Gureckis, 2004;, simple exemplar, prototype, and rule-based representations continue to be considered relevant, particularly in the context of research emphasizing different types of category learning and use (Markman & Ross, 2003;Yamauchi & Markman, 1998). Exemplar models have been widely shown to provide a better account of perceptual category learning than prototype or simple rule models (e.g., Medin & Schaffer, 1978;Medin & Schwanenflugel, 1981;Nosofsky, 1992). Evidence for simple rules continues to be found (e.g., Ashby et al, 1998;Erickson & Kruschke, 1998;Johansen & Palmeri, 2002;, but usually in conjunction with some other type of representation, such as exemplars or prototypes.…”

mentioning

confidence: 99%

Category Representation for Classification and Feature Inference.

Johansen

Kruschke

2005

Journal of Experimental Psychology: Learning, Memory, and Cogni

View full text Add to dashboard Cite

This research's purpose was to contrast the representations resulting from learning of the same categories by either classifying instances or inferring instance features. Prior inference learning research, particularly T. Yamauchi and A. B. Markman (1998), has suggested that feature inference learning fosters prototype representation, whereas classification learning encourages exemplar representation. Experiment 1 supported this hypothesis. Averaged and individual participant data from transfer after inference training were better fit by a prototype than by an exemplar model. However, Experiment 2, with contrasting inference learning conditions, indicated that the prototype model was mimicking a set of label-based bidirectional rules, as determined by the inference learning task demands in Experiment 1. Only the set of rules model accounted for all the inference learning conditions in these experiments.

show abstract

“…A question of crucial importance, in this regard, concerns the constraints by which natural concepts are characterized. It would be hard to believe that human culture has passed on categories with a structure that is not coordinated with the constraints of human information processing (Medin & Schwanenflugel, 1981;Sperber, 2000).…”

mentioning

confidence: 99%

Linear separability in superordinate natural language concepts

2004

View full text Add to dashboard Cite

Two experiments are reported in which linear separability was investigatedin superordinate natural language concept pairs (e.g., toiletry-sewing gear). Representations of the exemplars of semantically related concept pairs were derived in two to five dimensions using multidimensional scaling (MDS) of similarities based on possession of the concept features. Next, category membership, obtained from an exemplar generation study (in Experiment 1) and from a forced-choice classification task (in Experiment 2) was predicted from the coordinates of the MDS representation using log linear analysis. The results showed that all natural kind concept pairs were perfectly linearly separable, whereas artifact concept pairs showed several violations. Clear linear separability of natural language concept pairs is in line with independent cue models. The violations in the artifact pairs, however, yield clear evidence against the independent cue models.

show abstract

Linear separability in classification learning.

Cited by 242 publications

References 29 publications

Similarity-based categorization and fuzziness of natural categories

Similarity-based categorization and fuzziness of natural categories

Category Representation for Classification and Feature Inference.

Linear separability in superordinate natural language concepts

Contact Info

Product

Resources

About