Models of Concepts*

Cohen, Benjamin; Murphy, Gregory L.

doi:10.1207/s15516709cog0801_2

Cited by 217 publications

(30 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These negative findings suggest that it is time to explore an approach to conceptual combination that starts with the prototype representations themselves. Such a shift of emphasis has previously been argued for by Cohen and Murphy (1984). Hampton (1973), andThagard (1983).…”

Section: Rationale For a Representational Approachmentioning

confidence: 56%

Conceptual Combination with Prototype Concepts*

Smith¹,

Osherson

1984

Cognitive Science

144

View full text Add to dashboard Cite

This paper deals with how people combine simple, prototype concepts into complex ones; e.g., how people combine the prototypes for brown and apple so they can determine the typicality of objects in the conjunction brown apple. We first consider a proposal from fuzzy‐set theory (Zadeh, 1965), namely, that the typicality of an object in a conjunction is equal to the minimum of that object's typicality in the constituents (e.g., an object's typicality as a brown apple cannot exceed its typicality as a brown or as an apple). We evaluated this “min rule” against the typicality ratings of naive subjects in two experiments. For each of numerous pictured objects, one group of subjects rated its typicality with respect to an adjective concept, a second group rated its typicality vis‐à‐vis a noun concept, and a third group rated its typicality with respect to the adjective‐noun conjunction. In both studies, most objects were rated as more typical of the conjunction than of the noun. These findings violate not only the min rule but also other simple rules for relating typicality in a conjunction to typicalities in the constituents. As an alternative to seeking such rules, we argue for an approach to conceptual combination that starts with the prototype representations themselves. We illustrate one version of this approach in some detail, and show how it accounts for the major findings of the present experiments.

show abstract

Section: Rationale For a Representational Approachmentioning

confidence: 56%

Conceptual Combination with Prototype Concepts*

Smith¹,

Osherson

1984

Cognitive Science

144

View full text Add to dashboard Cite

show abstract

“…In some [enterprises], a concept is regarded as a definition of a term, or a statement of the defining conditions for membership in the class designated by the term. In others, it is regarded as a representation of a chunk of knowledge that can be used to categorize and understand a domain of objects, events, or processes (Cohen and Murphy 1984).…”

Section: Objects Concepts Meaning and Representationmentioning

confidence: 99%

“…While the advantages of fuzzy concepts have usually been presented in terms of represeptation (Zadeh 1965;Smith and Medin 1981;Cohen and Murphy 1984), the advantages to learning are also important. Consider again the effect of functional form on incremental learning.…”

Section: Advantage Of Gradation To Learningmentioning

confidence: 99%

“…Any algorithm for deciding class membership must assume some fixed method or else the prototype must be augmented with some other information. The representational inadequacy of prototypes is discussed in Smith and Medin (1981) and Cohen and Murphy (1984).…”

Section: Protovpe Formmentioning

confidence: 99%

“…These ideas have often been overlooked in cognitive models. For example, Medin and Smith (198 1) and Cohen and Murphy ( 1984) identify probabilistic representations with prototypes and do not consider a representation like our graded form [. Until recently (Rendell 19836;Holte 1985;Flann and Dietterich 1986;Porter 1986), most research has ignored the notion that representations for learning may differ from those for task performance.…”

Section: Protovpe Formmentioning

confidence: 99%

See 2 more Smart Citations

Similarity‐based learning and its extensions

Rendell

1987

Computational Intelligence

View full text Add to dashboard Cite

This paper synthesizes a number of approaches to concept representation and learning in a multilayered model. The paper emphasizes what has been called similarity-based learning (SBL) from examples, although this review is extended to address wider issues. The paper pays particular attention to requirements for incremental and uncertain environments. and to interrelationships among concept purpose, concept representation, and concept learning.One goal of the paper is to unite some of the notions underlying recent research, in an attempt to construct a more complete and extensible framework. This framework is designed to capture representations and methods such as those based on hypothesis search and bias selection, and to extend the ideas for greater system capability. This leads to a specific perspective for multilayered learning which has several advantages, such as greater clarity, more uniform learning, and more powerful induction.The approach clarifies and unifies various aspects of the problem of concept learning. Some results*are (1) Various concept representations (such as logic, prototypes, and decision trees) are subsumed by a standard form which is well suited to learning, particularly in incremental and uncertain environments;(2) Concept leaming may be enhanced by exploiting a particular phenomenon in many spaces-this phenomenon is a certain kind of smoothness or regularity, one instance of which underlies the similarity in SBL systems; (3) The paper treats the phenomenon in a general way and applies it hierarchically.This has various advantages of uniformity. For example the model allows layered leaming algorithms for concept learning all .to be instantiations of one basic algorithm. A single kind of representation (an instantiation of the standard form) is prominent at each level. The combination of representation and algorithm allows fast, accurate, concise, and robust concept learning.Le pdsent article fait la synthese d'un certain nombre de mtthodes de repdsentation et d'apprentissage des concepts dans un modtle multicouche. L'auteur insiste sur ce qu'on a appelt I'apprentissage par dttection dc similarit& (ADS) en pdsentant des exemples; son point de vue s'ttend toutefois ti des sujets plus vastes. Il p e t e particulibrement attention aux exigences relatives aux environnements incdmentiels et incertains, ainsi qu'B la codlation entre le but, la rep6entation et I'apprentissage des concepts.Cet article vise notamment i unifier certaines des notions qui sont ti la base des recherches dcentes, afin que soit tdifik un cadre plus complet et plus extensible. Cette structure est cogue pour reproduire des repdsentations et des mtthodes, comme celles bastes sur la recherche d'hypothtses et la stlection de polarisations, et pour tlargir les id& ayant pour but d'accroitre le potentiel des systkmes. Cela conduit i envisager une perspective sptcifique au niveau de I'apprentissage rnulticouche, ce qui pnkente plusieurs avantages : plus grande pkcision, uniformisation de l'apprentissage et augmentation de la ...

show abstract

Revisiting attribute diagnosticity in the context of product typicality

Gray

1996

Psychol. Mark.

View full text Add to dashboard Cite

Models of Concepts*

Cited by 217 publications

References 40 publications

Conceptual Combination with Prototype Concepts*

Conceptual Combination with Prototype Concepts*

Similarity‐based learning and its extensions

Revisiting attribute diagnosticity in the context of product typicality

Contact Info

Product

Resources

About