Academic journal article Cognitie, Creier, Comportament

Inductive Categorization: A Methodology to Examine the Basis for Categorization and Induction in Infancy

Academic journal article Cognitie, Creier, Comportament

Inductive Categorization: A Methodology to Examine the Basis for Categorization and Induction in Infancy

Article excerpt


Three experiments with a methodology called inductive categorization examined 12- 16-, and 20-month-olds' categorization of animals and vehicles with and without functional parts as well as their inductive inferences about the motion properties of the objects in these classes. The experiments showed that infants at 12 months use object parts to categorize after a brief learning period, infants at 16 months attend spontaneously to object parts to categorize, and infants at 20 months use object parts and other features to categorize. The experiments also revealed that 12-montholds have little knowledge about the motion properties of objects, 16-month-olds have associated specific object parts with those properties, and 20-month-olds have generalized from object parts to other features. Taken together, these experiments provide support for the domain-general approach to early concept development, and they are the first to show a relationship between inductive inference and categorization in infancy.

KEYWORDS: infancy, induction, categorization, object features.


One of the key functions of concepts - mental representations of the object, entities, and events in the worlds-is to provide a basis for categorization and generalization. Only a fraction of the entities, objects, features, and events in the world can be experienced directly; therefore, we must rely on categorization and inductive inference to determine which things in the world belong together or are alike in some way and how to generalize a specific observation to other instances. For example, on encountering a novel mammal that barks and has fur one would conclude that it belongs to the category of "dog", that it is selfpropelled, moves towards goals, and chases squirrels. As adults, we seamlessly and effortlessly accomplish these tasks-that is, we have carved nature accurately at its joints - yet how we reach this representational end-state currently remains unknown. The experiments in this paper were designed to address key, as yet unanswered, questions relating to the development of object concepts; namely, (1) what is the effect of high and low within-category part similarity on infants' categorization and induction?; and (2) what is the relationship between categorization and induction in infancy and is it informative about the content of infants' developing representations?

The last 20 years has witnessed a proliferation of research on the early development of categorization. One of the most fruitful avenues of this research has relied on the sequential touching or object manipulation procedure, in which the systematicity of infants' spontaneous successive touches to objects are interpreted as indicative of categorization. Using this paradigm, it was found that infants at 18 and 20 months categorize animals and vehicles as different (e.g., animals vs. vehicles) but not basic-level contrasts within these domains (e.g., dogs and horses) (Mandler & Bauer, 1988; Mandler, Bauer, & McDonough, 1991). According to Mandler and colleagues, these and similar data support the view that infants' categorization and induction are not based on surface features (see also Mandler & McDonough, 1996, 1998). They argued that exemplars within a superordinate domain (e.g., animals) share few surface properties and therefore categorization of such perceptually diverse stimuli must be based on conceptual categorization (e.g., animacy, movement abilities, or class relations), with properties perceptible in the input playing only a secondary role. A corollary of this view is that infants possess specialized processes, innate modules, or skeletal principles that facilitate rapid conceptual understanding and allow infants within the first year of life to learn about objects' surface properties as well as those that are sporadically available in the perceptual input (e.g. movement) (Gelman, 1990; Leslie, 1995; Mandler, 1992). …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.