к.ф.н., доцент Шингарева М.Ю., магистрант Мамбетова Г.Т.

Региональный социально-инновационный университет

To the Question of Frame structure

Frames provide the fundamental representation of knowledge in human cognition. In cognitive psychology, frames have received much attention in research on the essentially identical construct of schema [1]. Theorists who have attempted to articulate the structure of schemata have generally identified the same structural properties proposed for frames. However, much undesirable baggage has become associated with "schema." Psychologists frequently demonstrate the ubiquity of schemata in human knowledge. Yet, their studies rarely attempt to provide evidence for the structural characteristics of schemata proposed in more theoretical analyses (e.g., attribute-value sets, relations). As a result, "schema" is often criticized as being vague and unspecified. Moreover, "schema" has come to mean many different things to many different people. Most problematic is the frequent use of "schema" to mean a feature list prototype. Researchers sometimes assume that a schema is simply those features most common across a category's exemplars (e.g., Cohen, 1981; Markus, Smith, & Moreland, 1985; Posner & Keele, 1968). Because of these problems, we use "frame" to highlight the well-specified, structural properties common to formal analyses of frames and schemata.

Frames play a central role in constructing categories during planning [2]. We next examine three basic components of frames: attribute-value sets, structural invariants, and constraints. We assume that frames represent all types of categories, including categories for animates, objects, locations, physical events, mental events, and so forth. As we shall see, the representation of adjectives, adverbs, and quantifiers is feasible within the context of frames as well. Although these simplified examples keep presentation tractable, it is important to remember that constructing a complete conceptual frame for a single category is a challenging and sobering experience.

A cooccurring set of attributes constitutes the core of a frame. As a frame represents different exemplars, its attributes adopt different values. A fundamental task for frame theorists is to provide satisfactory definitions for attribute and value. We define an attribute as a concept that describes an aspect of at least some category members. For example, color describes an aspect of birds, and location describes an aspect of vacations. A concept is only an attribute when it describes an aspect of a larger whole. When people consider color in isolation (e.g., thinking about their favorite color), it is not an attribute but is simply a concept. Similarly, when people think about location in isolation (e.g., in geography), it is not an attribute. A concept is only an attribute when viewed as describing some aspect of a category's members. Color becomes an attribute when viewed as an aspect of bird, and location becomes an attribute when viewed as an aspect of vacation.

In this regard, the definition of attribute is extrinsic, depending on a concept's aspectual relation to a category.

By concept we mean the descriptive information that people represent cognitively for a category, including definitional information, prototypical information, functionally important information, and probably other types of information as well. In this regard, our use of concept vaguely resembles intension and sense. In general, we assume that frames represent all types of concepts, whether they are free-standing concepts or whether they are attributes.

What aspects of a category can be attributes? Clearly, this depends significantly on a category's ontological domain [3]. For physical objects, attributes are likely to include color, shape, and weight; whereas for events, attributes are likely to include location, time, and goal. Attributes are often parts of category members. As discussed by Chaffin and Winsto part is a highly polysemous relation. According to their analysis, part can refer to a physical part of an object (leg- chair), the material of an object (metal-ring), a member of a collection (flower- bouquet), an action in an activity (pitch-baseball), an object in an activity (food-eat), a location in an activity (destination-drive), and so forth. However, we assume that attributes can represent many other aspects of category members beside their parts. For example, attributes include evaluations (enjoyment-music), quantities (cardinality-family), costs (sacrifices-career), benefits (skills- education), and so forth.

The definition of value follows from the definition of attribute: Values are subordinate concepts of an attribute. Because values are subordinate concepts, they inherit information from their respective attribute concepts. In the frame for car, values of engine (e.g., four-cylinder) inherit properties of engine (e.g., consumes fuel, produces force). Values further inherit the extrinsic fact that they are an aspect of category members. Because engine is an aspect of car, its values are aspects of car as well. Values contain additional information not in their respective attributes, thereby making them more specific concepts. Four-cylinder and six-cylinder contain information that makes them more specific than engine and that differentiates them from each other.

Clearly, an infinite number of attributes could be constructed for a category [4]. In this regard, the human conceptual system is highly productive, although no person constructs all or even many of these potential attributes. Experience, goals, and intuitive theories play important roles in constraining attribute construction. If people experience different exemplars of a category, they may represent different attributes for it. For example, if cars have smog devices in one country but not in another, only citizens of the former country may typically represent smog device as an attribute of car. If people have different goals while interacting with exemplars, they may represent different attributes for them. For example, a wine connoisseur may represent wood used foraging as an attribute of wine, whereas someone who counsels alcoholics may not. If people have different intuitive theories about a category, they may represent different attributes for it. For example, people who know scientific theories of biology may represent genes as an attribute of animal, whereas people who do not know these theories may not.

Once particular attributes become represented for a category, they determine relevance. If two people represent a category with different attributes, they encode its exemplars differently. Different aspects of the exemplar are relevant, because the perceivers' respective frames orient perception to different information.

Clearly, the properties of frames are distinct and testable. In addition, these properties can probably be constrained further. For example, the capacity of working memory might constrain the number of core attributes in a frame to around five. Similarly, capacity and performance limits in human cognition might limit the recursive depth of frames and the length of constraint chains. In principle, developing a constrained and falsifiable theory of frames appears quite feasible. What remains is to implement such a theory and acquire rigorous empirical evidence for its components.

Literature

1. Barilett, F. C. (1932). Remembering: A study in experimental and social psychology. New York: Cambridge University Press.

2. Barsalou, L. W. (1991). Constructing categories to achieve goals. In G. H Bower (Ed ), The psychology of learning and motivation: Advances in research and theory (Vol. 27). San Diego: Academic Press.

3. Keil, F. C. (1981). Constraints on knowledge and cognitive development. Psychological Review. 88. 197-227.

4. Goodman. N. (1955). Fact, fiction, and forecast. Cambridge, MA: Harvard University Press.