Learning Theory

Schemata are psychological constructs that have been proposed as a form of mental representation for some forms of complex knowledge.

Bartlett's Schema Theory

Schemata were initially introduced into psychology and education through the work of the British psychologist Sir Frederic Bartlett (1886–1969). In carrying out a series of studies on the recall of Native American folktales, Bartlett noticed that many of the recalls were not accurate, but involved the replacement of unfamiliar information with something more familiar. They also included many inferences that went beyond the information given in the original text. In order to account for these findings, Bartlett proposed that people have schemata, or unconscious mental structures, that represent an individual's generic knowledge about the world. It is through schemata that old knowledge influences new information.

For example, one of Bartlett's participants read the phrase "something black came out of his mouth" and later recalled it as "he foamed at the mouth." This finding could be accounted for by assuming that the input information was not consistent with any schema held by the participant, and so the original information was reconstructed in a form that was consistent with one of the participant's schemata. The schema construct was developed during the period when psychology was strongly influenced by behaviorist and associationistic approaches; because the schema construct was not compatible with these worldviews, it eventually faded from view.

Minsky's Frame Theory

In the 1970s, however, the schema construct was reintroduced into psychology though the work of the computer scientist Marvin Minsky. Minsky was attempting to develop machines that would display human-like abilities (e.g., to perceive and understand the world). In the course of trying to solve these difficult problems, he came across Bartlett's work. Minsky concluded that humans were using their stored knowledge about the world to carry out many of the processes that he was trying to emulate by machine, and he therefore needed to provide his machines with this type of knowledge if they were ever to achieve human-like abilities. Minsky developed the frame construct as a way to represent knowledge in machines. Minsky's frame proposal can be seen as essentially an elaboration and specification of the schema construct. He conceived of the frame knowledge as interacting with new specific information coming from the world. He proposed that fixed generic information be represented as a frame comprised of slots that accept a certain range of values. If the world did not provide a specific value for a particular slot, then it could be filled by a default value.

For example, consider the representation of a generic (typical) elementary school classroom. The frame for such a classroom includes certain information, such as that the room has walls, a ceiling, lights, and a door. The door can be thought of as a slot which accepts values such as wood door or metal door, but does not accept a value such as a door made of jello. If a person or a machine is trying to represent a particular elementary school classroom, the person or machine instantiates the generic frame with specific information from the particular classroom (e.g., it has a window on one wall, and the door is wooden with a small glass panel). If, for some reason, one does not actually observe the lights in the classroom, one can fill the lighting slot with the default assumption that they are fluorescent lights. This proposal gives a good account of a wide range of phenomena. It explains, for example, why one would be very surprised to walk into an elementary classroom and find that it did not have a ceiling, and it accounts for the fact that someone might recall that a certain classroom had fluorescent lights when it did not.

Modern Schema Theory

Minsky's work in computer science had a strong and immediate impact on psychology and education. In 1980 the cognitive psychologist David Rumelhart elaborated on Minsky's ideas and turned them into an explicitly psychological theory of the mental representation of complex knowledge. Roger Schank and Robert Abelson developed the script construct to deal with generic knowledge of sequences of actions. Schema theory provided explanations for many experiments already in the literature, and led to a very wide variety of new empirical studies. Providing a relevant schema improved comprehension and recall of opaquely written passages, and strong schemata were shown to lead to high rates of inferential errors in recall.

Broad versus Narrow Use of Schema

In retrospect, it is clear that there has been an ambiguity in schema theory between a narrow use and a broad use of the term schema. For example, in Rumelhart's classic 1980 paper, he defined a schema as "a data structure for representing the generic concepts stored in memory" (p. 34). Yet he went on to state that "there are schemata representing our knowledge about all concepts: those underlying objects, situations, events, sequences of events, actions and sequences of actions" (p. 34). Thus, schemata are frequently defined as the form of mental representation for generic knowledge, but are then used as the term for the representation of all knowledge.

There are severe problems with the use of the term schema to refer to all forms of complex knowledge. First, there is no need for a new technical term, since the ordinary term knowledge has this meaning. In addition, if schema theory is used to account for all knowledge, then it fails. A number of writers have pointed out that schema theory, as presently developed, cannot deal with those forms of knowledge that do not involve old generic information. Thus, schema theory provides an account for the knowledge in long-term memory that the state of Oklahoma is directly above the state of Texas. However, schema theory does not provide an account of the new representation one develops of a town as one travels through it for the first time.

Therefore it seems best to use the term schema in the narrower usage, as the form of mental representation used for generic knowledge. However, if one adopts the narrower usage one has to accept that schemata are only the appropriate representations for a subset of knowledge and that other forms of mental representation are needed for other forms of knowledge. For example, mental models are needed to represent specific nonschematic aspects of knowledge, such as the layout of an unfamiliar town, while naive theories or causal mental models are needed to represent knowledge of causal/mechanical phenomena.

Schema Theory in Education

Richard Anderson, an educational psychologist, played an important role in introducing schema theory to the educational community. In a 1977 paper Anderson pointed out that schemata provided a form of representation for complex knowledge and that the construct, for the first time, provided a principled account of how old knowledge might influence the acquisition of new knowledge. Schema theory was immediately applied to understanding the reading process, where it served as an important counterweight to purely bottom-up approaches to reading. The schema-theory approaches to reading emphasize that reading involves both the bottom-up information from the perceived letters coming into the eye and the use of top-down knowledge to construct a meaningful representation of the content of the text.

Broad versus Narrow Use of Schema in Education

The problem with the broad and narrow use of the term schema surfaced in education just as it had in cognitive psychology. For example, in Anderson's classic 1977 paper on schemata in education, he clearly takes the broad view. He attacks the narrow view and says that it is impossible "that people have stored a schema for every conceivable scene, event sequence, and message" (p. 421), and that "an adequate theory must explain how people cope with novelty" (p. 421). However in a paper written at roughly the same time (1978), Anderson states that "a schema represents generic knowledge" (p. 67), and he adopts the narrow view systematically throughout the paper. In a 1991 paper on terminology in education, Patricia Alexander, Diane Schallert, and Victoria Hare note that the systematic ambiguity between the narrow and broad views has made it very difficult to interpret a given writer's use of the term schema in the education literature.

Instructional Implications of Schema Theory

A number of writers have derived instructional proposals from schema theory. They have suggested that relevant knowledge should be activated before reading; that teachers should try to provide prerequisite knowledge; and that more attention should be given to teaching higher-order comprehension processes. Many of these proposals are not novel, but schema theory appears to provide a theoretical and empirical basis for instructional practices that some experienced teachers were already carrying out.

Impact of Schema Theory on Education

Schema theory has provided education with a way to think about the representation of some forms of complex knowledge. It has focused attention on the role old knowledge plays in acquiring new knowledge, and has emphasized the role of top-down, reader-based influences in the reading process.