Feedback in Learning

A BRIEF HISTORY OF FEEDBACK

In the early twentieth century, the term feedback referred to electrical systems used in audio broadcast. For example, feedback was brought to public attention in a notorious series of lawsuits over a patent for a circuit that would amplify radio signals by feeding back the signal into the circuit in rapid oscillations. Engineers came to use feedback to refer to products of a mechanical or electrical system that allowed the system to regulate itself. For example, if a heating system warms air temperature beyond a specified point, a thermostat can switch the system off and restart the heating system once the air temperature cools. Feedback became an explicit and crucial engineering concern as industry and the military sought to increasingly automate during World War II.

In 1948 Cybernetics: Control and Communication in the Animal and the Machine, by Norbert Wiener (1894–1964) made an enormously influential articulation of the role of feedback in the operation of all varieties of mechanical, biological, psychological, and social systems. Coining the term cybernetics from the Greek for steersman, Weiner conceived of a new science to investigate how feedback and information dissemination regulate systems. With ideas both persuasive and prescient, Weiner imagined how feedback systems could be linked to complex calculating machines to create mechanical brains that demonstrate artificial intelligence, the machine learning and action that is responsive to dynamic features of a changing environment.

Simultaneous with conceptualizations of feedback in engineering, behaviorism emerged as a psychological paradigm which emphasized the consequences of an organism's behavior as the primary determinant of future behavior. Simply put, behaviorism argued that certain behavioral consequences act as reinforcers, increasing the subsequent frequency of behaviors that produce them. A child might increase, for example, the frequency of behaviors that lead to acquiring candy. Some consequences act as punishment, dramatically reducing the frequency of a behavior that produces them. Without reinforcing or punishing consequences, a behavior extinguishes, diminishes to some baseline frequency.

However, behaviorism struggled to account for the operation of some behavioral consequences and learning phenomena. How might it explain behaviors that persist in spite of punishment or which seem indifferent to reinforcement? How might behaviorism account for one-time learning and sudden insight, early acquisition of the grammar of a language, creativity, and actions that seem guided by abstract concept, not perceptible stimuli? Such phenomena implied the operation of some mental apparatus that could perform interpretative transformations on stimuli and produce new capabilities that are not so directly derived from simple behavior-consequence contingencies.

The combined influence of the cybernetics paradigm, weaknesses in behaviorism's account of learning, and developments in computer technology made psychology and education ripe for a so-called cognitive revolution. A new paradigm of humans as processors of information, borrowed from computing, supplanted behaviorism. Perceptible stimuli were no longer only considered cues or consequences for behavior, but informative signals that could be interpreted by mental apparatus, stored in meaning-preserving ways in memory, and used to inform future goal-directed action. The consequences of that action could then be fed back into the interpretive mental processor.

CONTEMPORARY VIEWS OF INSTRUCTIONAL FEEDBACK

At the beginning of the twenty-first century, the information-processing paradigm still provides a standard conceptualization of instructional feedback. For example, for the influential instructional psychologist, Robert Gagné (1916–2002), learning is the encoding of information from short-term into long-term memory storage. In subsequent practice, newly stored knowledge is retrieved and applied. Feedback, the consequences of knowledge applications, re-enters the cognitive system, causing potential alterations in goals, learning processes, and stored knowledge.

Feedback's conceptualization has evolved in two significant ways. First, the source of feedback is no longer presumed to be a teacher or even external consequences. Butler and Winne, for example, emphasize the importance of internal sources of feedback generated through self-monitoring as a critical engine of self-regulated learning. In the absence of teacher-provided feedback, capable learners could interpret the consequences of their actions and cognitions in light of learning goals and thus provide feedback to themselves.

Second, some researchers distinguish among types of feedback. Bangert-Drowns, Kulik, Kulik, and Morgan suggested that feedback can provide information about knowledge retrieval and application, emotional and motivational states, and strategic management of learning. It could indicate whether a performance is correct, explain the nature of an error, provide prototypical responses, or simply display the consequences of actions. Each of these likely requires different features to be optimally effective. Hattie and Timperley, extending Kluger and DeNisi's suggestion that feedback loops can be hierarchically nested one in another, proposed four levels of feedback: about the quality of task performance, about the cognitive processes used to accomplish the task, about the ways in which a learner could better manage learning engagement, and about the self. Evidence suggests that these different kinds of feedback will be differentially effective according to different standards; for example, information about the self may not provide effective performance enhancements because it does not indicate specific ways in which knowledge or learning processes can be adjusted.

OPTIMIZING FEEDBACK FOR LEARNING

Feedback in learning is difficult to research. It has been found to inform learners about the following:

The accuracy or efficiency of knowledge and its retrieval or application

The quality of mental operations applied to information

The quality of psychomotor performance

The viability of a hypothesis or expectation

The emotional or motivational state of the learner

The efficiency or quality of learning processes employed in a learning task

The management of learning processes

Enduring qualities and goals of self

The nature of relationships with co-learners and teachers.

Different feedback features might be differentially effective for different learners and different aspects of learning. Even apparently straightforward feedback depends on the interpretation of a learner. A learner could interpret the response “incorrect” as indicating a need to revise knowledge, a personal rejection, the confirmation of a hypothesis, or a hint to work harder.

Despite these difficulties, some generalities about feedback effects are possible. Kulhavy suggested—and Bangert-Drowns, Kulik, Kulik, and Morgan supported— that if feedback is available before a learner actively attempts to perform a task (presearch availability), the learner may not be cognitively and motivationally prepared to effectively use the feedback. Kulhavy also suggested— and subsequent research corroborates—that response certitude, the degree to which learners feel most sure of their knowledge, makes disconfirming corrective feedback more salient and thus more influential in learning. Kluger and DeNisi suggest that feedback most effectively enhances performance when it directs attention to motivational and task-specific goals, not self-related goals. Research on the timing of externally provided instructional feedback— immediate or delayed after some learner performance— has yielded highly varied results. Hattie and Timperley suggest that this variation might be accounted for by level of feedback; immediate task feedback might most benefit task performance, but delayed processing feedback might most enhance cognitive processing of information.

Clearly, feedback is a critical, ubiquitous, and complex feature of learning processes guided by teachers or directed by learners themselves. Educators would do well to create environments that support challenging standards and goal-setting and provide informative feedback. Students can also benefit from assistance regarding how to look for and use feedback and how to provide feedback for themselves. Future research will continue to clarify conditions that optimize feedback effects for different tasks and learners.