Study Notes - Memory and the brain - Chapter : Affective Memory

Source : Memory and the brain


Color coding : Violet - Question  Red - Claim  Blue - Evidence  Green - Assumption  Yellow - Constraints

Reinforcement, Reward: Appraisal and Affective Memory


In the last chapter, I have argued that attention is a necessary condition for perceiving and remembering, and that attention is initiated by an appraisal that this is good to know. Perception, aided by memory, eventually leads to action. But before we can approach anything or even want to do so, we must appraise it as good for us in some way: good to know, which leads to attention; good to have, which results in wanting and approaching it. We must also appraise a possible action as "good to do" and want to do it. No memory experiment will succeed unless the subject wants to learn, remember what he has learned, and wants to report it. Although modern cognitive psychologists are apt to disregard motivation in connection with memory experiments, the subjects must be motivated to do what the experimenter asks them to do. Given the docility of the usual subjects, psychology students, all the motivation they need is to know that participating in the experiment is required for a passing grade.

In animals, a psysiological state or drive (hunger, thirst) is assumed to impel the animal toward something (food, water) that will satisfy the drive. But both human beings and animals must be motivated to learn what the experimenter expects them to learn. Learning has usually been ascribed to the "law of effect" ( Thorndike, 1931), or the reinforcement of the correct response. In other words, the subject must be reinforced by each correct response to go on learning. Thorndike assumed that the correct response strengthens the stimulus-response association directly and automatically. He rejected the view that reinforcement merely determines the choice of response after correct recall. Later views of reinforcement have been divided between these two views ( Atkinson & Wickens, 1971). One view sees reinforcement as influencing the registration of memory (learning), the other, as determining the retrieval (performance). Some theorists are beginning to doubt that reinforcement theory, developed from animal research, can apply to human learning ( Estes, 1971). In answer, Voss ( 1971) points out that the real priority is to develop a satisfactory theory of human learning, and a theory of animal learning will take care of itself. He says: "Thus, even if one adheres to the continuity position, it does not necessarily follow that a reinforcement framework should be adopted for understanding reward and punishment in human learning; instead, it also would be possible that development of some other view may be fruitful, and even applicable to animal behavior" (p. 39).

Despite the vast research effort devoted to this problem for many years, none of the available reinforcement theories spells out just how reward or reinforcement produces its effect. True, the learner is rewarded and the response (or the association between learned items) is reinforced. But how does the reward promote learning, or how is the response reinforced? I propose to review briefly the more important motivational theories to discover whether any of the mechanisms proposed explains the function of reinforcement and could aid in the quest of identifying the pathways that mediate its effects.


REINFORCEMENT THEORIES


In an extensive review, Tapp ( 1969) classifies reinforcement theories, according to the main factor stressed, as motivational theories, stimulus theories and response theories. I follow him in this ordering.

Among motivational theories, Tapp points out, the drive-reduction theory has found considerable experimental support. According to this theory, the drive spurs the animal to learn a task; successful performance reduces the drive. But the theory cannot explain why "animals will learn a task . . . to taste a nonnutritive sweet substance . . . press bars to turn on lights or receive a puff of a novel odor . . . run mazes . . . to explore an empty goal box. . . . (or why) they will work to receive electric shocks or injections of minute amounts of chemicals into certain parts of their brain." (p. 390). None of these performances can reduce a physiological drive. The only way these experimental findings could be incorporated into the drive-reduction theory would be to postulate a new drive for every new reward, which, as Tapp says, would be rather uncomfortable.

The newer arousal theories ( Berlyne, 1967) hold that a stimulus or a need that moves the animal from an optimal state of arousal induces drive or tension; and anything that reduces this state again to an optimal level is reinforcing. Such theories avoid the paradox fatal to drive reduction theories that organisms sometimes seek arousal, and not drive reduction. But the concept of arousal, taken from the reported EEG arousal on stimulation of the brainstem reticular system, has little definition and suffers from lack of agreement as to what constitutes arousal. In addition, Berlyne's argument states that a reward is reinforcing when it can induce an optimal level of arousal. And the arousal potential of the reward is a function of the arousal level of the subject (his need). But how does the optimal (or near optimal) level of arousal reinforce?