Academic journal article The Psychological Record

Expectancies as Factors Influencing Conditional Discrimination Performance of Children

Academic journal article The Psychological Record

Expectancies as Factors Influencing Conditional Discrimination Performance of Children

Article excerpt

Thorndike's Law of Effect (1898, 1911) represented an initial attempt to explain a fundamental principle of behavior, namely, that an organism's actions within its environment are learned and maintained through the consequences of those actions. Thorndike's explanatory S-R "bonding" mechanism suggested that responses were catalyzed ("stamped in" or "stamped out") depending upon the organism's experience of satisfaction or discomfort. Later, learning theorists turned toward other accounts, most commonly, some form of mediational account. In these accounts, the behavior's consequent event appeared anticipatorily as a conditioned response, motive state, representation, or "expectancy" (Amsel, 1958; Gough, 1961; Hull, 1930; Meehl & MacCorquodale, 1951; Mowrer, 1947; Osgood, 1953; Rozeboom, 1958). One successful theoretical orientation was to view expectancy of outcomes as a cognitive phenomenon as suggested in the theories of Tolman (1932) and made explicit by MacCorquodale and Meehl (1953).

Results from animal experiments are consistent with both the existence and functional status of expectancies. A useful strategy for investigating whether subjects form expectancies for reward is the "differential outcomes" procedure. Typically used is a conditional discriminative choice task; subjects in such experiments learn to make one response in the presence of one discriminative stimulus cue and a different response in the presence of a second one. In the differential outcomes procedure, one reward follows the correct choices following one cue, whereas a different reward follows those correct choices made to the other cue. This procedure facilitates both (a) initial learning of conditional relationships (Overmier, Bull, & Trapold, 1971; Trapold, 1970) and (b) memory for the conditional cues in delayed matching to sample tasks (Brodigan & Peterson, 1976) when compared to performance reinforced in the more traditional manner of using the same outcome to reinforce all correct responses. Moreover, switching the scheduled rewards disrupts the choice behavior even though every correct response continues to produce a reward - albeit not the scheduled reward (Peterson & Trapold, 1980; see also Tinklepaugh, 1928). Such effects have been the basis for inferring that subjects have expectancies for specific rewards. This powerful training procedure has become the focus of an expanding series of experiments (for a review see Goeters, Blakely, & Poling, 1992).

Another functional significance of expectancies is manifest in a differential, determining effect on choice behavior. A procedure used to demonstrate this effect is the transfer of control procedure described in Figure 1. This procedure consists of two training phases (a) a conditioning phase in which subjects observe a stimulus paired with a particular reinforcer and (b) an instrumental learning phase in which subjects receive differential outcomes in a prior conditional discrimination task. In the critical transfer of control test phase, the same two options that serve as choice alternatives in the conditioned discrimination training task, again play that role, but now for a new conditional cue. The "new" cue is the stimulus that was simply paired with a particular reinforcer in the initial conditioning task. This allows one to test the prediction that in such arrangements (a) the pairing stimulus elicits an expectancy for the specific reward with which it was paired and (b) that this expectancy, in turn, guides the selection of the choice alternative stimulus that also lead to that reward. The theory also predicts that subjects who receive nondifferential rewards (i.e., random or common rewards) form a common expectancy, one which does not selectively lead to either response alternative. In studies using this procedure, performance in the transfer of control test was higher when subjects received differential outcomes than when they received nondifferential outcomes (Kruse, Overmier, Konz, & Ronke, 1983; Overmier & Lawry, 1979; Trapold & Overmier, 1972). …

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

Oops!

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.