Academic journal article The Psychological Record

Effects of Response Cost in Computerized Programmed Instruction

Academic journal article The Psychological Record

Effects of Response Cost in Computerized Programmed Instruction

Article excerpt

Two tenets of programmed instruction are that students should be permitted to proceed at their desired pace, and aversive contingencies should not be employed (Skinner, 1968). There is some evidence, however, that these fundamental assumptions may not provide the best conditions for learning.

When students are permitted to proceed at their desired pace, sometimes they respond so quickly that they make errors on material they know well. This behavior has been labeled racing (Crosbie & Kelly, 1993, 1994). If being permitted to proceed is a conditioned reinforcer, with escape from the learning situation the backup negative reinforcer, then faster responding produces the reinforcer more rapidly, and therefore is likely to be maintained. If, however, a delay is imposed after each frame (postfeedback delay), rapid responding is not immediately followed by presentation of the next frame, and therefore will not be adventitiously reinforced.

Crosbie and Kelly (1994) assessed whether postfeedback delays improve performance by stopping racing and promoting attention to study material in computerized programmed instruction. In that study, college students completed sets of Holland and Skinner's (1961) programmed text on behavior analysis in a computerized format. In Experiment 1 there were three conditions: (a) no postfeedback delay, (b) 10-s postfeedback delay for each frame, and (c) 10-s postfeedback delay for each frame answered incorrectly. Noncontingent delay produced better performance than no delay and contingent delay. To determine whether performance increased because subjects studied the material during delay periods, in Experiment 2 the three conditions were (a) no postfeedback delay, (b) 10-s postfeedback delay for each frame, and (c) 10-s postfeedback delay for each frame with the screen blank during the delay period. Noncontingent delay produced better performance than no delay, but there was no difference in performance between no delay and noncontingent delay blank screen. Hence, noncontingent delay improved performance because subjects used delay periods to study.

In a subsequent study with similar procedures (Kelly & Crosbie, 1997), baseline performance was increased from 64% to 79% by using shorter sessions. Even with higher baseline performance, noncontingent delay produced 7% more correct answers than no delay, and that advantage was increased to 13% at posttest, and 17% at follow-up. Unfortunately, noncontingent delay took 25% longer than no delay, and subjects occasionally complained that they were forced to wait after frames they answered correctly. Noncontingent postfeedback delay is effective, but it may take too long.

Punishment is another procedure that might stop racing and promote attention to study material. Programmed instruction has been conceptualized as an elaborate form of discrimination training (Skinner, 1968), and punishment of incorrect responses is one of the most effective ways to improve discrimination performance (Getsie, Langer, & Glass, 1985). This has been shown consistently with laboratory animals (Fowler, Hochhauser, & Wischner, 1981), intellectually disabled children (Harris & Tramontana, 1973), intellectually normal children (Miller, Moffat, Cotter, & Ochocki, 1973), impulsive children (Hemry, 1973), hyperactive children (Cunningham & Knights, 1978), and adults (Matthews & Shimoff, 1974). The functions of punishment most useful in learning are its ability to decrease error rates, slow an organism's rate of responding (Borresen, 1973; Donahue & Ratliff, 1976; Tindall & Ratliff, 1974), and get the organism to attend carefully to stimulus features in the environment (Balaban, Rhodes, & Neuringer, 1990; Muenzinger, 1934; Ratliff & Root, 1974). Each of these attributes is discussed below.

Punishment of incorrect responses plus reinforcement of correct responses produces lower error rates than does reinforcement alone (Boe & Church, 1968; Hemry, 1973; Trent, 1983; Warden & Aylesworth, 1927; Wischner, Fowler, & Kushnick, 1963; Witte & Grossman, 1971) or punishment alone (Brackbill & O'Hara, 1958; Wright & Smothergill, 1967). …

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.