Effects of punishment for errors on discrimination learning by humans.

Reinforcement and punishment during programmed instruction

“…The amount of time that the individual takes to respond when stimuli are presented during a discrimination trial is called response latency. When punishment contingencies for incorrect responses are in effect, response latencies reliably increase (Brent & Routh, 1978;Erickson, 1970;Hoge & Stocking, 1912;Muenzinger, 1938;Munson & Crosbie, 1998). For example, Erickson (1970) examined the performance of 48 college students on a two-choice discrimination procedure as a function of feedback in the form of punishment for the incorrect response.…”

“…When punishment contingencies for incorrect responses are in effect, response latencies reliably increase (Brent & Routh, 1978;Erickson, 1970;Hoge & Stocking, 1912;Muenzinger, 1938;Munson & Crosbie, 1998). For example, Erickson (1970) examined the performance of 48 college students on a two-choice discrimination procedure as a function of feedback in the form of punishment for the incorrect response. Subjects were required to discriminate between two rectangles of light that were positioned along a vertical axis, and the location of the stimuli along this axis was the relevant property of the discrimination task.…”

“…Researchers examining the effects of punishment and reinforcement on learning typically used a two-choice discrimination procedure (cf. Penney & Lupton, 1961) to examine the effects of reinforcement and punishment on children and adult learning (Brackbill & O'Hara, 1958;Cunningham & Knights, 1978;Donohue & Ratliff, 1976;Erickson, 1970;Harris & Tramontana, 1973;Miller et al, 1973;Moses et al, 1979;Mueller & Gumina, 1972;Penney, 1961;Penney & Lupton, 1967;Ratliff & Root, 1974;Tindall & Ratliff, 1974;Witte & Johnson, 1973). The conclusion that punishment produces increases in correct responding during a two-alternative discrimination procedure could be an artifact of the experimental procedure: the punishment of one of two available responses necessitates an increase in the probability of the only alternative response (cf.…”