Verbal discrimination reversal as a function of overlearning and percentage of items reversed: An extension

Acquired (conditional) equivalence: A basis for response-set effects in verbal-discrimination reversal performance.

“…However, in the Control condition, 11 out of 12 5"s made more errors on reversed than on nonreversed pairs and more total errors occurred on reversed than on nonreversed pairs. This latter observation is consistent with previous results (Paul, 1968) and suggests that, in the absence of differential equivalence-class cues (e.g., colors), the predominant tendency is for the occurrence of preshift responses to all pairs. The presence of the two different colors in the Homogeneous and Heterogeneous conditions seemed to have counteracted the tendency to emit preshift responses to all pairs, possibly by reducing the incidental equivalence of preshift responses, which is attained in original VD learning (Paul, 1970).…”

“…II, reversed and nonreversed pairs were respective members of experimentally established equivalence classes; the equivalence mediator (label) could identify shift status of each pair. The experiments support a conditional equivalence analysis of verbal-discrimination shift tasks.It has been suggested (Paul & Paul, 1968) and subsequently verified (Paul, 1970) that during verbal-discrimination (VD) learning, the correct stimulus alternatives acquire the dispositional characteristics of an equivalence class (i.e., a new response learned to one of these items can generalize to the others). Thus, if in a VD reversal (transfer) task, nonreinforcement of a correct preshift response activates a suppression tendency (i.e., "self-instruction to suppress the nonreinforced response"), then that suppression tendency may be expected to generalize to all other response items in the equivalence class.…”

“…Thus, if in a VD reversal (transfer) task, nonreinforcement of a correct preshift response activates a suppression tendency (i.e., "self-instruction to suppress the nonreinforced response"), then that suppression tendency may be expected to generalize to all other response items in the equivalence class. This analysis helps to explain a number of VD transfer phenomena, e.g., one-trial VD reversal-shift performance (Paul, 1966(Paul, , 1968) and onetrial elimination of preshift responses in a three-alternative VD shift (Paul, Callahan, Mereness, & Wilhelm, 1968), etc.Conventional operations for establishing equivalence classes were not explicitly present in the aforementioned experiments. Hence, to strengthen the analysis it is desirable to demonstrate that when such operations are present, VD-reversal performance is affected in the expected way.A 50% reversal shift, ordinarily a difficult task, should be made considerably easier if 1 The present experiments were supported, in part, by Grant GB7397 from the National Science Foundation.2 Requests for reprints should be sent to Coleman…”

“…Thus, if in a VD reversal (transfer) task, nonreinforcement of a correct preshift response activates a suppression tendency (i.e., "self-instruction to suppress the nonreinforced response"), then that suppression tendency may be expected to generalize to all other response items in the equivalence class. This analysis helps to explain a number of VD transfer phenomena, e.g., one-trial VD reversal-shift performance (Paul, 1966(Paul, , 1968) and onetrial elimination of preshift responses in a three-alternative VD shift (Paul, Callahan, Mereness, & Wilhelm, 1968), etc.…”

Effects of overtraining on verbal discrimination shifts: A differentiation-suppression hypothesis.

“…It is reasonably well-established that, compared to criterial training (CT), overtraining or postcriterial training (PCT) facilitates performance on a variety of "reversal tasks" (e.g., concept reversal- Bogartz, 1965;Caul & Ludvigson, 1964;discrimination reversal-Wolff, 1967;and verbal discrimination reversal-McClelland, 1943;Paul, 1966Paul, , 1968. Findings of this type contradict expectations from incremental, single-stage association models, which predict that interference with shift (second task) responses should increase as a direct function of degree of original (first task) training.…”