Verbal discrimination reversal in a whole/part re-pairing transfer paradigm

Abstract

Whole/part transfer and re-pairing of rights and wrongs were employed in a study of verbal discrimination reversal learning. The whole/part and re-pairing procedures were used to create reversal conditions (0%, 50%, or 100% reversal) that were sirnilar at the outset of reversal learning in that all correct alternatives on the reversallist were higher in situational frequency than their respective incorrect alternatives. Similarly, reversal lists were constructed that were sirnilar in that all incorrect alternatives on the reversal list were higher in situational frequency than their respective correct alternatives. The results were consistent with the existing literature on verbal discrimination reversal, as performance was irnpaired when items reversed functions (50% and 100% reversal). There was little evidence that uniform frequency relations between rights and wrongs affected ease of reversallearning.The purpose of the present experiment was to investigate whether frequency discriminations were dominant in verbal discrimination reversal learning. Recent research and theorizing, while not entirely inconsistent with frequency theory (Ekstrand, Wallace, & Underwood, 1966), have at least implied that acquisition of areversal or partial reversal verbal discrimination list involves learning mechanisms other than frequency discrimination (e.g., Paul, 1968;Paul & Callahan, 1972; Pau1, Hoffman, & Dick, 1970;Raskin, Boice, Rubel, & Clark, 1968). The present experiment was initiated to determine if frequency units acquired in a prior verbal discrimination task influenced verbal discrimination reversal transfer.The rationale for this experiment developed from an extension of studies on prefamiliarization training and verbal discrimination learning (VDL). It is known that familiarization training on items that are to serve subsequently as correct alternatives facilitates VDL rela tive to familiarization training on incorrect alternatives or both correct and incorrect alternatives (e.g., Lovelace, 1969;Underwood & Freund, 1968;Wallace & Nappe, 1970). In the present study, differential familiarization training was given to rights and wrongs in a VDL task so that Ss could transfer to VDL reversal conditions with either a11 the new rights greater than wrongs in situational frequency or a11 the wrongs greater than rights in situational frequency. For this purpose, three departures from standard VDL procedures were required. First, pairs were presented differing numbers of tim es prior to transfer; in the present case, 12 VDL pairs appeared eight times each and 12 VDL pairs appeared two tirnes each in the first stage of the experiment. Second, a whole/part transfer *This research was supported by Grant MH 21895 frorn the National Institute of Mental Health. The authors arc grateful to Michael D. Murphy for his assistance in the preparation of materials used in this experiment. Requests for reprints should be sent to William P. Wallace, Department of Psychology, University of Nevada, Reno, Nevada 89507. task was employed. The f...