Transfer of learning in avoiding false memory: The roles of warning, immediate feedback, and incentive

Abstract

Participants learned semantic associates and were tested in the Deese–Roediger–McDermott (DRM) paradigm. A multiple-trial learning procedure with warning, feedback, and monetary incentive was employed to reduce false memory. Results showed that there was a progressive reduction of false memory over the trials as well as a generalization of learning to new lists of words in avoiding the critical nonpresented words attributable to explicit warning, feedback, and incentive, respectively. Both the feedback and the monetary incentive had an effect beyond what an explicit warning could obtain. In addition, participants were found to achieve false-memory reduction by enhancing the activation of the critical words and the monitoring process, rather than by strengthening the verbatim processing of the list words. It was concluded that false memory may not be as impervious to correction as was believed insofar as an effective training method is applied.

An earlier version of this paper was presented at the 45th Annual Meeting of the Psychonomic Society in 2004 in Minneapolis, MN. The research was supported by an NIH MBRS-SCORE grant No. 516803 to J.J. We thank Evan Heit, Jeffrey Neuschatz, Gabriel Radvansky, and an anonymous reviewer for valuable comments on earlier versions of this paper. We also thank Yolanda Matus, Dawn Rogers, and Ryan Zimmerman for help with testing participants and Ken Buckman, Theresa Norman, Ramon Guerra, and Chuck Prather for allowing their nonpsychology class students to participate in our experiments for extra credit.

Notes

¹ The reason why no examples of the critical words were shown to the participants in the GW condition was that a plain initial warning matched the initial warning used in the feedback conditions and therefore would allow an assessment of the postresponse feedback effect in a GW versus feedback comparison. We expect that, had examples of critical words been shown to the participants in the GW condition, the overall false-memory rate might have been lower than obtained for that condition.

² In computing c value, when the hit or false alarm rate was 1 or 0, 0.01 was subtracted from or added to it, respectively.

³ We thank Evan Heit for suggesting the bias and sensitivity measures that we used in this paper.