McDermott, K.B. & Roediger, H.L. (1998). Attempting to avoid illusory memories: Robust false recognition of associates persists under conditions of explicit warning and immediate testing. Journal of Memory and Language, 39, 508-520.

Introduction

-Previous research by the authors has shown that presentation of words that are related to a non-presented word can induce high levels of false recall.

-Now the authors ask: "What is the source of the associatively induced false memory?"

-The authors offer two possibilities to answer that question: failure of reality monitoring (conscious awareness of critical word in study phase) or the critical associates are primed due to spreading activation in the semantic network (lack of conscious awareness of critical word in study phase).

-However, they seem to favor the first hypothesis and offer several points that support the failure of reality monitoring account.

-The critical associates (falsely recalled words) are classified as remembered by subjects, thus indicating that the subject can recall some specific aspect of the word's presentation (which never happened), It seems subjects are consciously aware of the critical associate during the study phase and then later falsely identify it as an external event instead of an internally genereated event.

-If the critical associates come to mind during the study phase then one would expect priming for these words on a perceptual implicit memory test.

This indeed is the case although not the extent as words which are actually presented. Once again this points to the explanation that false memories, in this paradigm, are due to the failure of reality monitoring.

-So this is what they did about it: They tested reality monitoring judgments directly in three experiments. In all three experiments they followed the same critieria. They informed subjects about the dangers of false recall and which words to look out for. The subjects were given a sample list to demonstrate the phenomenon. They presented the critical associate in half the lists and informed the subjects that the critical associate would appear half the time.

Experiment 1

-Purpose: To determine if subjects can encode accurately and remember whether the critical associate was presented or not presented.

-Participants: 24 Rice students, partial fulfillment of a course requirement.

-Methods: The subjects heard 12 associative lists, in half of the lists the critical associate was presented in the remaining lists it was not presented. In addition to the 12 lists heard, 6 lists were unheard and used for counterbalancing purposes. Study of all 12 lists was followed by a recognition test containing 18 test words, each of which was a critical associate. Six of the critical associates were studied, 6 were not studied but their related lists had been presented, and 6 were not studied and their related lists had not been presented.

-Results: Subjects were highly accurate in identifying critical associates that were presented. However they were poor at identifying critical associates that were not presented, making the mistake of saying these items were presented. In sum, when the critical associate had not been presented in the study list, subjects were very poor at determining so.

-Where to go next: Maybe subjects could make the discrimination immediately after the list was presented but not after hearing 12 lists and hearing test instructions. Maybe the retention interval is too long, so....

Experiment 2

-Purpose: To determine "whether subjects could discriminate between instances in which the critical associate was and was not presented."

-Subjects: 16 subjects from Washington U. Subjects were paid five dollars to play.

-Methods: Twenty lists were chosen for this experiment. Subjects were given a one-item recognition test following each list presentation. The item tested was always the critical ssociate.

-Results: Most non-presented items were correctly identified as items that were not presented. However on over one-third of the trials subjects incorrectly identified the non-presented critical associates as having been presented.

-Where to now: This makes our authors wonder if their warning instructions are having any effect on the sujects. maybe the memory illusion isn't under conscious control, so...

Experiment 3

-Purpose: To determine if waring instructions are effective at reducing the memory illusion.

-Subjects: 32 Washington U undergraduates.

-Methods: Only half the subjects were warned about the dangers of the false memory phenomenon. Almost identical procedure to experiment 2 except that four words were tested after each list instead of one.

-Results: Warning instructions did affect the responses for the standard studied words and for the critical non-presented items. However there is not difference between the two conditions for the critical presented words. In sum, subjects are good at determining when the critical word had been presented but subjects are not any better at this task when warned. More importantly though, resluts show that the probability of false recognition of the critical non-presented associate was weakened by the warning instructions. So warning instructions are somewhat helpful in studying memory illusions. In addition, it was found that warned subjects produced fewer false alarms to the critical associates but at the expense of lower hit rates.

Discussion (interesting tidbits):

-When critical associates are included in the study list, subjects are better at identifying as a presented item than the standard studied item.

-Subjects have only limited success in bringing the illusion under conscious control, and appear to be unable to eliminate it.