Memon & Stevenage's (1996) review of the cognitive interview suggests that it yields many errors and confabulations as well as accurate information. We argue that the demand characteristics of the interview persuade witnesses to provide details in their reports that are not consciously recollected. We suggest limiting errors by instructing witnesses to indicate, during the interview, whether they "remember" (consciously recollect) or "know" (believe, but do not recollect) that a given detail occurred.
2. Memon & Stevenage note that, although witnesses generally produce more accurate facts about a witnessed event during a cognitive interview than during a standard interview, there is a concomitant increment in errors. This increase in overall output, without an associated increase in accuracy, is reminiscent of the effect that hypnosis is theorized to have on memory recall. The explanation for this effect in hypnosis research has been that participants lower their criterion for reporting mental events as memories (e.g., Dywan & Bowers, 1983) such that any increase in the number of accurate responses (i.e., greater hit rate) is accompanied by more errors and confabulations (i.e., greater false alarm rate). It is possible that a similar shift is occurring with the cognitive interview, particularly because there are several factors present in the setting that may bias witnesses, causing them to increase the length of their reports. For example, during the interview, witnesses are asked to "report everything," without regard for how trivial or seemingly unimportant the information may be. Depending on the degree to which these instructions are emphasized by the interviewer, witnesses may attempt to reconstruct those aspects of the event that cannot be consciously recollected, which may lead to schema-driven errors and confabulations. Indeed, Memon & Stevenage point out (paragraph 6) that errors and confabulations often seem to derive from witness' schemas about the event, as if they are attempting to "fill in the gaps" of memory.
3. Other factors are present in the interviewing setting that are also likely to persuade witnesses to increase the amount of detail in their reports. For example, Memon & Stevenage indicate that one important influence on reports given during the cognitive interview is the knowledge that the interviewer appears to possess about the event. If the interviewer seems to be completely naive, then witnesses are inclined to report more information than if the interviewer appears knowledgeable. One reason for this may be that witnesses are inadvertently pressured, in the interview setting, to produce a sensible, coherent account of the event in question. If interviewers are naive about the event, then witnesses may not monitor their statements as closely, as they would if interviewers are knowledgeable, because no check for accuracy is in place. However, it is not clear to what extent witnesses monitor their statements even in the presence of a knowledgeable interviewer. Further demand for detailed reports from witnesses occurs if the interviewers behave in a manner indicating a strong desire for a successful interview, as Memon & Stevenage point out. Presumably, a successful interview is one where witnesses produce a detailed, consistent and reasonable account of the event in question and not one where there are large gaps in the narrative.
4. Memon & Stevenage argue that many errors occur when the interviewer is directing specific questions to the witness following the free report. Here again, errors may be occurring because witnesses are trying to conform to the demand to produce details. The sheer act of probing particular parts of the free report at length may, in some cases, inadvertently induces witnesses to report information that they cannot consciously recollect, but which is coherent with the gist of the free report.
5. Of course, not all of the factors that lower the witness' response criterion can be eliminated. For example, at some point during the interview, witnesses will have to be questioned about specific details that are of particular importance, such as the type of weapon employed in an armed robbery. However, if it is the case that the demands placed on witnesses undergoing a cognitive interview induce them to lengthen and fill out their reports, then it is also likely that they will produce details that are inaccurate unless some check is in place.
6. Some recent research, that has been focussed on capturing people's state of consciousness as they remember events, may provide interviewers with this check. This research has isolated techniques that separate those times when remembering is accompanied by conscious recollection from those times when it is not. For example, Tulving (1985) asked participants, during a memory test, to indicate whether a reported item was "remembered" or "known." If they responded that an item was "remembered," then they were indicating that something about the prior encounter with the study item was consciously recollected; in a sense, they believed that they were reliving the prior encounter. Conversely, if the item was "known," then memory for the item was unaccompanied by recollection.
7. Since Tulving's (1985) early study, the remember/know paradigm has generated a fairly substantial body of research, mostly with recognition tasks, which has revealed a number of important details about these judgments. First, remember/know judgments are dissociated by several different variables (e.g., levels of processing, word frequency, intentional versus incidental study instructions, attentional resources available at study; see Gardiner & Java, 1993, for a review). These dissociations are considered evidence for qualitatively different subjective states during retrieval. Often, the "remember" and "know" states of consciousness have been associated with the conscious recollection and familiarity components underlying dual-process models of recognition memory (e.g., Jacoby & Dallas, 1981). Second, remember and know judgments are not synonymous with "sure" and "unsure" states of confidence, although they are often correlated. For example, Gardiner and Java (1990) compared word and nonword stimuli in a recognition memory study and found that word stimuli produced more "remember" than "know" responses, whereas the reverse was true of nonwords. However, when they replicated the experiment, but replaced "remember" and "know" response categories with "sure" and "unsure," there were more "sure" judgments for nonwords than for words, but similar proportions of "unsure" responses across the two classes of stimuli. This suggests that when people make "know" judgments, they may not necessarily lack confidence, only a conscious reliving of the study episode during retrieval. Third, false alarm rates are typically lower for "remember" judgments than for "know" judgments (see Gardiner & Java, 1993 and Rajaram & Roediger, in press for reviews and for exemplary data). There are, however, some exceptions to this general rule. For example, Roediger and McDermott (1995) found a very high rate of false "remember" judgments in a word-list paradigm, and Roediger, Jacoby and McDermott (in press) found that the proportion of false "remember" judgments exceeded false "know" judgments in a misinformation paradigm. Nonetheless, in the vast majority of cases, where special circumstances designed to produce high levels of false, conscious recollection are absent, false "remember" judgments are rare. For example, in one recognition memory experiment, conducted in the first author's laboratory, that employed the paradigm (Higham & Cochrane, in preparation), an overall false alarm rate of 29% was obtained. However, only 26% of these errors (8% of total responses) were false "remember" judgments.
8. We believe an important new avenue of research, that may improve the efficacy of the cognitive interview, will involve the implementation of some version of the remember/know paradigm during the questioning process. As indicated above, many of the errors that participants produce may be due to the demand characteristics of the interview circumstance and the pressure to fill in gaps in the report so that a coherent narrative is produced. It is unlikely that many such errors are accompanied by conscious recollection of the event, although only empirical testing will decide this for certain (e.g., see Roediger et al., in press). The remember/know distinction potentially could be applied during either the free report or while witnesses are undergoing more specific questioning. In either case, witnesses need only indicate whether some aspects of the context are remembered along with the event itself. For example, if a witness reports, during an interview, that a thief pointed a gun at the victim, he may be reporting this because he noticed that the robber had a wedding ring on his finger. If so, the report would be accompanied by a "remember" judgment. Conversely, if no such contextual information comes to mind, then this aspect of the report would receive a "know" judgement. Note that in both cases, the witness may be quite certain that a gun was present, but the state of consciousness during retrieval is different.
9. One promising possible outcome of research employing the remember/know paradigm would be to find that the errors and confabulations that characterize the cognitive interview are accompanied by a state of consciousness that is different from the accurate statements. One possibility is that the errors will be manifest as "know" judgments, but that accurate responding will be manifest in either "remember" judgments alone, or in both "remember" and "know" responses. Should this finding be obtained, it would be reasonable to assume that the more contextual information that the witness can retrieve (e.g., the gun's appearance or what the thief was saying when it first appeared), the more likely it is that the memory is accurate. Under circumstances where a great deal of importance is placed on the accuracy of the witness' account (which may, in fact, be the norm; see Memon & Stevenage, paragraph 29), only those reports accompanied by "remember" judgments should be considered.
10. It may also be necessary to consider the results of applying the remember/know distinction to hypnotic interviews as well as cognitive interviews. As outlined above, the effect of both procedures may simply be to produce response criterion shifts. Conceivably, if some of the errors and confabulations are "filtered out" by examining only reports accompanied by conscious recollection, perhaps both procedures will yield improved recall. This, too, is an empirical issue, but if the conclusion is to be that it is specifically the application of laboratory-based memory principles to the cognitive interview that improve recall in that setting, then the comparison to hypnotic interviews will be necessary. Hypnosis does not employ these principles; so if accuracy in recall is enhanced by employing the remember/know paradigm with both procedures, then the claim cannot be that it is specifically the application of cognitive principles to the cognitive interview that provides the benefit. Indeed, it may only be necessary to fine tune the increased output of suggestible or imaginative witnesses, by more closely analyzing their state of consciousness, to obtain enhanced recall.
Dywan, J. & Bowers, K.S. (1983). The use of hypnosis to enhance recall. Science, 222, 184-185.
Gardiner, J.M. & Java, R.I. (1990). Recollective experience in word and nonword recognition. Memory and Cognition, 18, 23-30.
Gardiner, J.M. & Java, R.I. (1993). Recognizing and remembering. In A. Collins, M.A. Conway, S.E. Gathercole & P.E. Morris (Eds.), Theories of memory (pp. 163-188). Hillsdale, NJ: Erlbaum.
Higham, P.A. & Cochrane, R.M. Processing fluency can cause false recollection. Manuscript in preparation.
Jacoby, L.L. & Dallas M. (1981). On the relationship between autobiographical memory and perceptual learning. Journal of Experimental Psychology: General, 110, 306-340.
Memon, A. & Stevenage, S.V. (1996). Interviewing Witnesses: What Works and What Doesn't? PSYCOLOQUY 7(6) witness-memory.1.memon.
Rajaram, S. & Roediger, H.L. (in press). Remembering and knowing as states of consciousness during retrieval. In J.D. Cohen & J.W. Schooler (Eds.), Scientific approaches to the study of consciousness. Hillsdale, NJ: Erlbaum.
Roediger, H.L., Jacoby, J.D. & McDermott, K.B. (in press). Misinformation effects in recall: Creating false memories through repeated retrieval. Journal of Memory and Language.
Roediger, H.L. & McDermott, K.B. (1995). Creating false memories: Remembering words not presented in lists. Journal of Experimental Psychology: Learning, Memory & Cognition, 21, 803-814.
Tulving, E. (1985). Memory and consciousness. Canadian Psychology, 26, pp. 1-12.