To the Editor:—

Recently, two studies 1,2reported probably the first evidence of a relation between memory for words presented during general anesthesia and depth of hypnotic state, as measured by the electroencephalographic Bispectral Index (BIS). 3The studies seem to diverge with respect to whether explicit (also direct, voluntary, or conscious) or implicit (also indirect, automatic, or nonconscious) memory for intraoperative events decreases with increasing depth. It is argued herein that, when the results of the first study are reanalyzed along similar lines as in the second study, both studies support the conclusion that explicit memory is primarily affected by depth of hypnotic state.

In both studies, the measurement of BIS was combined with a new procedure for separating explicit and implicit contributions to memory performance. The Process Dissociation Procedure 4uses two opposing memory test conditions. When the test regards word completion, a participant is instructed (i.e. , the inclusion instruction) either to complete a word stem (e.g. , BA . . .) as much as possible to a previously presented word (e.g. , BASIS), or to avoid such old words (i.e. , the exclusion instruction) and replace them as much as possible with new words (e.g. , BAKER). When no old word can be given as completion, the participant is instructed simply to complete the stem with the first word that comes to mind (e.g. , FO . . . : FOCUS). Even in the exclusion condition, however, some old words (i.e. , BASIS) may be completed despite the instruction not to do so, which represents a form of memory that is apparently not under conscious control (i.e. , implicit memory). Inclusion performance roughly equals the sum of explicit and implicit memory performance. The performance difference between the two opposing test conditions thus reflects control over memory performance (i.e. , explicit memory).

The first study 1examined a large sample of trauma patients with highly variable depths of hypnotic state. An individual BIS value could be calculated for each word per participant because every word was repeated 40 times in a 3-min interval. The subsequent study 2was performed on patients undergoing emergency cesarean section who received only relatively light levels of anesthesia. This study 2found some explicit, but no implicit, memory performance postoperatively at these light levels of anesthesia (average BIS, 76.3). Lubke et al. , 1in contrast, concluded in the trauma study that there was no evidence for explicit memory over the whole BIS range (from 20 to 97) and that, consequently, the small decrease in “general memory performance” (i.e. , inclusion performance) at low BIS ranges should be attributed to a dependence of implicit memory on depth of hypnotic state. It should be noted, however, that the latter conclusion was not based on a direct test of implicit memory performance at different levels of anesthesia but was inferred indirectly from the aggregate measure of implicit and explicit memory (i.e. , inclusion performance). If this combined memory measure decreased with depth of hypnotic state and there was no overall (i.e. , over all BIS values) explicit memory performance, then only implicit memory can be responsible for this decrease. The absence of significance, however, cannot be seen as evidence in favor of the null hypothesis. Relatively few words in the trauma study were associated with high BIS values. The overall difference between inclusion and exclusion performance only narrowly missed conventional significance levels. Only a small increase in average BIS value in the trauma study (average BIS, 54 ± 14) in the direction of the BIS levels of the cesarean section study (average BIS, 76.3 ± 3) probably would have sufficed also to yield significant overall explicit memory performance in the first study.

In the cesarean section study, direct tests on both explicit and implicit memory performance were performed, and a fully significant explicit memory effect was obtained because, in this study, most words were processed at a light level of anesthesia. Applying the direct comparison of inclusion and exclusion performance (in a binomial test) of the cesarean section study to the performances at categorized BIS levels in the trauma study, however, may solve the apparent discrepancy between the conclusions of the two studies. Inclusion performance at categorized BIS levels was reported by Lubke et al.  1in their table 2. Exclusion performance at categorized BIS levels in the trauma study 1was calculated for the purpose of this letter (table 1). Binomial tests revealed that, only at the highest BIS levels, inclusion and exclusion differed significantly (BIS 80.1–97: z = 1.75, P < 0.05; BIS 70.1–80: z = 3.01, P < 0.002), indicating some level of control over memory performance (i.e. , a form of explicit memory 4). At the same BIS level as in the cesarean section study, explicit memory performance thus also has been obtained in the trauma study. The indirect reasoning in the trauma article that there was no explicit performance and that, consequently, any dependence of memory on BIS should be accounted for in terms of implicit memory therefore seems to be contradicted by the finding of significant explicit memory only at higher BIS levels.

Table 1. Number of Target Hits (%) Observed in the Stem Completion Task with Exclusion Instructions, Number (%) of Target Words Presented during Surgery, and Probability of Explicit and Implicit Memory Performance from the Trauma Study, 1Calculated According to Jacoby, 4All at Categorized BIS Levels

Base rate completion performance was 0.32 in the inclusion condition and 0.33 in the exclusion condition. Without previous presentation, base level explicit and implicit performances should be 0.0 and 0.33, respectively.

BIS = Bispectral Index.

Table 1. Number of Target Hits (%) Observed in the Stem Completion Task with Exclusion Instructions, Number (%) of Target Words Presented during Surgery, and Probability of Explicit and Implicit Memory Performance from the Trauma Study, 1Calculated According to Jacoby, 4All at Categorized BIS Levels
Table 1. Number of Target Hits (%) Observed in the Stem Completion Task with Exclusion Instructions, Number (%) of Target Words Presented during Surgery, and Probability of Explicit and Implicit Memory Performance from the Trauma Study, 1Calculated According to Jacoby, 4All at Categorized BIS Levels

The conclusion drawn from these studies has practical relevance if anesthesiologists want to set BIS during surgery at the highest level for which no memory of intraoperative events is probable. According to the reasoning of Lubke et al. , 1even the highest BIS level in the study would suffice to prevent explicit memory. The present reinterpretation of the data suggests that the chance of finding explicit memory may be reduced sufficiently only below a BIS level of 70. The form of explicit memory obtained in the two studies 1,2probably should not be equated with free recall of intraoperative events, 2but the successful exclusion at testing implies that patients can make some type of conscious reference to the presentation of the words during anesthesia, when explicit memory is cued by word stems. Theoretically, implicit memory also may be harmful by affecting postoperative well-being and recovery without the patient being aware of its source. It is plausible that implicit memory may be disrupted at high levels of distraction 5or at deep levels of hypnosis, but the present data do not provide a clear indication of the BIS level at which implicit memory disappears. Seeking the lowest BIS level also to avoid implicit memories may bring hazards to the patient. Although it is still unclear 6whether the Process Dissociation Procedure 4is able to provide pure and independent estimates of explicit and implicit contributions to memory, the two studies 1,2at least seem to support a dissociation between the Process Dissociation Procedure estimates of explicit and implicit memory.

The author thanks Gitta Lubke, M.A., Free University Amsterdam, Amsterdam, The Netherlands, Phil Merikle, Ph.D., University of Waterloo, Canada, and Jeroen Raaijmakers, Ph.D., University of Amsterdam, Amsterdam, The Netherlands, for helpful discussions and for their assistance in performing the additional data analysis.

Lubke GH, Kerssens C, Phaf RH, Sebel PS: Dependence of explicit and implicit memory on hypnotic state in trauma patients. A nesthesiology 1999; 90: 670–80
Lubke GH, Kerssens C, Gershon RY, Sebel PS: Memory formation during general anesthesia for emergency Cesarean sections. A nesthesiology 2000; 92: 1029–34
Rampil IJ: A primer for EEG signal processing in anesthesia. A nesthesiology 1998; 89: 980–1002
Jacoby LL: A process dissociation framework: Separating automatic from intentional uses of memory. J Mem Lang 1991; 30: 513–41
Wolters G, Prinsen A: Full versus divided attention and implicit memory performance. Mem Cogn 1997; 25: 764–71
Curran T, Hintzman DL: Consequences and causes of correlations in process dissociation. J Exp Psychol Learn Mem Cogn 1997; 23: 496–504