Editorial II

Gone but not forgotten—or was it?

R. A. Veselis1

1 Department of Anesthesiology, Box 24, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, NY 10021, USA and Weill Medical College of Cornell University E-mail: veselisr{at}mskcc.org

Implicit memory has been described as ‘gone, but not forgotten’, as it is a form of memory we do not know we have.1 This type of memory is only evident by changes in behaviour after exposure to a stimulus that we do not recall. This issue of the British Journal of Anaesthesia contains an exemplary study by Deeprose and colleagues2 which investigates implicit memory formation during anaesthesia, more specifically regarding the influence of surgery on this phenomenon. Implicit memory occurs at the opposite end of the spectrum from the dreadful experience of awareness, and observation of such is frequently referred to as ‘learning during anaesthesia’. As Deeprose and colleagues explain, ‘learning’ is a term best reserved for explicit memory, and ‘priming’ is more appropriate for unconscious learning or implicit memory. The techniques of intraoperative suggestion and the influence of the stressful and fearful surgical experience on postoperative events and recovery depend on understanding the nature of this phenomenon. The problem in this field of investigation is the extreme difficulty of obtaining reproducible results.3 This problem is added to by the unknown nature of modulating factors on this type of memory formation. In clinical studies, many confounding variables are present that could account for different results. If only we knew what these variables were, and how they influenced memory formation! One of the main confounds, the depth of anaesthesia, is addressed by the collection of bispectral index (BIS) data. The study by Deeprose and colleagues2 proposes to address another confound—the presence of surgery. The excellent methodology used to assess implicit memory should serve as an example for future studies in this area. The results of their study demonstrate that, despite this good methodology, investigation in this field is not for the faint of heart.

If one considers the literature as a whole, it seems that there are relatively equal numbers of studies supporting either side of the question of whether memory formation can or cannot occur during anaesthesia. Academic controversy exists over how to classify this type of memory: is it implicit, is it weak explicit, or is it some other form of memory? For example, what form of memory is retrieved under hypnosis?4 What is clear is that this memory is not a consciously recollectable form of memory, such as that which occurs in awareness. In this editorial, the term ‘implicit memory’ will be used to describe this phenomenon.

All things considered, it seems likely that memory formation can occur during clinically adequate levels of anaesthesia under some circumstances, which are yet to be defined. The problem is that even the same group of investigators cannot necessarily reproduce positive findings in subsequent studies.5 6 Added to this is the fact that a given set of results may have quite different interpretations depending on subtle nuances in the data. Only by continuing advances in methodology and analysis of data will a rational approach to this question be possible.

As the incidence of learning during anaesthesia is so low, the biggest problem is that of noise or random effects. Even a quick look at Figure 1 in the paper of Deeprose and colleagues1 reveals the typically tremendous amount of variability in such data. The critical question is whether the signal (a memory) is reliably different from noise (the placebo or control effect). In the present study, the answer to this question is ambiguous. As discussed subsequently, if the data are looked at in one way, an effect is present. But looking at it another way, it is not. Conservatively, one would reject the hypothesis that an effect is present. But it is tremendously important to our understanding of the anaesthetic state if an effect is really present. Recent evidence supports the plausibility of such an effect.

As discussed by an increasing number of authors, there is a sound physiological basis for the intensity of fearful memories—those that involve processes mediated through the amygdala.7 8 Fear-mediated memories are of a different nature from those not modulated by amygdalar influences. It is very likely that the surgical experience provides a setting where one might expect such modulation of memories. Importantly, at least in a volunteer setting, this modulation can be nullified by the relatively simple intervention of beta-blocker administration.9 10 With the advent of more reliable intraoperative monitoring of the hypnotic and pain components of anaesthesia, the ability to administer beta-blockers rationally will continually increase. Manipulation of central adrenergic tone with agents such as dexmedetomidine may also be important in this regard. Thus, relatively simple pharmacological intervention could quite possibly have a profound influence on the surgical experience—at least our unconscious memories and visceral experience of it.

Can the brain perceive stimuli or, more importantly, form memories during behavioural unresponsiveness? There is ample evidence from electrophysiology and brain activation studies, during both natural sleep and drug-induced unresponsiveness, that cortical processing of stimuli is possible.1114 The formation of contextually rich, explicit memories (those typified by recall of how we got to work today, for example) requires the participation of multiple distributed brain structures, such as the hippocampal complex and prefrontal and parietal regions, amongst others.15 The function of these distributed regions is depressed during anaesthesia, and this correlates well with the lack of this type of explicit memory formation during adequate anaesthesia. Sensory brain regions, however, seem to be more resistant to these depressant anaesthetic effects.12 16 Of note is the fact that the molecular machinery for memory consolidation is present in perisensory regions,17 and that these regions also demonstrate working memory processes.18 Thus, rudimentary memory formation may be possible in these brain regions alone. The state of deep sedation or anaesthesia provides an environment where weak memories may be strengthened, for example by waves of electrical activity similar to those occurring in sleep.19 It may be that the amygdala is particularly active in this state, and its modulatory influence thus increased.20

Thus, although results of this study are not definitive, the findings reported occur in the context of a physiological milieu in which memory formation might be possible. An excellent aspect of the methodology to be highlighted is the pilot testing of procedures, including a determination of the sensitivity of various memory tests. After all, if a test cannot detect implicit memory in subjects without drug, how can we possibly interpret results with drug? It seems that the commonly used word-stem completion task used to detect perceptual priming works well in this regard.

Does implicit memory correlate with anaesthetic depth? Such a relationship would seem to make sense, and the study of Lubke and colleagues5 in trauma patients seems to demonstrate this relationship. Using a word-stem completion test for material presented during anaesthesia, they found an increase in hit rate to targets (words presented during anaesthesia) in patients with higher BIS values. A first glance at similar data in the study of Deeprose and colleagues indicates that the same correlation is present. However, Figure 1 reveals that the tendency towards increased hits to targets probably represents a decreased rate of hits to distractors (words not previously presented, and referred to as ‘baseline’ performance—a somewhat confusing terminology as it does not refer to data collected before experimental intervention but rather to data collected concurrently with performance for targets). This is the critical control effect of the ‘baseline’ performance against which memory formation is compared. In Lubke’s study,5 this baseline performance did not change with BIS value, but it did so in the study of Deeprose and colleagues. No compelling explanation comes to mind, though Deeprose and colleagues offer some possible explanations. Once again, the vicissitudes of research in this area are demonstrated. The important take-home point is the utter necessity of a control procedure, in this instance an accurate determination of baseline performance.

Deeprose and colleagues suggest that implicit memory is related to the main experimental manipulation—that of surgery. This idea is certainly attractive, being based on the large body of literature regarding enhanced memory formation in a fearful setting, mediated by the effects of stress hormones on the amygdala. However, the presence of surgery was not the only manipulation present: fentanyl was administered before the surgery-absent period, but not before the surgery-present time period; surgery-present always followed surgery-absent, thus introducing a time confound; and different doses of propofol were present for the two time periods in question, even though the hypnotic effect was constant, as assessed by BIS. In addition, as mentioned at the beginning of this editorial, though the incidence of implicit memory in the surgery-present time period was different from 0 (one way of looking at the data), it was not different from the surgery-absent time period (another way of looking at the data). Though these results may be disappointing in their vagueness, the exemplary methodology for memory testing is once again emphasized.

Slowly and painfully, research in this area is making progress. Studies performed in the harsh environment of the operating theatre are necessary to make results clinically relevant. However, the incidence of memory detection is so low that results hover between ‘maybe present’ and ‘maybe absent’. Further insights can be obtained by careful study in an artificial environment, such as in volunteer studies. In this setting a reproducible event, such as priming based on emotive content, can be studied using much more controlled interventions. Additional insight can be gained by powerful techniques such as functional neuroimaging.

Each approach mines data and insight more slowly and with more difficulty than initially anticipated, but eventually the two lines of investigation will meet, with no small amount of jubilation.

Acknowledgement

The authors received funding from the NIH, R01 GM58782, Bethesda, MD, USA.

References

1 Sebel PS. Memory during anesthesia: gone but not forgotten? Anesth Analg 1995; 81: 668–70[ISI][Medline]

2 Deeprose C, Andrade J, Varma SV, Edwards N. Unconscious learning during surgery with propofol anaesthesia. Br J Anaesth 2003; 92: 000–000

3 Ghoneim MM, Block RI. Learning and memory during general anesthesia. Anesthesiology 1997; 87: 387–410[ISI][Medline]

4 Chortkoff BS, Gonsowski CT, Bennett HL, et al. Subanesthetic concentrations of desflurane and propofol suppress recall of emotionally charged information [see comments]. Anesth Analg 1995; 81: 728–36[Abstract]

5 Lubke GH, Kerssens C, Phaf H, Sebel PS. Dependence of explicit and implicit memory on hypnotic state in trauma patients. Anesthesiology 1999; 90: 670–80[CrossRef][ISI][Medline]

6 Lubke GH, Kerssens C, Gershon RY, Sebel PS. Memory formation during general anesthesia for emergency cesarean sections. Anesthesiology 2000; 92: 1029–34[ISI][Medline]

7 Maratos EJ, Dolan RJ, Morris JS, Henson RN, Rugg MD. Neural activity associated with episodic memory for emotional context. Neuropsychologia 2001; 39: 910–20[CrossRef][ISI][Medline]

8 Buchel C, Dolan RJ. Classical fear conditioning in functional neuroimaging. Curr Opin Neurobiol 2000; 10: 219–23[CrossRef][ISI][Medline]

9 Harmer CJ, Perrett DI, Cowen PJ, Goodwin GM. Administration of the beta-adrenoceptor blocker propranolol impairs the processing of facial expressions of sadness. Psychopharmacology (Berlin) 2001; 154: 383–9[CrossRef][ISI][Medline]

10 Cahill L, Prins B, Weber M, McGaugh JL. Beta-adrenergic activation and memory for emotional events. Nature 1994; 371: 702–4[CrossRef][ISI][Medline]

11 Czisch M, Wetter TC, Kaufmann C, et al. Altered processing of acoustic stimuli during sleep: reduced auditory activation and visual deactivation detected by a combined fMRI/EEG study. Neuroimage 2002; 16: 251–8[CrossRef][ISI][Medline]

12 Veselis RA, Reinsel RA, Feshchenko VA, Beattie BJ, Akhurst T. Auditory rCBF covariation with word rate during drug-induced sedation and unresponsiveness: an H2O15 study. Brain Cogn 2003; in press

13 Cote KA, Campbell KB. P300 to high intensity stimuli during REM sleep. Clin Neurophysiol 1999; 110: 1345–50[CrossRef][ISI][Medline]

14 Plourde G, Joffe D, Villemure C, Trahan M. The P3a wave of the auditory event-related potential reveals registration of pitch change during sufentanil anesthesia for cardiac surgery. Anesthesiology 1993; 78: 498–509[ISI][Medline]

15 Eichenbaum H. A cortical–hippocampal system for declarative memory. Nat Rev Neurosci 2000; 1: 41–50[CrossRef][ISI][Medline]

16 Heinke W, Fiebach CJ, Schwarzbauer C, et al. Sequential effects of propofol on functional brain activation induced by auditory language processing: an event-related functional magnetic resonance imaging study. Br J Anaesth 2003; in press

17 Kraus M, Schicknick H, Wetzel W, et al. Memory consolidation for the discrimination of frequency-modulated tones in mongolian gerbils is sensitive to protein-synthesis inhibitors applied to the auditory cortex. Learn Mem 2002; 9: 293–303[Abstract/Free Full Text]

18 Tallon-Baudry C, Bertrand O, Fischer C. Oscillatory synchrony between human extrastriate areas during visual short-term memory maintenance. J Neurosci 2001; 21: RC177[Abstract/Free Full Text]

19 Buzsaki G. Memory consolidation during sleep: a neurophysiological perspective. J Sleep Res 1998; 7: 17–23.[CrossRef][ISI][Medline]

20 Maquet P, Peters J, Aerts J, et al. Functional neuroanatomy of human rapid-eye-movement sleep and dreaming. Nature 1996; 383: 163–6[CrossRef][ISI][Medline]