Is priming during anesthesia unconscious?

Abstract

General anesthesia provides an alternative to typical laboratory paradigms for investigating implicit learning. We assess the evidence that a simple type of learning—priming—can occur without consciousness. Although priming has been shown to be a small but persistent phenomenon in surgical patients (Merikle & Daneman, 1996) there is reason to question whether it occurs implicitly due to problems in detecting awareness using typical clinical signs. This paper reviews the published studies on priming during anesthesia that have included a measure of awareness or of anesthetic depth. We conclude that perceptual priming, but not conceptual priming, takes place in the absence of conscious awareness.

Introduction

Anesthesia, is by definition, the loss of sensation and conscious awareness. However, the first attempts by Horace Wells in the mid 1800s to use an inhalational anesthetic during surgery were accompanied by reports of awareness and pain. Modern day anesthetic techniques have advanced considerably, and patients typically receive a combination of agents to provide unconsciousness, analgesia, and muscle relaxation. Nonetheless, the problem of awareness during surgery has not been overcome and remains a rare, but potentially devastating complication. The most recent study of awareness measured by post-operative recall was conducted in the United States, and indicated an occurrence of 0.13%, comparable to the incidence reported in other countries (Sebel et al., 2004). Twenty million anesthetics are given per year in the United States, giving an estimated 26,000 cases of awareness. The incidence is higher in some types of procedure, such as cardiac surgery (Phillips, McLean, Devitt, & Harrington, 1993), trauma surgery (Bogetz & Katz, 1984), and cesarean section (Lyons & MacDonald, 1991). Common reports of awareness during surgery include perceptions of terror, paralysis, pain and helplessness, and psychological sequelae including re-experiencing the traumatic event, avoidance and hyper-arousal, and consistent with chronic post-traumatic stress disorder (Osterman et al., 2001).

Despite the relatively low incidence of reported awareness during anesthesia, there has remained suspicion that patients may process information whilst apparently unconscious. In an early study investigating unconscious perception, Levinson (1965) anesthetized 10 dental patients and mid-way through the operation, staged a mock crisis in which he exclaimed, “Stop the operation. I don’t like the patient’s color. His/her lips are much too blue. I’m going to give a little more oxygen.” Following this, the surgery continued as routine and all patients are reported to have made an uneventful recovery. However, under hypnosis one month later, four of the patients repeated verbatim Levinson’s statement, and another four had some recall for intra-operative events. Although this study is in many ways methodologically flawed (Levinson conducted both the mock crisis and the hypnosis, for example), it provided a startling demonstration that patients may continue to process information during anesthesia.

Recent research has confirmed that auditory information presented to patients during general anesthesia may prime, or temporarily activate, existing representations and thus enhance performance on post-operative implicit memory tasks in the absence of explicit recall (for reviews, see Andrade, 1995, Ghoneim and Block, 1992, Ghoneim and Block, 1997). Although there are as many negative as positive findings in the literature, a meta-analysis by Merikle and Daneman (1996) found highly significant memory when patients were tested within 12 h of surgery (12 studies with a total of 708 patients; effect size (r) = .23, p < .001). Thus, priming during anesthesia is a small but significant phenomenon and potentially provides useful insight into unconscious processing in the human brain.

However, the extent of the impairment of consciousness during anesthesia—anesthetic depth—has until relatively recently been difficult to ascertain. Anesthetic depth is the balance between the depressing effects of anesthetics and the increased sympathetic activation resulting from surgery, and thus fluctuates during an operation. In many early studies, such as those in Merikle and Daneman’s meta-analysis, anesthetic depth was not measured beyond observation of typical clinical signs, such as heart rate, respiratory rate, sweating, movement, and tears. This is problematic as the physiological response to surgery may not be related to the state of consciousness (Antognini and Schwartz, 1993, Rampil et al., 1993). In addition, when neuromuscular blockers are used, these clinical signs are invalid indicators of anesthetic depth due to paralysis. Moerman, Bonke, and Oostings (1993) reported that experienced anesthesiologists were unable to distinguish between cases of anesthetic awareness (with post-operative recall) and control cases (without post-operative recall) based on anesthetic charts reporting standard clinical signs.

So awareness during anesthesia is typically very difficult to detect at the time of stimulus presentation, and neither can it be determined retrospectively using post-operative memory tests. Explicit memory on recovery is suggestive of awareness during anesthesia, but can be an unreliable indicator because memory tests are not process-pure. Implicit memory may contribute to performance on a putative test of explicit memory, and vice versa. The relative purity of a test of explicit memory is inversely related to its sensitivity. A free recall task (“Can you tell me anything you overheard during your operation?”) will be fairly resistant to implicit memory contamination but is an insensitive measure of memory. Conversely, a forced-choice recognition task will detect weaker, less easily retrieved explicit memories but is susceptible to implicit memory influences, particularly when participants feel they are guessing. A correct answer to the question “Did you hear X or Y during surgery?” may reflect explicit recognition of X (or Y), and thus explicit processing of the item during surgery, or it may reflect implicit priming of that item in memory. Because memory tests do not provide pure measures of the explicit and implicit memory processes they are intended to measure, it is more accurate to use the term ‘direct test’ for a test that explicitly asks participants to retrieve a memory and ‘indirect test’ for one that assesses memory surreptitiously, for example by asking participants to respond to cues with the first word that comes to mind. However, we have used ‘explicit memory test’ and ‘implicit memory test,’ respectively, for these tasks, for simplicity of exposition and in keeping with the terminology in the literature we are reviewing, but we have taken care to specify the nature of the tasks used so that readers may assess their relative process purity.

In contrast to explicit memory, implicit memory for intra-operative stimuli on recovery tells us nothing about whether those stimuli were processed consciously or unconsciously during surgery. Incidental or even fully unconscious processing may lead to implicit memory formation, but so may conscious processing, as is the case in the many demonstrations of preserved implicit memory in amnesia. Thus, implicit memory for intra-operative stimuli may result from periods of awareness that, owing to the amnestic effects of anesthetics or general forgetting, are not available to conscious recollection on recovery.

The difficulty of assessing anesthetic depth and hence the adequacy of anesthesia raises an important question: Is priming during anesthesia occurring in brief moments of undetected awareness, or does it reflect truly unconscious processing? Thus, this field, like the mainstream implicit learning literature, has suffered from problems of detecting awareness. Typical laboratory studies of implicit learning attempt to show learning in the absence of awareness of the stimuli (subliminal presentation, e.g., Marcel, 1983a, Marcel, 1983b) or of co-variations between stimuli (e.g., artificial grammar learning, Reber, 1967; hidden co-variation detection, Lewicki, Hill, & Sasaki, 1989; and control of dynamic systems, Berry & Broadbent, 1984), in conscious participants. There is continuing debate, mainly focusing on methodological controversies, as to whether participants really are unaware of the information that contributes to enhanced performance at test. As a result, it has been difficult to determine whether unconscious perception actually exists or whether findings are attributable to weak conscious effects (see Shanks & St John, 1994). Problematically, most implicit learning paradigms have to resort to testing awareness retrospectively, typically through verbal report. Because participants are conscious, ‘on-line’ awareness testing would increase the risk of them becoming aware of the critical information.

Anesthetic paradigms have the potential to avoid this problem by ensuring that patients are unconscious, unaware of all sensory information. There are some reasons to believe unconscious priming may nonetheless be possible. Studies by Münte and colleagues suggest continued auditory processing during deep anesthesia, shown by the presence of mismatch negativity in the EEG response to ‘oddball’ auditory stimuli (Gross et al., 2004, Quandt et al., 2004). If at least a small amount of auditory processing persists when people are anesthetized, then memory formation for this information may be facilitated by the high levels of catecholamines released in response to surgery and known to act via the amygdala to facilitate memory formation for emotional stimuli (Deeprose et al., 2004, Stapleton and Andrade, 2000).

Advances in anesthetic depth monitoring and cognitive psychology have been combined to provide a more robust methodology than used in early priming during anesthesia research, and may now potentially reveal whether priming or even learning of new information persists in the absence of consciousness. There are now several approaches available to monitor depth of anesthesia, ranging from assessment of response to command to the analysis of EEG activity and auditory evoked potentials. Assessment of response to command reflects a behavioral measure of anesthetic depth, whereas monitors measuring EEG activity, including auditory evoked potentials, are electronic indices of anesthetic depth. Attempts to validate the electronic indices of anesthetic depth have focused on correlations with behavioral indices (e.g., response to command, Observer’s Assessment of Alertness and Sedation Scale) and clinical variables. Electronic indices are thus indicators of the probabilistic state of consciousness, rather than indicators of the actual state of consciousness. Recent comparisons of several electronic indices suggest they are fairly comparable in ability to predict consciousness (e.g., Bruhn et al., 2003, Muncaster et al., 2003). However, little research has been directed towards comparing the ability of each monitor to detect or predict subtle cognitive processing, such as priming and memory formation.

Theoretical and methodological developments in experimental psychology indicate the need to select appropriate memory tests for priming during anesthesia research. One issue is whether perceptual or conceptual priming is tested. Perceptual priming refers to enhanced processing fluency, e.g., increased accuracy in identifying a word masked in background noise or the generation of a word from a fragment. Conceptual priming refers to activation of related knowledge, for example, presenting the word “banana” during a study phase may increase the tendency to name “banana” when later asked to name types of fruit. Whereas there is some evidence that above chance performance on perceptual implicit memory tests may reflect unconscious priming (e.g., Hutchinson, Neely, & Neil, 2004), there is ‘virtually no evidence that priming on conceptual tasks reflects unintentional and unconscious memory’ (Butler & Berry, 2001, p. 195). Thus, there is little to be gained in using conceptual implicit memory tests (as many early anesthesia studies did) in attempting to detect unconscious priming. In addition to using monitors of anesthetic depth during stimulus presentation, the problem of contamination between explicit and implicit tests may be minimized with the selection of appropriate memory tests. Using comparable implicit and explicit memory tests in conjunction improves the likelihood of accurately detecting the implicit or explicit nature of memory (Reingold & Merikle, 1988). Jacoby’s (1991) process dissociation procedure also offers a promising technique for separating the contributions of conscious and unconscious influences on task performance (e.g., word stem completion or word fragment completion). Rather than comparing performance between implicit and explicit tasks, the procedure aims to measure the within task contributions of conscious and unconscious recollection in responding (Jacoby, 1998). We suggest the combination of measures of awareness (depth and awareness monitors) at the time of stimulus presentation with post-operative behavioral measures (memory tasks) provides a powerful test of whether or not priming during anesthesia is unconscious.

In an attempt to evaluate whether priming during anesthesia is truly unconscious and thus establish general anesthesia as an arena for investigating implicit learning more broadly, this paper reviews studies of priming during surgical anesthesia that have attempted to detect awareness or used a monitor of anesthetic depth during word presentation, and measured implicit memory for the words on recovery. We take implicit memory for intra-operative stimuli to be evidence for implicit priming when measures of awareness or depth reveal no signs of awareness during word presentation and there is no evidence of explicit memory for the stimuli.

Anesthetic techniques vary widely and for simplicity we have kept details of the techniques to a minimum in the text, summarizing them instead in Table 1. We do however note when opioid and benzodiazepine regimes have been used because although such regimes provide analgesia and amnesia for surgery, they do not reliably prevent patients becoming conscious during surgery (Russell, 1993).