Two further studies suggest that the situation is in any case more complex (see Andrade et al. 1999). In one study (Andrade et al. in preparation), we hypothesized that clinical studies tend to reveal memory for stimuli played during anesthesia, whereas volunteer studies tend not to, because the patients in the clinical studies are anxious about their operation and their anxiety promotes learning. Surgical stimulation may also promote learning. To test these hypotheses, we played words to patients shortly before they went into the operating theater and tested memory for those words on recovery from the anesthetic after surgery. During word presentation, one group of patients was lightly sedated with propofol, another group was anesthetized with propofol, and a third group was anesthetized with propofol and the words were played during intubation. This design allowed us to compare learning during anesthesia that was light because only a small dose of anesthetic was given, with learning during anesthesia that was light because the stimulating effects of intubation counteracted the sedative effects of the anesthetic.
We presented two types of word: the "yellow banana green pear" phrase for comparison with Jelicic et al.'s (1992a) clinical study, and Finnish words that patients did not comprehend. We tested memory directly, by asking participants to judge whether they recognized presented and new Finnish words, and indirectly by asking them to generate examples of fruits and colors and to rate how much they liked different Finnish words. This preference rating task measures the mere exposure effect, an increase in liking of previously encountered stimuli compared with new stimuli. Zajonc (e.g., Murphy and Zajonc 1993) has argued that this change in emotional response is more resistant to lack of consciousness than are cognitive changes picked up by tasks such as category generation. Our pilot study confirmed that recognition of the Finnish words was impaired by dividing attention at study, whereas the mere exposure effect was the same with full or divided attention.
Patients recognized and preferred the Finnish words presented before propofol infusion, compared with distracter words that had not been presented. However, none of the three groups showed statistically significant memory for category examples or Finnish words presented during propofol infusion. This study provides another example of consciousness without subsequent memory, and leaves unsolved the puzzle of why some studies seem to have demonstrated memory for words presented during clinical anesthesia. There was a hint of memory for words presented during intubation, suggesting that potentially painful stimulation promotes learning in some way.
Our third study therefore investigated memory for words played during minor surgery with propofol sedation or propofol anesthesia. Patients either were sedated with a relatively small dose of propofol supplemented with alfen-tanil, or were lightly anesthetized with a larger dose of propofol and somewhat less alfentanil (details will be reported in Stapleton and Andrade in press). The sedated patients remained responsive to command during surgery. All patients were played a list of common English words before propofol infusion and another list when surgery began. On recovery, memory was tested by playing word stems and asking patients to complete them. Traditionally, this word stem completion task is considered a measure of unconscious memory, because participants are asked to complete the stems with the first word that comes to mind. However, if no word comes to mind, participants may search conscious memory of the studied words for a suitable completion, so the test is not process-pure—it may pick up conscious as well as unconscious memory.
A solution is to encourage conscious retrieval of studied words but, for half the test items, to forbid participants to use the consciously retrieved words to complete the word stems. On this half of the test, the exclusion condition, unconscious memory increases the tendency to complete stems with presented words but conscious memory impedes it. On the other half of the test, the inclusion condition, both types of memory increase the number of completions with studied words. This so-called method of opposition, or process dissociation procedure, is widely used in cognitive psychology to estimate the contributions of conscious and unconscious memory to performance on a memory test (Jacoby et al. 1993). There is debate about the validity of the underlying assumption that the conscious and unconscious memory contributions are independent (e.g., Curran and Hintzman 1995, 1997; Jacoby et al. 1997; Russo and Andrade 1995), but the advantage of the technique is that it assesses conscious and unconscious memory using a single memory test.
In our study, patients used conscious and unconscious memory to complete word stems belonging to words presented before propofol infusion. For the words presented during surgery, their completions appeared to be driven by conscious memory. This finding may be artifac-tual, because in the exclusion condition patients may have rejected any words that seemed familiar rather than only those words they consciously recalled. A tendency to exclude all common words would show up as a difference in responses to distracter items in the exclusion and inclusion conditions. This difference was not observed, baseline performance being comparable in the two conditions. However, if unconscious memory for presented words led to those words being excluded because they seemed familiar, this would lead to an overestimate of conscious memory, because those words would be treated in the calculations as though they were consciously recalled, and there would be a corresponding underestimate of unconscious memory.1 Nonetheless, the interesting finding is that the patients in this study learned words presented during propofol infusion, even though the lightly sedated participants in our earlier studies had failed to do so. The amount of learning was comparable in the sedated and anesthetized patients.
Together, our three studies suggest that learning during anesthesia occurs only when surgery is present. We found no evidence of learning during sedation or light anesthesia when words were presented to patients before surgery, or to volunteers. By contrast, we observed learning during deeper anesthesia when words were presented during surgery. In our studies, surgery appears to be more critical for learning than is consciousness. Surgery is promoting learning in some way that is independent of its effect on depth of anesthesia.
A study by Lubke et al. (1999) indicates that level of consciousness influences learning during surgery. Lubke et al. played words to patients during surgery and tested their memory on recovery, using a word stem completion task with inclusion and exclusion instructions, as we did in our third experiment. They used a new EEG measure called the bispectral index to measure the depth of anesthesia at which each word was presented. This is the first study to assess the importance of depth of anesthesia for memory of individual stimuli. The outcome was a statistically significant but weak correlation between the bispectral index when words were presented and memory for the words on recovery. Lubke et al. tested trauma patients undergoing emergency surgery because their condition makes it difficult to maintain a stable depth of anesthesia during the operation. This meant that they could study the effects on learning of a wider range of depths of anesthesia than would be possible to obtain ethically in other patient groups. It is conceivable that the degree of surgical stimulation also varies widely in trauma surgery, compared with the minor surgery undergone by the patients in our studies. It would be interesting to know to what extent the variation in surgical stimulation explained the remaining variance in learning in Lubke et al.'s study.
Animal studies suggest that the effect of surgery on learning is due to increases in catechol-amine release in response to tissue damage. Catecholamine levels would have been high in Lubke et al.'s trauma patients. Catecholamines enhance memory retention (McGaugh 1989) and may promote learning during anesthesia. Weinberger et al. (1984) and Dariola et al. (1993) have shown that anesthetized rats cannot learn pairings between fear-provoking and neutral stimuli. This is not surprising if anesthetics antagonize the long-term potentiation of NMDA synapses that underlies learning. However, Weinberger et al. and Dariola et al. showed that injections of adrenaline enabled fear conditioning despite anesthesia (but see El-Zahaby et al. 1994, for a failure to replicate). A challenge for Flohr is to explain how catecholamines interact with the NMDA-mediated cell assemblies that he hypothesizes underlie consciousness and long-term memory. Perhaps we are dealing with two learning mechanisms here: one that is closely coupled to consciousness and is dependent on plasticity in NMDA synapses, and another that is independent of consciousness and dependent on or influenced by catecholamines. This second mechanism may be the one responsible for the apparently preserved learning of simple stimuli and stimulus pairings during anesthesia.
Was this article helpful?