Detection of Auditory Input, Unconscious or Implicit Memory

Although the patient may not overtly recall a stimulus or an event, auditory input can register in the brain during apparently adequate surgical anesthesia. Auditory and verbal input must be "meaningful" for it to register in the patient's memory. Frequently, hypnosis or other cues may be needed to elicit recall.

Levinson^[63] performed the classic study of detection of meaningful auditory input under anesthesia. Ten volunteers undergoing dental surgery were given thiopental followed by nitrous oxide and diethyl ether. Monitoring the EEG for an irregular slow-wave-high-voltage pattern allowed the anesthetist to maintain a similar depth of anesthesia in all patients. This EEG pattern was considered equivalent to moderate to deep ether anesthesia. During surgery, the anesthetist provided verbal stimulation to the patient in the form of an intraoperative crisis by verbally stating that cyanosis was present and then treated appropriately. All 10 patients had no spontaneous recall of the simulated intraoperative crisis. Under hypnosis, however, four patients could remember the frightening words in exact detail. An additional four remembered someone speaking to them. All eight became anxious and either emerged spontaneously from their hypnotic trance or refused to continue exploring the event. One subject had activation of the EEG pattern when the intraoperative crisis occurred, but no recall of the event.

Blacher^[64] described his efforts to duplicate Levinson's experiment by using noxious, threatening verbal stimuli and hypnosis for subsequent recall. He reported similar findings, but he did not complete the project because he considered it too inhumane. Benign verbal stimuli did not produce significant evidence of auditory recall. He postulated that a noxious or crisis event was required. Bennett and coworkers^[65] randomly assigned 33 subjects to receive either no intraoperative verbal stimuli or a personalized, but nonthreatening instruction to pull on their ear during a postoperative interview. Anesthesia was maintained with nitrous oxide, halothane, or enflurane. All subjects given the intraoperative suggestion did not recall the event. The incidence of pulling on the ear during the postoperative interview was significantly higher in patients given the intraoperative message than in those not given the message. This and other studies^{[66] [67]} suggest that unconscious memory may occur during general anesthesia, as shown by the effect of intraoperative suggestion on post-operative behavior. Other studies with similar methodology, however, demonstrate no unconscious memory.^{[68] [69] [70]}

In a volunteer study examining the effect of subanesthetic concentrations of nitrous oxide and isoflurane on memory and responsiveness, Dwyer and colleagues ^[71] demonstrated that memory was decreased by increasing concentrations of each anesthetic. Both conscious and unconscious memories of the information presented during anesthetic administration were prevented by 0.45 MAC isoflurane, but they were not completely prevented by 0.6 MAC nitrous oxide. Although this volunteer study had significant methodologic rigor, it did not address the impact that surgical stimulation could have on anesthetic concentrations required to prevent conscious and unconscious memory. The stimuli in this study involved only verbal and visual input, stimuli less intense than surgical manipulation.

Veselis and colleagues^[72] used the best current clinical pharmacology methodology to characterize the pharmacodynamics of anesthetic drugs used during sedation when the patient is still awake and cooperative, but amnestic. The authors evaluated the effects of midazolam, propofol, thiopental, and fentanyl on volunteer participants' memory of words and pictures at equisedative plasma concentrations. Controlling for sedation was essential in this research because sedation itself can produce amnesia resulting from inattention to the stimuli presented for later recall. Equi-sedative concentrations were as follows: midazolam, 64.5 ng/mL; propofol, 0.7 µg/mL; thiopental, 2.9 µg/mL; and fentanyl, 0.9 ng/mL. The plasma concentrations of drugs that result in 50% of maximal effect (Cp₅₀ ) for loss of memory to words were 56 ng/mL for midazolam, 0.62 µg/mL for propofol, 4.5 µg/mL for thiopental, and 3.2 ng/mL for fentanyl. Large effects on memory were produced only by propofol and midazolam. Thiopental had mild memory effects, whereas fentanyl had none. These findings have relevance for examining the choice of drug for sedation, but they cannot be translated into an anesthetic situation with general anesthesia, surgical stimulation, and prevention of intraoperative recall. Although certain drugs used in anesthesia have very specific effects in combating recall (e.g., scopolamine, benzodiazepines), we still do not know whether routine use of these drugs will guarantee lack of intraoperative recall.^[73]