SEARCH

SEARCH BY CITATION

In this issue of Anaesthesia, Russell shows that, using the isolated forearm technique (IFT), patients' responses to verbal command during isoflurane-based anaesthesia [1] (just as recently demonstrated with propofol-based anaesthesia [2]) did not correlate well with EEG-based measures of ‘depth of anaesthesia’ (the bispectral index, BIS). The implication is that an IFT-positive response represents ‘consciousness’ (strictly, ‘wakefulness’ [3]) and therefore, the BIS is erroneous. The IFT appears a simple, yet powerful research tool and these results add to increasing concerns over limitations of EEG-based technology to monitor anaesthetic depth [4].

In this year of the 5th National Audit Project (NAP5) on Accidental Awareness During General Anaesthesia [5, 6], it is apposite that Russell's last paper [2] was accompanied by ‘pro’ and ‘con’ commentaries [7, 8]. These focussed mainly on technical limitations, namely whether IFT is cumbersome or potentially distracting in practice, with a paucity of trial evidence. Here, I focus on an entirely different aspect of IFT. Rather than issues of utility or efficacy, I wish to discuss a more fundamental question: what does it mean when a patient squeezes the anaesthetist's fingers to command during general anaesthesia?

The answer seems obvious. Russell states: “a patient's sensible intra-operative motor response to a command… would indicate that the patient is conscious” (my emphasis) [1]. But things are not always as they seem, as this conclusion is based on the knowledge that (a) there are no mind-altering drugs involved (e.g. anaesthetics) and (b) this response is consistent with other responses, to make it ‘sensible’. If, after some unspecified drug intervention, you squeezed my finger to command, but did not at all respond to being kicked in the shin, how should I properly interpret your inconsistent responses? This combination of responses is patently not ‘sensible’ and therefore, I might conclude, due to aberrations induced by the drug. This pertinent hypothetical scenario resembles the IFT-responders in Russell's papers [1, 2]. Readers may have noted that, remarkably, about a third of patients responded to simple command during apparently adequate anaesthesia; even more remarkable was the fact that none of them moved spontaneously with any purpose in response to surgery. The motor response to simple command cannot therefore be sufficient information to allow us to conclude that a patient is conscious.

Furthermore, an IFT-positive response implies that the anaesthetic dose is inadequate. One might predict, therefore, that intentionally administering this same inadequate dose should consistently result in persisting consciousness and therefore, a response to command. However, there are (at least) two situations when this prediction is not fulfilled, making interpretation very ambiguous.

The first is when patients have not received neuromuscular blocking drugs at all. A formal trial does not appear to have been undertaken, but at equivalent concentrations of anaesthetic that yield IFT-positive responses, non-paralysed patients do not respond to simple command. (Just one non-paralysed paediatric patient has apparently ever been described as doing this [9], but I have been unable to reproduce this in a series of adult patients so far (unpublished results)).

The second is during emergence from anaesthesia. Russell's own data show that at the end of surgery, when the administered anaesthetic concentration is much lower than when the patient responded to command during surgery, the patient no longer responds [1, 2]. If anaesthesia induces unconsciousness in a dose-dependent manner, why is a lower dose apparently more effective at eliminating response to command than a higher dose?

In other words, the really important questions in IFT studies relate not to the ‘positive’ effect as to why patients respond, but rather to the ‘negative’ effects:

  1. why do patients not signify their consciousness spontaneously during surgery?
  2. why do patients not respond to command at equivalent anaesthetic depths when not paralysed at all?
  3. why do patients not respond to command at lower concentrations of anaesthetic at the end of surgery?

Interpreting lack of spontaneous response to surgery

  1. Top of page
  2. Interpreting lack of spontaneous response to surgery
  3. The IFT: a Turing test?
  4. Using functional magnetic resonsance (fMRI) in the ‘isolated brain’
  5. Conclusions
  6. Competing interests and acknowledgements
  7. References

It is logically unsound to confine statements about patients' conscious state to the positive motor responses to command; some comment is also required on their lack of response to stimuli like surgery. Dangers lie in assuming that interpretation of these ‘negative responses’ exactly corresponds to interpretation of positive responses. Russell claims: “In non-paralysed patients [i.e. during IFT], response to command unequivocally indicates consciousness[1] (my parenthesis and emphasis). The temptation is to assume that the corollary is correct, namely that a lack of response to simple command indicates unconsciousness. This leap of logic would be falling into the trap of ‘denying the antecedent’ with respect to the relationship between responsiveness (R) and consciousness (C) [10, 11]. The fallacy is: ‘If R then C; if not R therefore not C’. This logic was underlined in the title of Sanders et al.'s review; ‘Unresponsiveness ≠ unconsciousness’ (the full logic would have read: ‘Just because responsive = conscious in one brain state, it does not follow that unresponsive = unconscious in another brain state') [12].

If you don't spontaneously cross the road to meet me I shouldn't conclude only that you must be drugged. You may simply not have seen me (even though I think you were in a reasonable position to do so). You may have chosen to ignore me (finding someone more interesting instead). I can only speculate why you did not cross. Similarly, it is impossible to know why IFT-responders do not move spontaneously to alert the anaesthetist that they are aware. We can only speculate:

  1. the level of surgical stimulus is inadequate or too uninteresting to hold the IFT-responder's attention (whereas the verbal command compels). This is akin to Carpenter's ‘cortical threshold’ model, wherein the subject is often aware of only one of several objects in the visual field, and saccadic eye movements are initiated to bring the focus of attention to that object [13, 14];
  2. the IFT-responder is dreaming (note that some authors regard dreaming as partial consciousness [12]) and satisfactorily incorporating surgical stimuli into their dream; a verbal command, however, is sufficiently specific and disruptive to evoke a response;
  3. the IFT-responder is in a state akin to Parkinson's disease where there is poverty of spontaneous movement, yet a near-normal movement to command [15]. Why this would be so is unknown, but anaesthesia may be influencing the ability to initiate movement, regardless of neuromuscular blockade [16];
  4. the IFT-responders spontaneously try to move the wrong (paralysed) arm, but on command are able to move the ‘correct’ (unparalysed) arm as instructed. (Russell describes patients ‘searching for their bodies’, aided by command) [7, 16];
  5. in the IFT-responder, the anaesthetic has abolished some aspects of consciousness (the ability to attend to surgery) whilst retaining others (attention to command). If this is the case, it is further intriguing that in Russell's studies, IFT-responders do not respond to a nerve stimulator [1] (which is used experimentally to generate exquisite pain [17]), but only to the anaesthetist's voice;
  6. in a variant of the above speculation, Sanders et al. have introduced the notion of ‘environmentally-connected consciousness’ as a state of partial consciousness induced at some levels of anaesthesia [12], arguing that the IFT-responder is sufficiently ‘connected’ to the environment to respond to verbal command, but to a degree insufficient to move spontaneously. One problem with this interpretation is that the reverse equally applies: namely, that the IFT-responder is ‘disconnected’ from the environment such as to fail to respond to surgery, but not to a degree that prevents response to verbal command.

The brain states implied by each of these speculations may not be mutually exclusive; however, none of them resemble what we recognise as ‘consciousness’. IFT-responders may have retained some limited capacity for responsiveness to simple command, but lack of spontaneous movement (and near-absence of any recall of events) indicates that anaesthetic drugs have not been entirely ineffective.

If a motor response to command is insufficient evidence alone of consciousness, and if non-response to stimuli is also difficult to interpret, what might then constitute sufficient evidence?

The IFT: a Turing test?

  1. Top of page
  2. Interpreting lack of spontaneous response to surgery
  3. The IFT: a Turing test?
  4. Using functional magnetic resonsance (fMRI) in the ‘isolated brain’
  5. Conclusions
  6. Competing interests and acknowledgements
  7. References

Alan Turing (1912–54), the Cambridge mathematician credited as breaker of the Enigma code during World War 2, is regarded as the inventor of the modern-day computer (termed then a ‘universal Turing machine’) [18]. Turing believed that as these machines became more complex, they would acquire a level of logical processing resembling human consciousness, and that this was testable using an ‘imitation scenario’ [19]. The ‘Turing test’ consists of an interrogating human, separated from a computer and a fellow human, each in separate rooms. The first human interrogates the two by whatever appropriate means and if or when s/he cannot distinguish the hidden human from the computer, then the computer has reached the same level of sophisticated thinking (consciousness) as the human. Thus we establish consciousness in another entity when we recognise (in the poet A. J. Ayer's words) the entity's responses as ‘empirically indistinguishable from normal human responses’ [20].

The IFT may be viewed as an opportunity to apply a Turing test. The nature of the unknown mental state induced by anaesthesia might be ascertained using ‘Turing-type’ interrogation. To conclude the patient is normally conscious we would require all of: (i) a spontaneous, goal-directed motor response; (ii) an appropriate motor response to simple commands; and (iii) appropriate motor responses to more sophisticated questioning. The last may involve interrogation about known likes/dislikes, feelings, or choice options. Clearly, the IFT-responders fail step (i); regardless of responses to steps (ii) and (iii), they fail the Turing test and therefore, they cannot be said to be normally conscious.

Anecdotally, practitioners who regularly employ the IFT claim to have had sophisticated ‘conversations’ with IFT-responders, but there are no extensive descriptions in the literature. Essential elements of step (iii) would be to ask IFT-responders questions to establish why they are not responding to the surgery, or command them to move their fingers in response to surgery. Nonetheless, an IFT-response is a dramatic event (as is any movement during anaesthesia) and the pragmatic reaction should be to increase the anaesthetic dose, because these movements (while falling short of signifying full consciousness) may represent a form of ‘pre-consciousness’; a propensity to becoming conscious if nothing were done. This hypothesis is potentially testable by the logically rigorous (but ethically interesting) protocol of making no intervention after an IFT-positive response.

The above argument assumes that the Turing test is watertight – but not everyone agrees. On its 50th anniversary, Saygin et al. described it as “one of the most disputed topics in artificial intelligence, philosophy of mind, and cognitive science[21]. We cannot tell if a mathematical problem has been solved by human or machine, or whether we are playing chess against a grandmaster or a computer; computers already ‘pass’ the Turing test in these restricted cases. There are (at least) two types of ‘reverse’ Turing tests that further confound interpretation. A human can pretend to be a computer to fool another human, or a computer can try to identify humans from machines [22]. This last paradigm, known by the acronym CAPTCHA (Completely Automated Public Turing test to tell Computers and Humans Apart), is frequently used to filter access to websites by presenting aspiring entrants with artistic depictions of a combinations of letters and numbers (e.g. image; only humans can correctly identify the distorted characters. Bottros et al. employed an inverse Turing test in which an anaesthetist (‘human’) estimated the BIS value from a raw EEG, comparing this with another (hidden) BIS value (BIS-test) by reference to a third (also hidden) reference BIS (BIS-reference) [23]. For the clinically relevant range of 40–60, human estimates agreed better with the BIS-reference estimate than did a second BIS-test monitor; or in Turing-terms, the BIS-reference was fooled into thinking that the human was the machine! [23]

Using functional magnetic resonsance (fMRI) in the ‘isolated brain’

  1. Top of page
  2. Interpreting lack of spontaneous response to surgery
  3. The IFT: a Turing test?
  4. Using functional magnetic resonsance (fMRI) in the ‘isolated brain’
  5. Conclusions
  6. Competing interests and acknowledgements
  7. References

Owen and colleagues attempted to assess whether patients ‘locked in’ by persistent vegetative state (PVS) might retain capacity to respond to command [24-27]. The experimental paradigm focussed on ‘isolated brain’, rather than ‘isolated forearm’, responses to command. First, fMRI scans in normal subjects identified specific activity patterns triggered by commands to imagine playing tennis versus walking in a house. Remarkably, they obtained the same unique signatures when several PVS patients were given the same commands. Tempting though it is to conclude that PVS patients can retain cognitive capacity for voluntary mental activity, Nachev and Hacker (a neuroscientist and philosopher, respectively) have noted the logical fallacy of such a conclusion [28]. It is committing the fallacy of ‘affirming the consequent’ in logic to conclude that, if in a normal brain state a command to imagine activity (A) results in neural fMRI signature (S), then observing S in another situation implies activity A. Expressed differently: it is invalid to argue ‘If A then S; if S therefore A’ (unless A was established as the only sufficient condition for S). For example, a heart rate of 35 min−1 is normally a sign of good health, yet this heart rate after a myocardial infarction indicates quite the reverse [29]. Moreover, in Owen's studies it cannot be known if the brain activity patterns were in fact unique to the commands given; they might also be generated by any number of other untested imaginations. Most importantly perhaps, it is unknown if the same brain activity patterns are obtained when either normal or PVS patients spontaneously imagine these tasks as compared with imagining them to command.

Conclusions

  1. Top of page
  2. Interpreting lack of spontaneous response to surgery
  3. The IFT: a Turing test?
  4. Using functional magnetic resonsance (fMRI) in the ‘isolated brain’
  5. Conclusions
  6. Competing interests and acknowledgements
  7. References

I do not conclude (in contrast to Russell [1, 2]) that during IFT, a motor response to simple command unequivocally implies consciousness (it is as equivocal of consciousness as an ‘isolated brain’ response to simple command in PVS). Instead, I propose the hypothesis that IFT-responders are in a specific state of partial anaesthesia (‘dysanaesthesia’), characterised by the ability to respond to simple command, but not to many other environmental stimuli such as surgery (i.e. functionally disconnected from these). This is largely consistent with notions of ‘disconnectedness’ or ‘unbinding’, previously proposed in different ways by others [12, 30]. This hypothesis is testable using sophisticated interrogation during IFT (‘Turing testing’), by brain imaging and perhaps by detailed analysis of patients' reports. With respect to the last, Noreika et al. examined patients' experiences recalled with different hypnotic agents and described some reports consistent with what I term ‘dysanaesthesia’ [31]. The patients' reports arising out of NAP5 may be illuminating in this regard [5, 6].

The following reasoning suggests that the proposed dysanaesthetic state is broadly acceptable as a patient experience. Russell tests for its presence using IFT, finding ~37% patients [12] to be IFT-responders; he then deepens their anaesthesia to avoid adverse effects and recall. Presumably, an equivalent proportion arises in the clinical practice of all anaesthetists but, because they do not employ IFT, they do not identify dysnaesthesia and they do not specifically deepen anaesthesia. Yet overall, the proportion of patients who report awareness during general anaesthesia is very low (~1:15 000) [5]. This means either that IFT-responsive dysanaesthesia is rarely associated with recall even if untreated or, even if it is recalled, it is generally benign. The third possibility is that it is so distressing that patients are too frightened to report it [32], but this now seems very unlikely. There are ~3 million general anaesthetics administered in the UK annually, of which ~1 million probably receive neuromuscular blockade [33]; therefore ~370 000 patients will be potentially IFT-positive, dysanaesthesia cases (undetected and untreated peri-operatively). If this experience is so distressing as to cause phobic avoidance, then over the last 10 years, almost ~4 million UK residents are silently suffering in this way. I do not think the problem (although a genuine one [34]) is of this order of magnitude. Therefore – reductio ad absurdum – untreated IFT-positive dysanaesthesia must be generally benign.

With only 14 of > 8000 senior anaesthetists in the UK apparently employing IFT, it is not a popular technique [5] (none of ∼300 consultant anaesthetists use it in the Republic of Ireland (NAP5, unpublished results)). As a research tool, however, it provides a useful – and arguably necessary – benchmark against which to assess more technically based monitors (such as EEG-based devices) [1, 2]. Russell's papers raise more questions than they answer about the fundamental nature of human consciousness and responsiveness, some of which are philosophical and explored above. If the anaesthetic community genuinely wishes to understand the mechanism of general anaesthesia, a full discussion of these questions is unavoidable.

Competing interests and acknowledgements

  1. Top of page
  2. Interpreting lack of spontaneous response to surgery
  3. The IFT: a Turing test?
  4. Using functional magnetic resonsance (fMRI) in the ‘isolated brain’
  5. Conclusions
  6. Competing interests and acknowledgements
  7. References

I thank my professorial colleague at St John's College, Oxford, Peter Hacker (Emeritus Research Fellow in Philosophy), for helpful discussions on the ideas contained above. I am Clinical Lead of NAP5: the views expressed are personal and not the views of NAP5 or the Royal College of Anaesthetists, or the Association of Anaesthetists of Great Britain & Ireland. No other competing interests or external funding declared.

References

  1. Top of page
  2. Interpreting lack of spontaneous response to surgery
  3. The IFT: a Turing test?
  4. Using functional magnetic resonsance (fMRI) in the ‘isolated brain’
  5. Conclusions
  6. Competing interests and acknowledgements
  7. References