Sep 23 2009

Learning in the Comatose

Neurologists are often confronted with patients who have a disorder of consciousness and are asked to predict their probability of recovery. The more severe the neurological damage the easier prediction is, but there are many patients in the gray zone – they have enough damage that they may never have a significant recovery, but there is also enough preserved function that they may improve – even to the point of crossing an important functional milestone, for example being able to communicate.

Traditionally we have relied upon the clinical exam to determine the the level of impairment – for example, whether a patient is in a persistent vegetative state (PVS) or a minimally conscious state (MCS). However, there are limits to the exam (patients, for example, may be unable to move because they are paralyzed rather than due entirely to lack of consciousness) and a recent study has shown that errors in the distinction between PVS and MCS are not uncommon.

Making this distinction is also becoming increasingly important as neuroscientists develop possible techniques to treat patients with disorders of consciousness to help them recover. For example, patients have been treated with transcranial magnetic stimulation or with implanted chips to stimulate brain activity. It is not inconceivable that within 10-20 years we will be able to augment brain function in those in a MCS and help them regain significant consciousness.

Efforts are therefore underway to develop better tools for assessing residual conscious function in patients in a PVS or MCS and to correlate these findings with recovery (for better prediction). For example, functional MRI scanning has been used to look for conscious processing in those in an apparent PVS.

Now, researchers have employed a clever technique to look for conscious processing in comatose patients – exploiting the learned response, or Pavlovian conditioning. They played one of two tones to patients, one tone was followed by a puff of air against the cornea which could produce a blink response. They established a baseline response in conscious patients and compared this to a group of patients in a PVS or MCS and also to an unconscious control group – intact patients under anaesthesia with propofol. They looked for a conditioned response to the tone, anticipating the coming puff of air.

They found a significant conditioned response in the patients with disorders of consciousness, more than the propofol group but less than the conscious control group. This is especially interesting because up to now it has been thought that a conditioned response requires conscious processing.

The authors acknowledge two basic ways to interpret these results. The first is that learning a conditioned response is not dependent upon conscious processing. It may be mediated partly by unconscious processing. I think this may be very likely, especially in light of other evidence for unconscious learning.

The other possibility is that some of the patients in this study had more preserved consciousness than was evident on clinical exam. This too is plausible, and in fact both explanations may have some truth and may vary from patient to patient. Preserved conscious processing, in my opinion, should not be equated with conscious awareness. Patients in PVS and MCS may have some conscious processing, but not enough to generate any kind of awareness that they will remember.

While these results are interesting, the biggest question is whether or not these findings can be used to make predictions about individual patients. The researcher therefore looked at this. First, they compared their findings to various clinical assessments, but only found one that correlated with the demonstration of a conditioned response – brain atrophy. This suggests that their results are meaningful – they are actually telling us something about brain function. However, I don’t think this helps us distinguish between conscious vs unconscious processing contributing to the conditioned response – as overall brain atrophy will correlate with damage to both types of processing in the brain.

Further, and most significantly in my opinion, the researchers compared the degree of conditioned response in individual patients to their outcome over the next 6 months to 2 years, and they found again that there was a good correlation. Conditioned response was 86% predictive of later improvement in clinical function – some improvement in the specific grade of coma.

As always I must caution that we are talking about a modest improvement in neurological function – patients went from a PVS to a MCS, or to a higher degree of MCS, or in some cases to being severely disabled but able to communicate. I don’t want to minimize the significance of such improvement to the patient and their family, but I also don’t want to give the impression that some of these patients “woke up” and had anything close to normal neurological function.

This study needs to be replicated to see if this effect is reliable, and if so it may add to our options for assessing patients with disorders of consciousness. There is unlikely to be one definitive test of consciousness any time soon, but if we can combine the clinical exam, anatomical studies, electrical activity (EEG), response on functional MRI scanning, and this new effect – the ability to generate a conditioned response – we will get more and more accurate in our assessments.

Increased accuracy will in turn enable us to give families better information with which to make decisions about ongoing care, and eventually may help us predict which patients will be good candidates for experimental treatments, like implantable chips.

14 responses so far

14 Responses to “Learning in the Comatose”

  1. Xalxuffaschon 23 Sep 2009 at 10:32 am

    This is interesting to me for a few reasons. First, one of my loves of experimental psychology is the single subject design. Effectively each person in the experiment serves as a control for their own changes in behavior. Here it is a little moot, as there is no reason to expect a tone eliciting an eye-blink in any subjects. The other thing single subject designs stress is repeated measurement, which is just as it sounds take a ton of measures of the behavior.

    More importantly it the assumption of a lot of human behavior requiring conscious awareness of the task. There is a lot of evidence in the animal world of Pavlovian (respondent) conditioning in animals that could not in any form be argued to be conscious. In much of the human literature it is hard to separate out the behavior from conscious awareness because the conditioned response they are trying to elicit is extremely easy to notice. I don’t know of a single person who would not know they just blinked when air was puffed in their eye or kicked their leg when you whack the patellar tendon. Interestingly the same kind of thing applies to “conscious” or operant behavior. The things that we are most noticeably in control of are very obvious, like arm movements. However, people can have direct control over behavior for something they were never even possibly aware of. There was as series of studies done by Ralph Hefferline in the 60’s doing just that, gaining control of a very measurable behavior that was not detectable to the person. (Attached is the url of a representative artticle) I just think we need to be very careful on saying whether some behaviors require conscious processing if there is no possible way to separate the two.

    http://www.pubmedcentral.nih.gov/articlerender.fcgi?tool=pubmed&pubmedid=13963702

  2. Eric Thomsonon 23 Sep 2009 at 3:17 pm

    The fact that the body of a spinalized cat can learn via Pavlovian conditioning suggests that it isn’t a good indicator of consciousness. 🙂 (Reference below)

    Using general anesthesia control was nobel, but it is also hard to interpret. It’s like pulling the plug on your computer to show that its CPU can’t do addition. While the anesthetized subjects didn’t learn anything, it is hard to interpret because general anesthesia doesn’t just affect consciousness, but pretty much everything.

    Reference
    Durkovic RG, and Damianopoulos EN. Forward and backward classical conditioning of the flexion reflex in the spinal cat. J Neurosci 6: 2921-2925,

  3. llewellyon 23 Sep 2009 at 4:03 pm

    Upon reading the title, I assumed this article would be about those dreaded after-lunch classes …

  4. kvsherryon 23 Sep 2009 at 10:29 pm

    Steve-

    You wrote, “…They established a baseline response in conscious patients and compared this to a group of patients in a PVS or MCS and also to an unconscious control group – intact patients under anaesthesia with propofol.”

    I have two questions about this. 1. What was the Riker score that they were sedated to? Because I have used Propofol boluses to bring some of my more unstable neuro patients to a ‘1’ while I do invasive things like deep suctioning and these patients do nothing at that time.
    2. Were these patients only on Propofol or were they also on another anesthesia adjunct?

  5. HHCon 24 Sep 2009 at 2:45 am

    I like the post about conditioned learning in a comatose state. I worked with a comatose patient once which we cued to shake when he responded to his name being called. Perhaps this is simple recall but this was useful for assessing his mental status. I would think this post could be filed with the scientific articles about conditioning with respect to physiological processes we don’t normally control but we are surprised when we do.

  6. Joeon 24 Sep 2009 at 8:12 am

    Prediction is always difficult, especially about the future. I just heard a woman interviewed (she has a new book) and she related the story of her mother, who had bipolar disorder. The mom was unconscious after carbon monoxide poisoning and a neurologist said she would probably not recover. In the end, she recovered better than ever- the bipolar disorder is gone.

  7. Wicked Ladon 24 Sep 2009 at 12:53 pm

    “Preserved conscious processing, in my opinion, should not be equated with conscious awareness.”

    Interesting. What’s the distinction?

  8. nohayeson 24 Sep 2009 at 12:59 pm

    “I just heard a woman interviewed (she has a new book) and she related the story of her mother, who had bipolar disorder. The mom was unconscious after carbon monoxide poisoning and a neurologist said she would probably not recover. In the end, she recovered better than ever- the bipolar disorder is gone.

    My mother-in-law is bi-polar and I can attest that fact that bi-polar individuals can seem completely and totally normal for a while. There can be years between “episodes”.

    I’m very skeptical of the claim made above.

  9. nohayeson 24 Sep 2009 at 1:25 pm

    “The fact that the body of a spinalized cat can learn via Pavlovian conditioning suggests that it isn’t a good indicator of consciousness. (Reference below)”

    I did a websearch on the article and I don’t understand what I found. Does “spinalized” mean that the study utilized the cat’s spine, as in… the rest of the cat wasn’t there?

    “Using general anesthesia control was nobel, but it is also hard to interpret. It’s like pulling the plug on your computer to show that its CPU can’t do addition.”

    I understand what you’re saying but it’s wrong in two ways.

    First, you can’t say it’s like “pulling the plug” without having an expertise in the exact pharmocology (is that the right word here?) of propofol.

    Here again, I did a websearch and, wikipedia (if this article is credible) says that it causes “potentiation of GABA-A receptor activity”, which is totally over my head, and that it blocks the sodium-channel, which would inhibit signal transmission along an axon (Intro Anatomy). It also said that it interacts with the endocannabinoid system, which is also over my head. Any specialists out there want to give me a free education on the topic? 🙂

    Second, the propofol group was the baseline, not the control. If you wanted a baseline of processer activity in a damaged computer, performing your diagnostic on a computer that isn’t plugged in would be a great way to do it.

  10. Eric Thomsonon 24 Sep 2009 at 2:50 pm

    Nohayes: in spinalilzed cats the spinal cord is cut so it is no longer connected to the brain. As they say in the article I cited, they sectioned the spinal cord at T10 (10th thoracic vertebrae). The spinal part of the cat still learns (that is, the part not connected to the brain). This is quite well established, so Pavlovian conditioning is not a marker of consciousness.

    Whether we describe the anesthetized case as a control or baseline is beside the point: the point is that the lack of learning under anesthesia doesn’t show that consciousness is required for Pavlovian conditioning.

    If you want to show that someone lacking X can’t do S, it isn’t enough to remove X, Y, Z, and Q from that person and show they can’t do S. It would be like pulling the plug on your computer to show the CPU can’t do addition. You aren’t just disabling the CPU, but all sorts of other things. I wasn’t being literal in my pulling the plug metaphor, obviously. If you want to show that a person without a brain can’t dance, you wouldn’t do it by removing the kidney, brain, and liver.

    So, my computer analogy was just an analogy, but it is a good one even though obviously there are disanalogies. General anesthetics don’t simply inhibit consciousness, but have global effects on brain function (they can even inhibit reflexes if given in large enough doses). In this case, propofol, it acts as an inhibitor of activity (I did a quick PubMed search, and it seems the locus of activity isn’t known: some general anesthetics act at more specific regions than others, but most are fairly general).

  11. Steven Novellaon 24 Sep 2009 at 3:01 pm

    Eric,

    The authors do acknowledge that the propofol group is a problematic control for their experiment because of exactly what you say – the more generalized effects of the anaesthetic beyond just decreasing consciousness.

    Steve

  12. Eric Thomsonon 24 Sep 2009 at 3:54 pm

    I am overall impressed with the paper. That they did any controls is impressive, much better than that paper where they saw responses in PVS patients to requests to play tennis (what if you did see the same response in anesthetized patients).

    So the control was good in that if they had observed such conditioning that would have been conclusive evidence that consciousness wasn’t required. Not observing such conditioning leaves things rather open (though as Steven and others and I have already pointed out it is pretty clear that classical conditioning doesn’t require consciousness).

    As you (Steven) already stressed, what is most impressive is the correlation between learning and recovery of consciousness. The clinical side of this paper is much more interesting than the theoretical implications it might have for consciousness research.

  13. nohayeson 24 Sep 2009 at 5:57 pm

    “If you want to show that a person without a brain can’t dance, you wouldn’t do it by removing the kidney, brain, and liver.”

    Gotcha.

    Thanks for the clarification on spinalized cats.

    Additionally, it would appear from Dr. Novella’s comment that the propofol group was a control. I guess I had it wrong.

  14. artfulDon 24 Sep 2009 at 6:19 pm

    I guess you got the analogical inference wrong too.

Trackback URI | Comments RSS

Leave a Reply

You must be logged in to post a comment.