Is It Impossible to "Measure" Conscious Feeling?

It has been an unusual pleasure to participate in this remarkable conference. I am grateful to Stuart Hameroff and Al Scott for the opportunity or challenge to speak before quite a few individuals whos works I hold in the highest respect. It has been excellent to hear so many of them summarized and updated this past week.

I've never before intentionally spoken in public on a subject I know so little about, nor without experimental data to show.  At the very end of the meeting we are all pretty used up from a week of hearing things sometimes different from what we think we know. I propose to provide semi-humorous relief.  Only "semi-": I am actually serious about someof the points I will try to make, but I hope you will  relax and enjoy and laugh with me at some of the purposely extravagant thought-experiments I mention.

I was trained as an engineer, and I am accustomed to validating first approaches to any problem with order-of-magnitude estimates of the quantities involved. That might not be possible at first, but the attempt at least shows where first effort is needed. In the riddles I try to address this morning, I can't see that any kind of quantitation is possible, maybe even in principle.  Something  essential is missing, at least from my thinking on the subject, and I am here to get that fixed.  [Note later: didn't get fixed. There was no comment, and this essay with figures was deleted from the MIT Bradford Press book. I remain today in the same state of ignorance.]  In pointing out the quantifiability of thresholds and onset times for awareness of specific stimuli, Professor Libet pointed to the importance of sticking to questions that can be decided by experimental observation. That seems the key to the cumulative success of scientific inquiry in the past few centuries. Some of the questions that I imagine partly motivate this meeting --- certainly some of the ones that keep nagging at me about subjectivity --- don't seem decidable by observation.

I have not been to a meeting on subjects like this since --- believe it or not --- before the Kennedy assassination. At that time I was a student of Frank Rosenblatt, at Cornell University with a wire-wrap gun in my hands for fixing Perceptrons. But Perceptrons didn't work very well, and after Minsky and Papert did them in I decided an attack on the problems of associative memory and perception was not timely, at least for me. I turned to other problems and went to other scientific
meetings.

My habit in those other meetings has been to come with a new clearly formulated question, criteria for laboratory decision, and an answer from experiments (my own, or published data from other labs) or at least from computations on a physically-based quantitative model. This time I arrive at the meeting with only questions, probably the same questions that motivate some of the concerns of many of the wonderfully diverse people in this room.

What intrigues me most about these questions is that I have been unable to think of any way even in principle to answer them empirically. That is usually the telltale indicator that I am confused, and that when I get my head together the so-called question will look in retrospect pretty silly and certainly not in need of an answer. This has been going on without noticeable progress since I was and adolescent worrying the priest and nuns in my school for answers.  They gave help only indirectly, by loan of Thomas Aquinas and other texts that did not satisfy. I got more help through reading and re-reading of Olaf Stapledon in high school, of psychophysical experiments and Vedanta obscurantism in college, then from university courses in psychology and philosophy ... never felt much wiser for any of it. Ten years ago I started asking my professional colleagues why is it that scientists never talk about these things? Why does "awareness" seem slipperier than superfluid helium yet more indisputably real than anything we deal with so successfully by science? There seemed then very little to read and I felt that colleagues felt uncomfortable with my insistent naggings. I found the posthumous book of my former teacher Max Delbruck and published a review of it, hoping to invite correspondence. The ferment of publication about conscious awareness in the last several years and now the surprisingly large and persistent attendance at this meeting reveal that I am not the only one who goes about his business as a scientist with such "unscientific" questions shamefully lurking in the closet. There has been a massive coming-out-of-the-closet comparable to the Gay Rights movement. Some of the people I most admire as scientists arehere this week brazenly  trying to contribute.

So last fall I tried to prepare an abstract for this meeting. I tried and tried and never solved any puzzle or paradox, and honorably declined to submit the draft abstract. I fell back into my usual pursuits and ended up building an optical tomography arrangement for the study of 3-dimensional waves like this

(3d color slides).

The  key features of such waves are topological, unlike waves familiar in acoustics, electromagnetics, and so on. What gives them their unique character and persistence is a topological subtlety that I first saw explictly drawn in Roger Penrose's pictures of his twistor. Their manifestation in the partial differential equations of excitable media involves no quantum mechanics but it is still the topology of phase singularities that underlies their remarkable behavior in the lab. Such waves occur in all excitable media ... e.g., Stuart Hameroff showed you Jim Lechleiter's slide, cast in the format I established for plots of such waves in chemically excitable media --- but it was in cortex of frog egg. Similar pictures have turned up more recently about the calcium waves by which  astrocytes in the hippocampus and in the suprachiasmatic nucleus communicate over long distances. The brain, crudely speaking, is an excitable medium. So is heart muscle. Such topologically peculiar electrical events occur in the thick wall of the heart at the onset of malignant arrhythmias. This was only my personal nutty theory when I described it in Scientific American 10 years ago, but lots of dog experiments in the last several years have borne it out as a common mechanism of sudden cardiac death. Here is a 2d cross-section of a pair of phase singularities induced by a topologically-premeditated experiment in the dog's heart (not pertinent to the real topic here: omitted from the transcript).

And how does "death" come about from nonlinear waves? When their topology is wrong, they spoil the coherent pumping of the heart. The result, even in perfectly normal hearts in perfectly healthy individuals, is cerebral ischemia, just as James Whinnery  showed us in Navy fighter pilots.  By stages that he showed so nicely, consciousness is lost, irreversibly after a few more minutes. In the unmonitored (and so uncorrected) cases this is called Sudden Cardiac Death. In this way I was led again to questions about consciousness.

Another main occupation of my laboratory for ten years was experiments on circadian rhythms, initially in the fruitfly Drosophila. These led to human body clocks and the suprachiasmatic nucleus of the hypothalamus, that Owen Flanagan mentioned in this connection. It regulates the onset of sleep and arousal from sleep. Just as I found in the heartbeat, body clock timing turned out to be dominated by topologically peculiar organizational principles, such as you see in this 3-d plot of my first experiments to expose the central singularities.  A plot like this recently proved central to the non-linear dynamics of human jet-lag. I use it during travels, so I have not experienced jet-lag for 10 years.  This singularity is also central to the sleep-wake cycle --- what I called the Tides of Consciousness in some papers at the outset of that decade.

It turns out, curiously, that the duration of human sleep adjusts spontaneously to the phase of the neural clock in the suprachiasmatic nucleus at the moment sleep begins. And there is a critical phase at which future sleep duration discontinuously increases by 12 hours, thus discontinuously jumping across half the circadian cycle in which consciousness is never spontaneously restored (wake-up does mnot occur). Abstract models make some sense of these things, and in fact predicted some of them. The recent experiments demonstrating phase singularities in the human body clock were done explicitly to test those theories. But the most that can come of such things is an understanding of the timing of consciousness, on a scale a million times coarser than Progessor Libet's.

These adventures taught me a lot about spatial and temporal patterns of activity in excitable media, and I am prepared to believe that conscious experience is a certain pattern of action potentials and other activities in such a medium. But nothing I have learned yet hints how any such pattern can be equivalent to subjective experience. Nothing hints what differences among patterns correspond to what differences of sensation.

I would feel more comfortable if I knew of a measure of the intensity or quality of consciousness, or at least an answer to the question of whether it could in principle be measured. Not just to find some qualitative YES/NO correlate of being conscious, like having the eyes open, or some feature of EEG, but I would like to know the necessary logical and dimensional features of any useful quantification of the extent or magnitude of conscious experience, on the supposition that if I can't even think of one then there is something basically wrong with my whole approach from the outset. And ideally, to conceive at least the germ of an idea how to compare such experiences, i.e., to measure, at least to be able to make order of magnitude estimates of the simplest essentials of conscious experience. Or at least to know why it can't be done.

Francis Crick and Christof Koch a few years ago presented a cogent discussion of how to go about such things. They settled on visual consciousness as the best place to start, because it is so wonderfully accessible to experiments. That seems to me absolutely right and I follow their work with the keenest interest and admiration. But I wanted to think about something that I imagined at least might be computationally simpler: just plain raw feels, like pleasure and pain, which might not even involve the neocortex. Exquisite sensations like toothache or orgasm probably don't involve a lot of computational complexity, but they do get right to the essence of awareness.  If one were to approach awareness from that direction the first challenge might be to recognize what would be appropriate units of measure or methods of comparison. Right there, whammo, that's where I already got stuck. So I didn't submit the attempted abstract.

Three months later ---this week --- durng sabbatical I had to be in Tucson for a student's PhD defense, so I looked in to see if the planned meeting was still on. You can imagine my surprise Wednesday morning to discover the meeting, and a name tag waiting for me, and my name on the Sunday morning schedule! Cold sweat. The mix-up apparently occurred because the addresses listed beside my name are pretty old and not in use this past year. I turned in a belated abstract Friday --- not much more than a confession of perplexity, only questions. My main hope is that they are familiar questions that have been solved, unknown to me. I hope that by gathering the courage to embarrass myself here, I might reap from you the benefit of swift forced-draft enculturation in this area that so fascinates and preoccupies me, and might even be told answers to my questions or at least scientifically practicable approaches to answering them, or might be introduced to a different perspective from which they seem silly and go away. Certainly if contact with this powerful group can't straighten me out, I may as well just go back to my laboratory and continue doing  what I can do and leave such deep riddles for people better able to handle them.

The Box

The riddles that got me stuck 3 months ago come from wondering why there could not in principle exist a box with knobs and dials on it that I could point at you and take a reading related to the intensity of your awareness. Maybe a lamp on the box would brighten when you feel more acutely aware, when you experience the throbs of a toothache or some less mentionable very positive sensation. And it would dim when I point it at a dog, and it would go out when I point it at a cadaver or a tree. I suppose. Well, that is pretty silly --- maybe its like imagining a box that would assay the extent to which the thing in front of it is a wheelbarrow or is useful, or would measure how much music there is in a concert hall or how beautiful it is. Am I just confused to fantasize even the possibility of such measuring the extent to which there is "someone" in front of The Box who is feeling something? Or might MRI and PET scanners be its early versions? If this is a confused fantasy, what are morals and ethics about?

Ethics

If I am confused, then so is our whole society, because we base lots of practical applications on the implicit idea that we have done such measurements, or could, and that we already know the results, at least to order of magnitude. These applications constitute the domain of morals and ethics, words that have never before appeared in my seminars. But then I have never given one on a Sunday morning either. Maybe there is some connection. I learned on Sunday mornings 40 years ago that it is not good to cause pain; hitting a nail with a hammer is OK, but don't hit people. People feel pain, anils don't. Kicking dogs is also bad, but not so bad as kicking your little brother.  Killing non-verbal animals or people outside your language group for food is perfectly OK, at least with government authorization, but I won't even speak of alternatives closer to home. I think most people are prepared to admit non-human animals to the fraternity of feeling processes, but it is thought that their intensity of awareness is less than ours, so quite a few units of inconvenience or pain on their part fairly balance just one unit on our part, while we collectively optimize the greatest good for the greatest number. One aspect of the Green movement is a dispute over the magnitude of that ratio. How many baby seals and distressed mother seals are worth one lady's pleasure in a nice coat? Or let's factor out our own paternal feeling for big round dark eyes in cute faces. Consider the question in a ludicrously extravagant form: what are we doing when we step on ants? We suppose that because individual ants don't have a lot of neurons and no cerebral cortex at all, probably they don't mind being stepped on, even in large crowds. If we thought ants were sensitive philosophers and poets or at least felt sensation as acutely as a comparable volume of our own neurons, we might step on them less. It seems ridicuous to ask what is the empirical evidence by which we know they are not and do not. Because we are so sure the ratio is so large, no such evidence is required, unless maybe it came to a question of one child's momentary pleasure vs some truly astronomical number of ants tortured for a long time. But there is the point: we implicitly use some astronomical number, but  no one will tell you the exponent and there is absolutely no empirical evidence. Our practical ecological relations to other living animals connote implicit quantification of pleasure and pain. Could it be refined to something scientific? Maybe this practical understanding based on thousands of years of experience could provide a jumping-off point for thinking it through in modern scientific terms and coming to some designs for quantifying pleasure and pain as one more avenue of approach to better experimental understanding of consciousness, or at least that aspect of consciousness.

Units

This stone (shown for dramatic effect, not reproduced here) found at the Rosetta delta of the Nile symbolizes the present impasse: We need a translation scheme for passing among three scripts: between brain stimuli such as Penfield and Rasmussen first explored, and subjective experience written in our private feelings, and modern MRI and PET observations. One necessary ingredient for any such scheme must be a way to compare and measure subjective feelings. What should be the units of any useful measure of intensity of feeling? Something per unit time I suppose. If a person were instantly frozen at absolute zero like Hans Solo in Star Wars, his consciousness would presumably be suspended. Consciousness is a process so its measures are rates. Consciousness depends on action potentials and so on, and the rate of a given kind of consciousness must fall toward zero as the rates of all those physical processses fall toward zero in a coordinated way. (Ceteris paribus, of course.) I can't imagine anyone arguing with that.

But if intensity of awareness is a rate, then it might further be inferred that there is some cumulative measure of conscious experience to be gotten by integrating that rate over time.  When all the same action potentials and neurotransmitter releases have transpired, the same experience has occurred. Regardless whether it took centuries or milliseconds, the subjective experience is the same. Then if you further repeat the same experience 5 times you have 5 times the total amount of subjective experience. For example in the early 1960's suppose hippocampal resection patient HM meets Brenda Milnor 5 mornings in a row, and each time HM learns in surprise that his hippocampal operation was years ago. Does such record-player repetitiousness really amount to cumulation of experience in any useful sense? Even if the experience were more acute, let's say some painful event in the dental chair, if there were no memory of its prior recurrences I am not sure that there is any meaningful sense to integrating the rate of subjective experience. To do a simple integration over time would be a bit like purporting to enhance the experience of Beethoven's Fifth by playing it five times over. Something has gone wrong in the logic here.

So I want to back off and just characterize experience as the rate of some process, and nevermind conjectures about its integrability in time. [Not all functions in dynamical systems theory are integrable, after all.] What about integrability in space? We have to say what process we measure, and that means delimiting its spatial locale, e.g., to include 1 person's body. But suppose I take in two persons. Suppose they were perfect xerox copies, both independently reacting to the same joke. Is the experience doubled? Consider doing the fantasy thought-experiment this way: take just one person and double the volume of each nerve cell. It would seem to me doubtful that this doubles the subjective experience. Or let each cell be duplicated alongside the original and wired in parallel; that seems almost the same thing, and also not equivalent to doubling the experience. Why should the answer be any different if these two parallel sets of cells, etc., were now teased apart into disjoint volumes and called two persons? I think the answer is still the same: the experience is not doubled in any sense more meaningful than one person without memory hearing it twice. I already had to back off from that one, so it seems I have to back off from spatial integration too.

But if I back off from spatial integration then I don't even know how to specify what process I am trying to quantify: if I can't say 'the process in that volume' e.g., inside a particular city under nuclear attack or a particular person's skin, or left hemisphere or anterior cingulate cortex, then what am I to do? Must I say "I seek a measure on this kind of process, wherever and whenever in the universe it may happen"? That begins to sound like treating all electrons in space-time as the same one, there being no grounds of individuality for distinguishing them as distinct. Then I just seek to associate a kind of sensation with a kind of physical process, and admit it makes no sense to think of quantification. But if it can't be quantified, can we do science about it?

I don't know how to answer that, but it seems clear that at least this irreducible minimum demand must be met, if consciousness is to be quantified: That different kinds of process must be associated with different subjective intensities (subject to the rate factor). If functionalists are to be believed, each kind of process must be associated with some kind of experience, maybe an exceedingly dim experience in most cases, even no experience at all, but evidently in some cases, a pleasure or pain of some such intensity as we all know privately. What kind of process is pain?  That is too abstract a question to even begin thinking about an answer. Morally you might not want to do experiments. Maybe you could do experiments about pleasure, e.g., to find out what brain nuclei or cortical projections constitute the sensation of orgasm, which certainly is one of the most dramatic conscious experiences. I would imagine it involves a minimum of complicated calculation, maybe not necessarily even involving the cerebral cortex.

Presumably the question could be answered by PET or MRI observations such as we heard about from Eric Reimar.  Visual neurophysiologists have done the equivalent for the sensation of red, and so on, so it is clear there is a non-mysterious, non-fantastical way to begin finding out what sorts of process correspond to what sorts of experience. It is reassuring to find even that small a hint that this is not all just crazy talk.

Feeling Machines

But strange ethical implications of functionalism arise immediately. If we ever find out a sufficiently abstract
characterization of processes and their relation to subjective affect, then presumably we could build and operate devices --- other than our children and lab animals --- which have subjective experience. Accepted moral principles would then make it a no-no to implement painful processes artificially, just as on natural organisms. It would be important then to find out if, perhaps, algorithms running today in supercomputers happen to constitute dim pain. Because it will get more acute about 2-fold every few years as machine performance doubles every few years. And maybe, if such rates are integrable over space, more intense as the number of such machines around the world increases even faster. Maybe computers will need some sort of anesthetic. If that is not a crazy idea, I never heard one.

Conversely, I guess it would be virtuous to implement processes associated with joy and ecstasy. As a science fiction theme one might explore the consequences in a future culture dedicated to the construction and maintenance of such machines, much as past cultures have been dedicated to multiplying the abundance and energy of the human physiological machines that we know so well experience suffering joy under conditions that we all try to improve, at least for ourselves. Just as Egyptians 4000 years ago invested most of their social effort in pyramids, so another society might invest in comparably demanding machines: ever faster, nuclear-powered artificial minds specialized for the experience of pleasure at the highest wattage engineers can maintain. It could be considered as much a moral
imperative as minimizing pain.

This seems such a dubious proposition because there is no way, even in principle, for any such engineer to verify that that his labors do indeed produce any subjective experience. I think this is what Christoph Koch meant in his introductory transparency, item 5, if I remember right. He said there are some phenomena that simply lie outside the grasp of science. Since we can't say anything about them, we may as well remain silent about them and do things we can do: his point 4. Very sensible. He makes no fuss over it, but accepts it as obvious and gets on about the very productive work that he can do. That is also what I have done with my professional life. But this past several months, while going about those conservative activities, with the coming of this meeting in the back of my mind, I fantasized about reaching for what Christoph dismissed as point 5.

I have never before taken seriously the notion that there is anything real that cannot be inquired into by empirical means. But this month's exercise in deviation from the straight and narrow track has led me straight into such wierd thoughts that if anyone else had been foolish enough to say them at this meeting, I would write him off as bonkers and be amazed that such a distinguished group of scientists put up with him. I entreat you not to write me off. As scientists we draw our
paychecks for not writing off lunatic notions just on account of their sheer fantasticalness:  instead we try to make them into distinct paradoxes and use those as touchstones for detecting some mistake in our assumptions. It is a method that has worked for all of us in the past; in my own case it led to some pretty neat laboratory discoveries about biological clocks and spatio-temporal organization in excitable media. I would like to understand more about consciousness, in other words about subjective affect, if that is another attribute of excitable media. But is it really in principle not measurable, in other words not comparable between instances? Are there really questions --- meaningful questions to which we presume to know unverifiable answers with the largest practical applications such as war-making, pest
extermination, the meat industry, and slavery --- that simply cannot be inquired into in a scientific way? This has come to bother me so much in recent months of first gropings that I am willing  to make a laughingstock of myself here for the sake of getting some pointers from this distinguished group toward clearer thinking on these issues. I like this problem because I know it is going to teach me something, whichever way it comes out: I will either recognize a new way to
experimentally approach a subject of the greatest importance, or I will learn that there is at least one real, substantial problem, not a mere word game, which the present scientific world view simply cannot grasp. It seems to me not an ignorable problem because we act as though we did know quantitative values, and we use them in practical applications.

Leave-taking

You see now why I didn't submit an abstract, at least until challenged by finding my name and title on the printed schedule four days ago. Now I've accepted that challenge and I am ready to be led away to the booby hatch --- or maybe told I've simply fallen into standard philosophical errors, and that everyone here long ago passed that stage and agreed on their resolution. Tell me what that resolution is and then without any kind of violence I'll go peaceably back to my
optical tomography experiments.

Thank you very much for listening, and let me again thank our organizers for bringing us all together for this unique meeting.
 

Postscript 8 weeks later: I collected quite a lot of pointers to literature at the meeting and I have now read most of them (not yet all). Many are very scholarly, every edifying, very instructive about many things of interest. But I do not sense any improvement in my ability to deal with the phenomena I tried to discuss here. I cannot do any new tricks, I do not see any crucial experiments to try, nor do I see why the nagging questions I tried to articulate here should be dismissed as mere foolishness. Thus my experiment in coming-out-of-the- closet is not finished. I cannot compose a scientific paper on this topic, and to discipline myself against pretentiousness in that direction, I omit figures and citations (which would all be off the subject in any case, mere magician's misdirections) to ensure that this tape transcript gives no such appearance. I nonetheless venture a little further out by here preserving in a public place this plea for The Missing Ingredient, still hopeful that it might someday route unexpected illumination in my direction.

My address: winfree@email.arizona.edu