HETEROPHENOMENOLOGY VERSUS CRITICAL PHENOMENOLOGY
Max Velmans, Department of Psychology, New Cross,
Phenomenology and the Cognitive Sciences (in press)
Following an on-line dialogue with Dennett (Velmans, 2001) this paper examines the similarities and differences between heterophenomenology (HP) and critical phenomenology (CP), two competing accounts of the way that conscious phenomenology should be, and normally is incorporated into psychology and related sciences. Dennett’s heterophenomenology includes subjective reports of conscious experiences, but according to Dennett, first person conscious phenomenena in the form of “qualia” such as hardness, redness, itchiness etc. have no real existence. Consequently, subjective reports about such qualia should be understood as prescientific attempts to make sense of brain functioning that can be entirely understood in third person terms. I trace the history of this position in behaviourism (Watson, Skinner and Ryle) and early forms of physicalism and functionalism (Armstrong), and summarise some of the difficulties of this view. Critical phenomenology also includes a conventional, third person, scientific investigation of brain and behaviour that includes subjects’ reports of what they experience. CP is also cautious about the accuracy or completeness of subjective reports. However, unlike HP, CP does not assume that subjects are necessarily deluded about their experiences or doubt that these experiences can have real qualities that can, in principle, be described. Such experienced qualities cannot be exhaustively reduced to third-person accounts of brain and behaviour. CP is also reflexive, in it assumes experimenters to have first-person experiences that they can describe much as their subjects do. And crucially, experimenter’s third-person reports of others are based, in the first instance, on their own first-person experiences. CP is commonplace in psychological science, and given that it conforms both to scientific practice and common sense, I argue that there is little to recommend HP other than an attempt to shore up a counterintuitive, reductive philosophy of mind.
Keywords: heterophenomenology, critical phenomenology, Dennett, Velmans, first person, third person, psychology, science, consciousness, qualia, mind, subjective report, experience, behaviourism, behaviourism, reductionism
How can one understand the nature of conscious experience? To begin with, one needs a description of the phenomenology of consciousness and that, in turn, requires an investigative method, an appropriate descriptive language and an embedding theoretical framework that is sufficiently shared by a community of consciousness researchers for communal research to progress. Traditionally, it is assumed in such phenomenologies that experiences are, in essence, subjective and must, at least in the first instance, be described from a first-person perspective. In the present paper I focus on the current status of some of these methods in psychology. It is worth noting however that in its classical philosophical form, Phenomenology (with a capital P) goes further than advocating the use of first-person methods, suggesting that “the pure and transcendental nature and meaning of phenomena, and hence their real and ultimate significance, can only be apprehended subjectively” (Shorter Oxford English Dictionary, 2002 – my italics). While this advocates a “reduction” of a kind, it is opposed to any reduction of consciousness to a set of scientific facts or third-person theories. One the contrary, Phenomenology is a “method of reduction whereby all factual knowledge and reasoned assumptions about a phenomenon are set aside so that pure intuition of its essence may be analysed” (Ibid).
on the other hand has developed a more critical stance towards phenomenal
description and method, driven by the rather chequered history of trying to
apply phenomenological methods in practice—for example, the problems with
“experimental introspectionism” developed by Wilhelm
Wundt in the first experimental psychology
laboratory, at the
A little bit of history
For Wundt, the task of psychology was the scientific study of the "mind" and, for him, the "mind" was identical to consciousness. In his laboratory, controlled, measurable stimuli were used to bring about given conscious states. Rather like chemical compounds these states were thought to have a complex structure and the aim of experimentation was to analyse the entire structure into its fundamental, component elements. This was to be achieved by trained subjects carefully introspecting and reporting on their detailed, moment-to-moment experiences.
This categorising of conscious states presented a formidable task and extensive inventories were developed, for example in the laboratories of Külpe (1901) and Titchener (1915). However, in the early years of the 20th century, this programme fell into disrepute. How can one give a definitive list of the contents of consciousness? In his analysis of this period Boring (1942) noted that Külpe's laboratory discovered less than 12,000 distinct sensations, whereas Titchener's laboratory discovered more than 44,435! These differences appeared to be largely due to differences in how subjects had been trained to attend to and describe what they experience, and without agreement in the field about the fine details of the introspective method, disagreements between different laboratories were difficult to settle. Worse, given the privacy of individual experience and the sole reliance on subjective reports, introspective findings were difficult to falsify.
Other reasons for the demise of introspectionism had more to do with the prevailing, positivist intellectual climate. Psychologists were keen to reformulate their discipline along the lines of natural science. John Watson (1913) for example, argued that the subject matter of psychology should not just be restricted to humans, but should include other animals. The introspective method does not allow this for the reason that other animals cannot make verbal reports about what they experience. Nor, he argued, does it make much sense to speculate about what they experience. Psychology, therefore, should confine itself to a study of overt behaviours, the stimuli which produce them, and observable physiological functions such as the behaviour of nerves, glands, muscles and so on. Thus refocused, psychology would become a behavioural form of biological science, and it was hoped that this would make it more suitable for practical purposes—for example for the purposes of social engineering.
In short, "Psychology as a behaviorist views it is a purely objective experimental branch of natural science. Its theoretical goal is the prediction and control of behavior. Introspection forms no essential part of its method nor is the scientific value of its data dependent upon the readiness with which they lend themselves to interpretation in terms of consciousness" (Watson, 1913, p 158). Indeed, "The time has come when psychology must discard all reference to consciousness; when it need no longer delude itself into thinking that it is making mental states the object of observation..." (Ibid, p163).
Methodological versus ontological behaviourism.
Methodologically, there were clear advantages to be gained from this refocusing of psychological enquiry. Organism’s responses may be measured with precision and, being publicly observable, allow intersubjective agreement or the settling of disagreement. Watson's commitment to behaviourism, however, was more than methodological. In his view mental events are irrelevant to psychological enquiry—and some mental events are in any case nothing more than the behaviour of internal organs. For example, thinking (Descartes' prime exemplar of nonmaterial mind) is, for Watson, nothing more than minute muscular activity of the vocal tract.
Clearly, if inner variables such as consciousness or mind reduce to behaviour, and behaviour is entirely under stimulus control, then nothing is lost by restricting psychology to the study of responses and the stimuli that produce them. In this way methodological behaviourism, which is basically a thesis about how psychological research should be carried out, and analytical behaviourism a reductive thesis about the ontological nature of consciousness or mind, are mutually supportive. Consequently, behaviourist psychologists often adopted aspects of both positions.
Skinner (1953), for example, argued that talk about conscious, mental events is mostly vague and metaphysical. For example, if someone forgets something (an observable behaviour) we speak, metaphorically, of his ‘mind’ being ‘absent.’ Other mental accounts, he claims, simply restate the facts of observed behaviour and are, therefore, redundant. For example, "He eats because he is hungry" is, arguably, no more informative than to say "he eats". Such attempts to translate statements about mental events into statements about observable responses exemplify Skinner's ontological behaviourism.
Difficulties with behaviourist analyses of consciousness.
Watson's theory that thought is nothing more than the minute movements of articulatory muscles was heroically put to the test by E. M. Smith (1947) who temporarily paralysed all his muscular activity with curare. He reported afterwards that his ability to think and remember while paralysed was unimpaired - thereby falsifying the "minute muscle movement" theory of thought.
Ontological behaviourism is, in any case, counterintuitive. As Chappell (1962) pointed out, "If behaviorism were true, I could find out that I myself had a pain by observing my behavior, but since I do not find out that I have a pain, when I do, by observing my behavior ... behaviorism is not true" (p 10).
In any case, we are often not able to determine the mental states of others even if they make no attempt to conceal these states and their overt behaviour is clearly visible. Again, as Chappell comments, "If behaviorism were true I could always in principle find out when you had a pain by observing your behavior, but since I cannot always find out, even in principle, that you have a pain when you do, whereas I can always observe your behavior it follows that behaviorism is not true" (Ibid, p 10).
There are also many instances where overt behaviour is inconsistent with what one thinks, feels or otherwise experiences. For example, one may experience hunger without eating—which makes it difficult to argue that hunger is nothing more than eating behaviour. One can also eat in spite of the fact that one is not hungry; one may conceal or lie about one's intentions and so on. Even if one tries to express some experience faithfully in overt behaviour it is not always possible to do so. For example, the phenomenology of experience cannot always be unambiguously and exhaustively described in words ("translated into verbal behaviour"). This was, in fact, one of the stumbling blocks of introspectionism.
Given the many dissociations between conscious states and overt behaviour, the attempt to reduce conscious states to overt behaviour seemed ill conceived. As a methodology, radical behaviourism also failed to fulfil its mandate, “the prediction and control of behaviour”, for the reason that complex behaviours such as verbal descriptions of visual scenes (verbal responses to visual input stimuli) were not entirely under stimulus control, and could not, therefore be reliably be predicted or controlled. Given such problems it is not surprising that radical behaviourism, like Wundt’s experimental introspectionism, has largely passed into history.
Are mental states just "dispositions" to behave?
However, there were subtler versions of behaviourism, that to some extent internalised the causes of behaviour that were not so easily dismissed, such as Gilbert Ryle's (1949) suggestion that mental states reduce not to overt behaviour but rather to "dispositions to behave". This was elaborated by D.M. Armstrong (1968) who tried to apply such a dispositional behaviourist reduction to the qualia of consciousness, arguing (a) that mental states (of whatever kind) are nothing but states of a person apt for bringing about certain sorts of behaviour, and (b) that states of a person apt for bringing about certain sorts of behaviour are nothing but states of the brain. In this way, Armstrong tried to eliminate phenomenal qualia by means of a two-stage reduction, combining dispositional behaviourism with a form of physicalism. For example, according to Armstrong, perception is just,
".. a matter of acquiring capacities to make physical discriminations within our environment" (p83), and ".. nothing but the acquiring of true or false beliefs concerning the current state of the organism, body and environment" (p209). "Our perceptions, then, are not the basis for our perceptual judgements, nor are they mere phenomenological accompaniments of our perceptual judgements. They are simply the acquirings of these judgements themselves" (p226).
If Armstrong is right, there is nothing mysterious about perceptual qualia, as there is nothing about perceptions that is additional to the capacity to make discriminations based on the acquiring true or false beliefs about the organism and environment. Such a reanalysis, he argues, has two advantages. It both captures the "inner character of perception" and creates "a logical tie between the inner event and the outer behaviour" (Ibid, p248).
It is in the above historical context that Dan Dennett’s later heterophenomenology is best understood. According to Dennett (2003) heterophenomenology is just a conventional, third person, scientific investigation of brain and behaviour that includes subjects’ reports of what they experience. It is simply “a phenomenology of another not oneself.” The verbal responses of subjects (in which they describe their experiences) allow them “to collaborate with experimenters—making suggestions, interacting verbally, telling what it is like [for them to have experiences]” (p20). And, “This third-person methodology is … the sound way to take the first person point of view as seriously as it can be taken.” (p19).
How can a third-person methodology be the sound way to take the first-person point of view as seriously as it can be taken? Because it gives you a “catalogue of what the subject believes to be true about his or her conscious experience. This catalogue of beliefs fleshes out the subject’s heterophenomenological world … not to be confused with the real world. The total set of details of heterophenomenology, plus all the data we can gather about concurrent events in the brains of subjects and in the surrounding environment, comprise the total data set for a theory of human consciousness. It leaves out no objective phenomena and no subjective phenomena of consciousness.” (p20, my italics)
The crucial point is of course the last one. Does (third-person) heterophenomenology leave out the (first-person) subjective phenomena of consciousness? Dennett claims not—but like the radical and dispositional behaviourists before him, in order to make good that claim he has to argue that the subjective phenomena of consciousness are not what they seem to be. Dennett asks, “Just what kinds of things does this methodology commit us to? Beyond the unproblematic things all of science is committed to (neurons and electrons, clocks and microscopes…) just to beliefs—the beliefs expressed by subjects and deemed constitutive of their subjectivity (p20, my italics).”
Again, the crucial point is the last one—and it involves exactly the same reduction of experiences to beliefs proposed by Armstrong (1968) (see above), although in Armstrong’s case the beliefs are about the world, not about the experiences as such. While Dennett is willing to listen to what people have to say about their experiences, he is not prepared to believe what they say. What a subject believes to be true about his or her phenomenology is “not to be confused with the real world.” (see above) Because their subjective worlds are not real, subjects’ beliefs about their qualitative nature are false. Consequently, such beliefs are not about some independent subjective world or subjectivity, as there is no such world for them to be about—which makes subjects’ beliefs entirely, “constitutive of their subjectivity.” As brain science progresses, such beliefs about the nature of mind will be swept away much as science swept away the idea that evil spirits are the cause of illness and disease (personal communication – see Velmans, 2001).
Note that, although Dennett sometimes claims that he is not denying the existence of consciousness (cf Velmans, 2001), he is unquestionably denying the existence of what most individuals have in mind by the term “consciousness”, that is subjective, phenomenal consciousness. Dennett makes this eliminative intent perfectly clear in his other writings, for example when he points out that "Philosophers have adopted various names for the things in the beholder (or properties of the beholder) that have been supposed to provide a safe home for the colors and the rest of the properties that have been banished from the external world by the triumphs of physics: raw feels, phenomenal qualities, intrinsic properties of conscious experiences, the qualitative content of mental states, and, of course, qualia, the term I use. There are subtle differences in how these terms have been defined, but I am going to ride roughshod over them. I deny that there are any such properties. But I agree wholeheartedly that there seem to be." (my italics - Dennett, 1994, p129, see also Dennett, 1991, p372).
In sum, although Dennett tries to present heterophenomenology as a theory-neutral, scientific approach to the study of consciousness, he is far from neutral about the status of subjective qualities or qualia. He has already made up his mind, and, like his behaviourist predecessors, the third-person investigative method that he advocates and his eliminative ontology for first-person experiences are mutually supportive. Heterophenomenology “leaves out … no subjective phenomena of consciousness” only if there are no first-person phenomena to leave out.
In various ways, Dennett’s heterophenomenology (HP) bears a family resemblance to the critical phenomenology (CP) that I have described in Velmans (1999, 2000a,b)—although I would argue that the latter more closely reflects a practice that is theory-neutral about subjective, first-person evidence—as well as being common practice in psychological research. As with HP, CP includes a conventional, third person, scientific investigation of brain and behaviour, which includes subjects’ reports of what they experience. As with HP, the verbal responses of subjects (in which they describe their experiences) allow them to collaborate with experimenters—making suggestions, interacting verbally, telling what it is like for them to have experiences and so on. However CP differs from HP in that it does not assume that subjects are necessarily deluded and scientifically naïve about their experiences, or doubt that it really is like something for subjects to have experiences, or that these experiences have qualities that can, in principle, be described. On the contrary, the default assumption in CP is that subjects really do have subjective, conscious experiences, and if one tries to reduce such first-person experiences to third-person descriptions of brain and behaviour, one is likely to leave something out that is of central importance to the nature of consciousness. Unlike HP, CP is also reflexive. It is not just “a phenomenology of another not oneself.” Rather, it is a phenomenology of another and oneself. CP takes it for granted, for example, that experimenters have first-person experiences and can describe those experiences much as their subjects do. And crucially, experimenter’s third-person reports of others are based, in the first instance, on their own first-person experiences. This last point is largely missed or at any rate dismissed in discussions of heterophenomenology, and I will return to it below.
Why call this approach “critical phenomenology” rather than just “phenomenology”? Partly, to dissociate it from the classical, philosophical versions of Phenomenology mentioned at the beginning of this essay, in which third-person methods and third-person science are sometimes treated with the same distain as heterophenomenology treats first-person methods. Instead, critical phenomenology adopts a form of “psychological complementarity principle” in which first- and third person accounts of the mind are treated as being complementary and mutually irreducible. A complete account of mind requires both (Velmans, 1991a,b). While CP assumes subjective phenomena to be real, it remains cautious about phenomenal reports in that it assumes neither first- or third-person reports of phenomena to be incorrigible, complete, or unrevisable, and it remains open about how such reports should be interpreted within some body of theory. CP is also open to the possibility that first-person investigations can be improved by the development of more refined first-person investigative methods, just as third-person investigations can be improved by the development of more refined third-person methods. CP also takes it as read that first- and third-person investigations of the mind can be used conjointly, either providing triangulating evidence for each other, or in other instances to inform each other. Third-person observations of brain and behaviour for example can sometimes inform and perhaps alter interpretations of first-person accounts (subtle differences in first-person experience for example can sometimes be shown to have distinctive, correlated differences in accompanying neural activity in the brain). Likewise, first-person accounts of subjective experience can inform third-person accounts of what is going on in the brain—indeed, without such first-person accounts, it would be difficult to discover the neural correlates of given conscious experiences. In adopting the view that subjective conscious experiences are real, but our descriptions and understanding of them revisable, CP follows an entirely conventional critical realist epistemology that is widely adopted in science.
HP versus CP
Does heterophenomenology or critical phenomenology more closely describe actual psychological research? As noted above, if matters are viewed solely from the third-person perspective of an external observer, restricted (in both data and theory) to observations of the brain and (verbal or other) behaviour of a subject, there is little to choose between them. This is hardly surprising, as if, in behaviourist style, one adopts a pretheoretical commitment to including only third-person observable data and theory based on that data in one’s science, then third-person methods should leave nothing out. In debate with Chalmers and Goldman, Dennett (2001) challenged them to name any experimental investigation of consciousness that did not follow the heterophenomenological method. In an online dialogue with Dennett (Velmans, 2001) I reversed the challenge, asking Dennett to name any scientific investigation of consciousness that did not follow the critical phenomenological method. As one might predict (given the similarities in our accounts of third-person methods), Dennett was unable to do so. However, CP (unlike HP) is not confined to third-person methods. Do psychologists ever use first-person methods (in isolation, or in combination with third-person methods)? Of course we do. When setting up a laboratory experiment, say on perception, the very first thing one usually does is to try the experiment on oneself. For example, in setting up a study on visual masking the first thing one might check is whether the time of onset of the mask, its duration and its form prevent one’s own conscious perception of the stimulus. If it does not work for oneself, it probably won’t work for others. When studying the conditions under which illusions occur, if they don’t occur for oneself, one needs to alter the experimental conditions, and so on. These preliminary phases are commonly followed by pilot studies and carefully designed experimental trials that adopt third-person methods, which include subjective reports. Even so, the default assumption in such experiments is that subjective reports are reports about what subjects experience, and that (barring individual differences) these experiences resemble the experiences of the experimenters when they tried the experiment themselves. While it is true that, during the behaviourist years, some psychologists tried to present the entire procedure in third-person, behaviourist language that made no reference to what subjects or experimenters experience, this exclusion of subjective experience had more to do with what passed for “political correctness” at the time, than with what was really going on (see discussion in Velmans, 2000a, ch4).
Dennett (2003) recognises that experimenters commonly draw on their first person experiences in psychological research in the manner described above, and that this is not covered by heterophenomenology (in the sense that he defines it). Consequently he tries to argue that such first person investigations are preliminary to science rather than integral to it. For him, “Lone-wolf autophenomenology, in which the subject and the experimenter are one and the same person, is a foul, not because you can’t do it, but because it isn’t science until you turn your self-administered pilot studies into heterophenomenological experiments.” (p23) In my view this gerrymandering of the borders of science has more to do with the preservation of Dennett’s philosophical position than with the nature of psychological science. Scientists are experiencing human beings as much as their subjects. So why should something as obviously empirical as trying an experimental treatment on oneself before trying it on others be preliminary to rather than part of good science? In medical research, for example, this can be a brave (and honourable) thing to do. And what exactly makes observations of the effects of treatments on oneself less “scientific” than observing such effects on others? Aren’t both kinds of observation based on what we experience of others or ourselves? (See Velmans, 1999, 2000 ch8 for a detailed analysis of how the empirical method incorporates subjective experience.)
Returning to Dennett’s challenge to Chalmers and Goldman (above), there are in fact many examples of studies that explicitly use first-person methods (in the way that CP describes) where a translation into the exclusively third-person descriptions of HP look very forced. According to Dennett (2003) “Neither Chalmers nor Velmans has responded to my challenge to describe an experiment that is licensed by, or motivated by, or approved by ‘critical phenomenology’ but off limits to heterophenomenology” (p26). Actually, it was Dennett that did not respond to the examples offered at the end of our dialogue in Velmans (2001). One simple example is the study of pain. Pain is often taken to be a prime exemplar of a “mental event” within philosophy of mind. Yet it is one of the most highly researched subjects in medicine. Over the period 1966 to 1998, for example, the Medline database listed over 148,000 publications on pain and its alleviation. As noted above, Dennett explicitly denies the existence of qualia. But if the qualia of pain were not real (just false beliefs), why spend so much time, money and effort on their removal? And how could aspirin remove a false belief? Given its subjective nature, how is pain measured? By subjective reports involving verbal rating scales, numerical rating scales, visual analogue scales and questionnaires such as the McGill Pain Questionnaire. Significantly, even in 2005 no valid “objective” measure of pain experience (in the form of a physiological index) exists.
Equally difficult for heterophenomenology, over the last 20 years or so, there has also been a renewed interest in the development of first-person research methods that focus on “what it is like” for subjects in various situations of interest to investigators, for example with the expanded use of phenomenologically inspired qualitative methods that are used both in isolation and in conjunction with triangulating third-person quantitative methods in psychological research (see for example, Denzin & Lincoln, 2000, for a review). Complementary first- and third-person methods are also widely used (without embarrassment or apology) in much of neuropsychology, for example in the search for the neural correlates of consciousness using neuroimaging techniques. There have also been in-depth re-evaluations of how the use of such combined first- and third-person methods can be refined, for example in the field of neurophenomenology and more generally in cognitive neuroscience (see for example, readings in Varela and Shear, 1999, Velmans, 2000b, Jack & Roepstorff, 2003, 2004). All of these developments adopt a critical phenomenology that assumes first-person subjective experience to be both real and investigable—without minimising the methodological difficulties that such research can sometimes present.
In sum, one can be “critical” or cautious about how well or how reliably a subject can communicate the qualia of his or her subjective experience in experimental settings, without for a moment doubting their existence or claiming them to be something completely different to how they seem. One can study colour vision without doubting that there really are colours and that subjects really experience them, study pain without doubting that subjects really feel pain, and so on. Given this, Dennett’s heterophenomenology with its accompanying “qualia denial” looks like nothing more than an attempt to shore up his counterintuitive, eliminativist philosophy of mind.
Armstrong, D.M. (1968) A Materialist Theory of Mind,
Boring, E. (1942) Sensation and Perception in the History of Experimental Psychology,
Chappell, V.C. (ed.) (1962) Philosophy of Mind,
Dennett, D. (2001) The fantasy of first-person science. http://ase.tufts.edu/cogstud/pubpage.htm
Dennett, D. (2003) Who’s on first? Heterophenomenology explained. Journal of Consciousness Studies, 10(9-10), 10-30.
Ryle, G.(1949) The Concept of Mind,
Skinner, B. F. (1953) Science and Human Behavior,
Smith, S. M., Brown, H. O., Toman, J. E. P. (1947) ‘The lack of cerebral effects of d-tubocurarine’, Anesthesiology 8: 1-14.
Titchener, E. B. (1915) A Beginner’s Psychology,
F. and Shear, J. (1999) First person approaches to the study of
Velmans, M. (1991a) Is human information processing conscious? Behavioral and Brain Sciences 14(4): 651-669.
Velmans, M.(1991b) Consciousness from a first-person perspective. Behavioral and Brain Sciences 14(4): 702-726.
Velmans, M. (1999a) Intersubjective science. Journal of Consciousness Studies 6(2/3): 299-306.
M. (2000a) Understanding Consciousness.
Velmans, M. (2000b) A
psychologist’s map of consciousness studies. In M. Velmans (ed.) Investigating Phenomenal Consciousness: New Methodologies and Maps,
Velmans, M (2001) Heterophenomenology versus critical phenomenology: A dialogue with Dan Dennett. On-line debate at http://cogprints.soton.ac.uk/documents/disk0/00/00/17/95 /index.html.
Watson, J. B. (1913) Psychology as the behaviorist views it. The Psychological Review XX: 158-177.