Vol 1 (1994)

Table of Contents

Book Reviews

Consciousness Reconsidered by Owen Flanagan Full Text
Valerie Gray Hardcastle, Peter E. Pruim
The Computational Brain by P. S. Churchland and T. J. Sejnowski Full Text
Bruce Bridgeman
Stage Effects in the Cartesian Theater: A review of Daniel Dennett's Consciousness Explained Full Text
Kevin B. Korb
Consciousness and Behavior by Benjamin Wallace and Leslie E. Fisher Full Text
Bruce Bridgeman
Unilateral Neglect: Clinical and Experimental Studies edited by Ian H. Robertson and John C.Marshall Full Text
Robin Walker
Getting the Ghost out of the Machine: A Review of Arnold Trehub's The Cognitive Brain Full Text
Luciano da Fontoura Costa
Consciousness: Philosophical Issues Volume 1, edited by E. Villanueva Full Text
Matthew Elton



Vagueness, Semantics, and the Language of Thought Summary Full Text
Richard DeWitt
In recent years, a number of well-known intentional realists have focused their energy on attempts to provide a naturalized theory of mental representation. What tends to be overlooked, however, is that a naturalized theory of mental representation will not, by itself, salvage intentional realism. Since most naturalistic properties play no interesting causal role, intentional realists must also solve the problem of showing how intentional properties (such as representational properties), even if naturalized, could be causally efficacious. Because of certain commitments, this problem is especially difficult for intentional realists such as Fodor. In the current paper I focus on the problem as it arises for such realists, and I argue that the best-known solution proposed to date is inadequate. If what I say is correct, then such intentional realists are left with an additional and substantial problem, and one that has generally not been sufficiently appreciated.
Searle on the Brink Summary Full Text
Selmer Bringsjord
In his recent The Rediscovery of the Mind John Searle tries to destroy cognitive science and preserve a future in which a ``perfect science of the brain'' (1992, p. 235) arrives. I show that Searle can't accomplish both objectives. The ammunition he uses to realise the first stirs up a maelstrom of consciousness so wild it precludes securing the second.
A Thoroughly Empirical Approach To Consciousness Summary Full Text
Bernard J. Baars
When are psychologists entitled to call a certain theoretical construct "consciousness?" Over the past few decades cognitive psychologists have reintroduced almost the entire conceptual vocabulary of common sense psychology, but now in a way that is tied explicitly to reliable empirical observations, and to compelling and increasingly adequate theoretical models. Nevertheless, until the past few years most cognitive psychologists and neuroscientists avoided dealing with consciousness. Today there is an increasing willingness to do so.
But is "consciousness" different from other theoretical entities like "working memory" or "mental imagery"? Some argue that under no circumstances can empirical science speak of consciousness as such, while others claim that the scientific goal is "knowing what it is like to be a bat" --- to share an organism's conscious experience (Nagel, 1974). The Bat Criterion is ominously reminiscent of the protracted debate on the consciousness of ants and amoebas that caused so much uneasiness in psychology around 1900. It seems to demand that we first solve the mind-body problem as a condition of doing sensible science, and thereby creates the risk of endless, fruitless controversy. The endless philosophical debate about consciousness helped trigger the Behaviorist revolution about 1913, which threw out the baby of consciousness with the bathwater of perennial, circular debate. We've been that way; let's not go back to it.
This paper maintains that the position of behavioristic denial is far too restrictive, but that the Bat Criterion is far too demanding --- that in fact, we only need to specify comparable pairs of psychological phenomena that differ only in the fact that one member of any pair is conscious, while the other is not. This "method of contrastive analysis" is a generalization of the experimental method, with consciousness as a variable whose interaction with other psychological and biological phenomena can be assessed in standard ways. As usual in science, this strategy is pragmatic: If it appears to yield sensible results, it can be a stepping-stone toward further understanding (Crick and Koch, 1992; Ericsson and Simon, 1984/1993).
This paper describes five sets of well-established pairs of phenomena that meet these criteria. Others are presented elsewhere, with more of a theoretical interpretation (Baars, 1983, 1988, 1993). Here I simply want to show that any adequate theory of conscious experience must satisfy these demanding but achievable empirical constraints.
Baars Falls Prey to the Timidity He Rejects:Commentary on Baars on Contrastive Analysis Summary Full Text
Selmer Bringsjord
Baars (1994) affirms Crick and Koch's (1992) position that the timidity most cognitive scientists show in the face of consciousness is ridiculous. Unfortunately, all three succumb to a variation on the timidity they deprecate. Furthermore, Baars' own method, ``contrastic analysis,'' is at odds with the computational conception of mind that dominates contemporary cognitive science.
1.1 Baars is right to affirm Crick and Koch's (1992) position that the timidity most cognitive scientists show in the face of consciousness is ridiculous. Unfortunately, all three succumb to a variation on the timidity they deprecate: they shrink back from trying to tackle the daunting ``explanatory chasm,'' eloquently encapsulated by T.H. Huxley (1866):
How it is that anything so remarkable as a state of consciousness comes about as a result of irritating nervous tissue, is just as unaccountable as the appearance of Djin when Aladdin rubbed his lamp.
1.2 Baars seems to say at one point that some investigators of consciousness have as a goal ``knowing what it's like to be a bat.'' But of course there are no such people; the search for a scientifically respectable account of consciousness is hardly vespertilian. What some investigators do aim at is conquering the explanatory chasm. And it's true that in order to conquer it we'll have to discover what information about the physical world explains the what-it's-like-to- be-an-X form of consciousness, and how it does so. [This is the observation with which Jackson's (1982) color scientist argument begins. It ends with the conclusion -- via reasoning I regard to be fundamentally correct -- that there is no information about the physical world which could even in principle suffice; see (Bringsjord, 1994a, 1992).] What Baars must mean, then, is that a scientifically respectable account of consciousness needn't meet the Bat Challenge: the account needn't allow us to conquer the explanatory chasm. But this, alas, is a timid position.
1.3 As Block (in press) points out, Crick and Koch turn timid too in the face of the explanatory chasm. Their famous theory, in a nutshell, is that a synchronized 35-75 hertz neural oscillation in the sensory areas of the cortex is at the heart of the consciousness with which Nagel and Jackson are concerned. But what, prey tell, is the principled and informative connection between synchronized 35-75 hertz neural oscillation and what it's like, say, to be a human writing under a deadline? The connection seems entirely arbitrary. In fact, it seems easy enough to imagine a being who enjoys such oscillation but is thoroughly devoid of an inner life. And of course anyone [like Bringsjord (1994b) and Block (in press)] who argues that zombies -- creatures who have brains operating as ours but no subjective awareness whatsoever -- are possible, will find the C-K theory to have absolutely nothing to do with the nature of what-it's-like-to-be-an-X consciousness.
1.4 Baars will no doubt reply that unlike C and K he is avowedly timid; he means to dodge the Bat Challenge (= BC). For that matter, Baars premeditatedly drops from consciousness discourse not only what-it's-like-to-be-an-X consciousness, but also the concept of self with which it's intertwined (Section 2.4). But why should we take these stunning shortcuts? Baars tells us -- correctly -- that tackling BC will preclude separating consciousness from self, but he never really tells us why this is objectionable. He also tells us that BC is ``ominously reminiscent of the protracted arguments about the consciousness of ants and amoebas that caused so much trouble in psychology around 1900'' (Section 2.5). But this will hardly justify the Baarsian shortcut for those inclined to take some risks. If we had been this risk-averse on the question ``What is a proof?'', we wouldn't have seen the Euclid-Aristotle-Hobbes-Frege-Turing progression from utter ignorance to today's high-speed computers [see (Glymour, 1992) for a nice account of this story].
1.5 Baars may at this point make explicit that which is mere subtext in this paper, viz., that avoiding a discussion of the self is desirable, because confronting the issues involved will lead to a protracted and difficult investigation. But this is the same anemic justification all over again; it's just the ``ants and amoebas'' worry in another guise.
1.6 Perhaps Baars will admit that he is certainly no gambler when it comes to progress on the explanatory chasm, while at the same time reminding us that his own approach does play significant dividends. Baars' approach, of course, is `contrastive analysis.' Is it fruitful? Well, what contrastive analysis yields is a list of properties to be associated with a conscious state; Baars gives us an up-to-date list in Section 8.6. The list includes the following properties: having adequate duration, being perceptual or imagistic, and being dependent upon voluntary control to retrieve and maintain them. But how do we know that this list isn't adventitious? How do we know that these properties have something to do with the nature of consciousness? I'm not pointing out that this list fails to allow us to bridge the explanatory chasm. I made that point above, and the Baars-Bringsjord dialectic has gone beyond it. My point now is rather that the dividends contrastive analysis is supposed to pay, in the absence of an attack on the Bat Challenge, may be exceedingly small -- because the list of properties which we associate with consciousness via this method may simply reflect the situation local to our own planet. Consider, for illumination of this point, the following parable:
1.7 Suppose that NASA's SETI (Search for Extraterrestrial Life) project pans out in the year 2011: in that year we discover intelligent life beneath the surface of Mars. It turns out that Martians, in addition to being wondrous spelunkers, are also silicon-, not carbon-based. Not only that, but Martians enjoy conscious states which lack most if not all of the properties on the contrastive analysis list: their conscious states are evanescent, aren't perceptual or imagistic (because Martians don't have sensors like ours, nor do they enjoy a correlate to our powers of visual imagery), and are independent of voluntary control (Martians, if you like, are at the mercy of conscious states which happen to them). Now, such beings are peculiar, but they don't seem to be impossible. In fact, the possibility that there are such creatures is precisely what has moved most of us to reject type materialism (put roughly, the view that mentation can't possibly happen if the physical substrate in question is different from ours) in favor of computational brands of functionalism (put roughly, the view that mentation requires the correct computational flow of information, which may be instantiated in any physical substrate) which form the core of traditional symbolic AI and cognitive science.
1.8 Of course, Baars may reply that the objectives he has in mind for contrastive analysis are exceedingly humble. He may not only agree that this method fails to bridge the explanatory chasm, but he may also agree that this approach fails to reveal the nature of consciousness. What, then, would be his thesis? It would presumably be that contrastive analysis produces a list of properties which happen to be associated with the brand of consciousness enjoyed by homo sapiens. To this I don't in the least object; I readily grant that he establishes this thesis. But for those who want to conquer the explanatory chasm, or at least live by the ``cognition is computation'' credo of traditional AI and cognitive science, a credo which is supposed to extend to intelligences that needn't be human, I find the thesis, again, unpalatably timid.
Working Definitions of "Non-Conscious":Commentary on Baars on Contrastive Analysis Summary Full Text
Greg Davis
Baars (1994) contends that we must ask answerable questions about consciousness, and that to do so will require definitions of consciousness that permit "contrastive analysis". I endorse this general approach, but find several of Baars claims of processing without consciousness, unconvincing. I show that a more cautious definition of "non-conscious" than Baars' need not impede experimentation and is more likely to enjoy universal agreement.
1.1 In his A Thoroughly Empirical Approach to Consciousness, Baars (1994) argues convincingly for accepting working definitions of consciousness that may be manipulated in the laboratory. Certainly, if experimental study of consciousness is to progress, some of its more mysterious aspects must be temporarily left to one side. Instead, experimentation must focus initially on more answerable questions, progressing from crude definitions of the object of study, to more refined ones. However, though I agree absolutely with Baars diagnosis, I am concerned that some of his claimed examples of non-conscious processing will fail to convince many in psychology. This area of research desperately needs to gain respectability amongst those working in other fields; a cautious but pragmatic approach to empirical data will optimise its chances of doing so.
1.2 For "contrastive analysis" (comparison of conscious representations with closely related non-conscious ones), evidence of processing without consciousness is required, and Baars draws on an wide selection of empirical evidence in support of his approach. He considers four types of representations to be without consciousness: these are referred to below as "unattended", "subliminal", "implicit" and "pre-perceptual" processes. Whether each of these types of processing may be productively considered "non-conscious", is now discussed.
2 Unattended Processing
2.1 Some experimental psychologists (e.g. Velmans, 1991) have been prepared to treat "unattended" processing as non-conscious and Baars is apparently prepared to make the same step. On the other hand, several authors including Crick (1994), see attention as simply "enriching" consciousness and believe that unattended stimuli may achieve some degree of conscious representation. This controversy does not appear to be reaching a conclusion, and it is difficult to see how it might do so, given that our current manipulations of both attention and consciousness are so very primitive. What is clear is that attention itself is rather poorly understood, and that equating "unattended" with "non-conscious" therefore promises to be a fruitless, possibly misleading exercise.
3 Subliminal Processing
3.1 When psychologists talk about "subliminal" processing, this is usually in reference to backward masking paradigms. When a word is flashed momentarily on a screen and an uninformative pattern superimposed upon it after a sufficiently short delay, subjects report being unaware of the presence of the stimulus and may not be able to guess above chance whether a word was presented or not. Under these conditions there is evidence that word meaning is still processed (e.g. Marcel, 1983). Theoretically however, masking may allow conscious perception but prevented registration of the word in sensory memory. Since at the moment we cannnot know whether a masked stimulus was not seen, or simply not remembered, it is difficult to see how backward masking studies may have any direct bearing upon our understanding of subjective experience (see Holender, 1986 for a review).
4 Implicit Processing
4.1 Several claimed demonstrations of processing without consciousness involve complex rule learning (see Schacter, 1987 for a review). When presented with stimuli which conform to very complex rules, such as artificial grammars, subjects perform surprisingly well but insist that they were unaware that they had processed such rules. These paradigms do not require stimuli to be presented briefly and so it may be assured the subject has seen and attended to the experimental stimuli. However, many researchers remain to be convinced that the existence of implicit processing has been established (e.g. Perruchet and Pacteau, 1990) and particular concerns persist regarding the sensitivity of measures of conscious processing in such tasks. Even if this problem were to be solved, it is not clear that implicit learning studies can tell us anything we do not already know. For example, we are not aware of the complex rules that our visual system uses to provide us with an accurate idea of an object's lightness, but we do perceive the subtle conscious effects of these processes; enhanced performance in implicit rule learning tasks might therefore simply be another example of consciousness reflecting the results of rule-processing, not the rules themselves. I conclude that equating "implicit" with "non-conscious" is unlikely to be informative at present.
5 Pre-Perceptual Processing
5.1 Probably the most promising approach to investigating conscious experience is the study of "pre-perceptual" processes. Our experience of visual phenomena seems to be modulated by aspects of perceptual set (Gilchrist, 1977) and in many cases only a subset of representations of a visual feature (e.g. motion) may correlate with what we see at any one time. Those representations which are not directly affecting visual experience may be considered as "pre-perceptual", and it is hoped that systematic study of when processes do and don't directly impinge upon subjective experience, may confine possible criteria for conscious representation. A simple, but convincing case (Logothetis and Schall, 1989), occurs when an upwardly moving grating is presented to one eye and a downwardly moving one to the other eye. In this case only downward or upward motion is seen by human observers, rather than a summation of the 2 signals. Only a subset of motion sensitive neurons in V5 of the rhesus monkey correspond to the monkeys experiences evidenced in its behaviours. Such approaches benefit greatly form the wealth of physiological and psychological data already available, and from the likelyhood of easily interpretable results; they are already providing useful insights into the processes underlying subjective experience.
6 Conclusion
6.1 In conclusion, I am convinced that conscious experience may be studied productively by systematic comparison of "pre-perceptual" processes with their conscious counterparts. However, I have also voiced the concerns of psychologists regarding the validity of claims that backward masking, attention or implicit learning studies may be directly informative about our experience. The scientific community is with some reason still sceptical that consciousness may be studied experimentally; it therefore seems appropriate to use the most cautious working definitions of "conscious" and "non-conscious" which do not impede experimentation unnecessarily.
6.2 Our criteria for a working definition of "non-conscious" should aim to maximally satisfy both the goals of universal acceptability in the scientific community and of permitting experimental study. As I have argued above, criteria which treat "unattended", "masked" or "implicit" representations as "non-conscious" will not be universally acceptable, whereas any criterion framed in terms of intentionality or qualia clearly cannot satisfy the aim of allowing experimental study of consciousness. If a salient aspect of a stimulus does not appear in detailed, guided, verbal report, given unrestricted viewing by non- brain-damaged subjects, it may be universally agreed that this aspect is not conscious. Paradigms used to study pre-perceptual representations are both able to utilise this most cautious working criterion for "non-conscious" and yet provide workable paradigms for use in the laboratory. I suggest that this more cautious approach should waylay the fears of those who see experimental study of consciousness as impossible, and yet it may still encourage progress to be made.
A Thoroughly Empirical First-Person Approach to Consciousness: Commentary on Baars on Contrastive Analysis Summary Full Text
Max Velmans
According to Nagel, bat consciousness is ``what it is like to be a bat.'' According to Baars (1994), we will never know what it is like to be bat, so this approach to consciousness does not allow the science of consciousness to progress. Rather, the nature of consciousness as such should be determined empirically, by contrasting processes which are conscious with processes that are not conscious. The present commentary argues that contrastive analysis is appropriate for finding the processes most closely associated with consciousness; but it will not illuminate the nature of consciousness as such. Unlike bat consciousness, human consciousness is accessible to humans introspectively (or through communication with others). Consequently, a complete science of consciousness needs to relate introspective, first-person accounts of consciousness to third-person processing models of the brain.
1.1 In his ``Thoroughly empirical approach to consciousness'' Baars contrasts Nagel's question ``What is it like to be a bat?'' with the ``method of contrastive analysis.'' According to Baars the ``bat criterion of consciousness'' asks what consciousness as such is like in an unanswerable way - unlike systematic contrasts between processes which are or are not conscious which enable one to discover what consciousness is like by experiment and inference.
1.2 One could hardly take issue with the usefulness of contrastive analysis, as the method (Hume's ``method of difference'') is as old as empirical science. How else, other than through the careful addition and/or subtraction of specific conditions could one ever find out the necessary and sufficient conditions (the causes) for given events? In the field of consciousness research this is the appropriate method for determining the necessary and sufficient conditions for consciousness within the brain specified either in physical (neurophysiological) or functional (information processing) terms.
1.3 However, whether contrastive analysis is the appropriate method for determining the nature of consciousness as such, or whether it will ever be possible to claim that consciousness is just a construct within an information processing theory (section 1.5) are different matters.
1.4 Nagel's bat criterion of consciousness was intended to be illustrative of the irreducibility of the ``first-person perspective.'' If one views a bat from a ``third-person'' external observer's perspective one might be able, in principle, to observe everything that there is to observe about bat physiology and behaviour. But one cannot observe how the bat experiences the world. No matter how extensive one's knowledge might be of bat physiology and behaviour one cannot know what it is like to be bat from the bat's point of view. Consequently, for Nagel, bat consciousness could never be just a construct within an information processing model. But he accepts that this irreducibility poses an insuperable problem for a science of consciousness - and if it does, Baars might be right to suggest that we could discuss bat consciousness forever without progress.
1.5 There are many students of animal consciousness that would take a somewhat different view. One might, for example, agree with Nagel that it is not possible to experience what it is like to be a bat, and that bat experience is not reducible to information processing, and yet hope to infer something about the nature of what the bat experiences on the basis of what one can observe. One might, for example, be able to infer whether or not the bat is in pain (see Dawkins, 1990, for an extensive discussion of this issue).
1.6 More to the point, the ``bat criterion'' does not apply to human beings. To understand what it is like to be a bat one might have to be a bat; but to understand what it is like to be a human being one has to be a human being. Baars' argument for reducing consciousness to an information processing construct relies on all more direct routes to an exploration of consciousness being closed. It might be that the consciousness of bats is inaccessible to humans. But it is absurd to suggest that humans have no access to human consciousness. To explore different conscious states one can experience them for oneself, or one can communicate about those experiences with other human beings. In short, for humans, contrastive analysis needs to be set not against a ``bat criterion'' but against a ``human criterion'' of consciousness.
1.7 In itself, of course, this would not be sufficient to establish a science of consciousness. One has to establish systematic methods of first-person enquiry and first-person accounts have to be related to third-person accounts (some of the complexities are discussed in Ericsson and Simon 1984; Velmans 1991a,b, 1993a,b). This is far from being an unscientific dream - the relating of first- to third-person accounts is as old as experimental psychology, in the study of psychophysics, perception, and so on. Indeed, some of Baar's criteria of consciousness come from first-person analysis although they are presented as the fruits of third-person empirical science. For example, his generalization that ``conscious experiences are most clearly articulated in the case of perceptual or quasi-perceptual phenomena'' derives from an introspective examination of such phenomena, as does the distinction between focal and fringe consciousness (Section 8.3). This applies also to ``the unity of consciousness'' (Section 8.4) and to the relative evanescence of images (Section 8.6). We think of consciousness as unified only because we generally experience it that way, images are generally experienced to be less well-articulated and stable than percepts, and so on.
1.8 In sum, a third-person contrastive analysis does not compete with a ``human criterion'' of consciousness and it certainly does not legitimise the reduction of consciousness to a construct within some cognitive theory (cf Velmans, 1991a,b, 1993a). Rather, first-person (introspective) and third-person (neural and information theory) approaches are complementary. A complete psychology of consciousness requires both.
Consciousness Requires Global Activation:Commentary on Baars on Contrastive Analysis Summary Full Text
James Newman
Baars' contrastive analysis approach offers an essential framework for differentiating conscious processing from the myriad unconscious functions carried out by the mind/brain. In applying this approach it is important to understand that consciousness is not something other than, but something in addition to the unconscious processes that precede and follow the momentary focus of awareness. We have argued elsewhere that neurologically this something is activation via a global attentional matrix which both: 1) controls access to consciousness by competing unconscious processors; and 2) integrates the multimodal representations of those processors which gain momentary access into unified, conscious percepts.
1.1 I think Baars (1994) does a credible job of outlining an empirically based set of criteria for the study of conscious phenomena. Indeed, one wonders at the timidity of psychology during the middle decades of this century (see Baars' initial quotation from Crick and Koch, 1990) given that consciousness was the central concern of experimental psychologists during William James' time.
1.2 Perhaps the greatest stumbling block for this earlier consciousness psychology was its a priori rejection of the existence of unconscious thought processes (Baars, 1986). It is this basic flaw in approaching consciousness which Baars' contrastive analysis serves to rectify. Indeed contrastive analysis, in my view, is more than an approach: it should be a fundamental constraint in consciousness research. Conceptually this constraint appears dichotomous: yes/no, either/or, but this can be misleading when we turn to functional and biological considerations.
1.3 This brings me to my first point: Consciousness is not something other than unconscious thought processes, it is something in addition to unconscious perception and representation. This fact is implicit in every example Baars gives of contrasting conscious and unconscious phenomena. We have developed a neural attentional model (Newman and Baars, 1993) which instantiates a number of the distinctions Baars makes. Based upon the model, I would offer a list of criteria for what consciousness entails in addition to the unconscious processes which provide its necessary context.
2 Conscious Processing Requires Global Activation
2.1 For a percept or image to become conscious requires an activation process beyond that generated by the stimulus itself. The traditional metaphors for this criterion has been consciousness as a threshold or searchlight providing the necessary luminance for a percept or idea to enter into awareness. Crick (1984) suggested the existence of a thalamic searchlight mediating such processes. Baars (1983; 1988) and Newman and Baars (1993) have extended this metaphor of a thalamically mediated searchlight to include cortical and midbrain areas contributing to the global activation of wide areas of the cortex. This extended reticular-thalamic activating system (Baars, 1988) is postulated to provide the supplementary activation required to represent the activities of modular processors as integrated, conscious contents.
2.2 This activation system is regulated by a global attentional matrix which controls access to a cortical global workspace. Various coalitions of unconscious processors compete for access to this global workspace (Newman and Baars, 1993). A basic criterion of this model is that consciousness is a limited-access system. It primary adaptive advantage is that it allows the organism to focus the processing resources of the central nervous system upon the particular stimulus most relevant to it in the moment. Viewed in isolation, this limited processing capacity of consciousness (and its concomitant slowness) would seem maladaptive. However, when the existence of myriad unconscious processes -- working outside of, and competing for access to consciousness -- is taken into account, the entire system can be seen to support the highly adaptive allocation of processing resources. Creating a single "stream of consciousness" is nature's solution to the problem of prioritizing experience, so that what is most dangerous or attractive or advantageous or interesting at that moment gains our undivided attention.
2.3 Based upon the above criteria we would predict that global activation is transient, yet continuous. By this I mean that conscious percepts and images are the tip of a highly fluid iceberg of supporting and competing unconscious processes. Global activation, then, involves multiple levels of activation and inhibition preceding (e.g. sentence parsing) and following upon (e.g. long-term memory) the momentary content to which these processes relate.
2.4 Conscious representations and processes are multimodal and unitive. This may seem paradoxical, yet it is quite consistent with the data Baars presents (e.g. Necker Cube imaging). By unitive I mean that consciousness naturally tends towards the integration into a single, unified percept of the outputs of those unconscious processors predominant in the global workspace at any one time. Attended percepts are normally multimodal. Mental images tend to integrate several stimuli (e.g. chunking). Brain research has shown that even stimuli presented in a single modality, like vision, are built up out of the activities of multiple processing areas specialized for edge detection, contrast, color, movement, etc. The CNS, then, requires some mechanism for binding the diverse representations generated by multiple unconscious processors into a coherent stream of consciousness. It is our contention that global activation, as described briefly above, provides the necessary basis for such binding. Global activation is not some vague metaphor, rather it is a widely studied property of neural activity (grossly represented by the EEG). Moreover, the neural global workspace model (Newman and Baars, 1993) delineates the anatomical and physiological substrates for this activation process with considerable specificity.
3 Conclusion
3.1 As these criteria and examples illustrate, any approach that hopes to define the nature of consciousness must develop contrasting perspectives for elucidating conscious and unconscious processes. These perspectives should be understood as complimentary rather than mutually exclusive, however.
The Dead Hand: Commentary on Baars on contrastive analysis Summary Full Text
Bruce Mangan
Behaviorism still threatens consciousness research. On the surface, Baars' "contrastive analysis" may look as if it reduces first-person consciousness to a third-person construct. But once its tacit behaviorism is isolated and overcome, contrastive analysis turns out to give empirical support to the primacy of the first-person stance for the scientific investigation of consciousness.
1.1 As a pragmatic method for organizing experimental results, contrastive analysis is extremely useful. It underlies Baars' A Cognitive Theory of Consciousness (1988), in my opinion the most important single book on consciousness since William James' Principles of Psychology (1890). But to grasp the great strength of Baars' method as he uses it, it is best to read his book.
1.2 Paradoxically, Baars' "A Thoroughly Empirical Approach to Consciousness" (1994) rests on a theoretical elaboration of contrastive analysis. This moves Baars toward a philosophical (or at least methodological) tar-baby: the first-person/third-person dispute that now sticks to so much thinking about consciousness.
1.3 In his theoretical treatment of contrastive analysis, Baars at times seems to reduce the notion of first-person consciousness to a third-person construct. On my first reading, I was inclined to interpret this point as Max Velmans (1994) does. Velmans writes that Baars' contrastive analysis looks like it makes consciousness "just a construct within an information processing theory" (Velmans, 1994, section 1.3). But it is one thing to equate consciousness with certain operations in a third-person methodology; it is quite another to use the first-person sense of conscious experience as the basis of a subsidiary third-person experimental research program. The former interpretation would, among other things, make Baars a gross functionalist, and open up just the sort of philosophical can of worms that Baars wants (rightly or wrongly) to keep tightly closed. The later interpretation, which I think the correct one, means that the results of contrastive analysis increase support for the SCIENTIFIC validity of the first-person standpoint. This has the consequence of undercutting functionalism, and more generally helps to extricate research on consciousness from the tacit influence of behaviorism and its gratuitous assumptions: e.g., that as a science, psychology must be a third-person enterprise and must always rest on an experimental foundation.
1.4 The dead hand of behaviorism still deforms the study of consciousness. In this case, behaviorism is the unacknowledged source of the terminology and assumptions which make Baars look like a reductive functionalist; and tacit behaviorism also works in other ways to obscure the implications of Baars' method, even, perhaps, from Baars himself.
1.5 Baars is, above all, concerned with justifying the application of contrastive analysis to experimental psychology. Current experimental psychology is, in the main, the direct descendent of behaviorism, and in many ways still follows behaviorist conceptions. The difference between today's "cognitive" experimental psychology and radical behaviorism is extremely small relative to the difference between behaviorism and the research methods (e.g. James') that it supplanted. Today's cognitive experimentalism and less radical forms of behaviorism hardly differ at all. Over sixty years ago the behaviorist Tolman (1932) was willing to use internal, mental constructs for theoretical purposes. Experimental psychology now simply has more technical tools than Tolman did --- chief among them a better set of cognitive modeling techniques and the possibility for computer simulation.
1.6 I should note that from Baars' standpoint the transition from behaviorism to cognitivism was a substantial break, not a slight shift of emphasis. (See Baars' [1986] excellent book on the history of the cognitive revolution.) We differ here by virtue of our different frames of reference. If the Soviet Union under Stalin were one's starting point, Khruchev's reforms would look substantial and Gorbachev's truly "revolutionary." But the underlying Marxist-Leninist structure of the Soviet state in fact stayed intact during these changes. In the same way, Baars seems to take radical behaviorism to be the natural point of comparison, and so even slight movement toward the study of cognition and, ultimately, consciousness, will indeed seem "revolutionary." Nevertheless, the underlying behaviorist assumptions have changed very little, and experimental psychology is still, at best, quasi- behaviorist.
1.7 Behaviorism maintained that the methods it supplanted were bankrupt, but this is very questionable. Baars (1994) also seems to echo the behaviorist version of history when he writes of "the endless philosophical debate about consciousness [that] helped trigger the behaviorist revolution" (abstract). And Baars' reading of Nagel shows, I think, a feeling on Baars' part that philosophy in itself is a source of confusion, not help. Yet for the great consciousness researchers active just before the rise of behaviorism (e.g., William James, Ernst Mach, Ewald Hering) philosophy was part of a very catholic method of investigating consciousness and, notably, its relation to neural activity. In retrospect, their method turned out to have been remarkably successful; certainly more successful than behaviorism turned out to be. Men like James, Hering and Mach are part of a largely forgotten Golden Age of consciousness research. And, significantly, they each brought a version of first-person method to science, and achieved substantial results with it -- even if we only use third-person criterion to judge their first-person methodological efficacy. (For elaboration of, and evidence for, the above assertions, see Mangan, 1993, and 1.15 below.)
1.8 Watson insisted that on the "assumption that there is such a thing as consciousness and that we can analyze it by introspection, we find as many analyses as there are individual psychologists" (Watson, 1930, p. 5). This is the infamous "intractability" argument so often used against introspection. It is false both as a matter of logic and of evidence. Logically, the mere existence of what may seem to be an intractable question in a field of study hardly means that the question OR the field is to be abandoned. Arguments about the existence of atoms went on quite literally for millennia and well into the 19th century. Should the notion of an atom, then, have been abandoned altogether? Should the much more inclusive attempt to understand the physical world (of which atomic theory was one part) have been abandoned? Even if, for sake of argument, it could be shown that some questions in introspective psychology were intractable, this would hardly mean that introspection in general was suspect, any more than it would mean that the experimental method as such is suspect because it cannot now resolve important scientific questions in, for example, cosmology. Nor am I aware of any study showing that the degree of philosophical controversy about consciousness in the late 19th Century was notably greater than for similar scientific and philosophical controversies, say, a century earlier or a century later.
1.9 It was not controversy in philosophy that behaviorism found so objectionable, but the inclusion of philosophy at all. Behaviorism just asserted that a scientific psychology had to be narrowly defined and rest on a strict experimental basis. It must never be forgotten that the early behaviorists could not point to any major research accomplishments of their own to justify their method. The behaviorists simply promised loudly and often that they could do much better science by, among other things, breaking the extremely close traditional link between what we today call psychology and philosophy. This meant that the proven methods of the 19th Century philosopher/scientists were abandoned. Behaviorism in fact failed; very few interested in consciousness research will dispute this. The behaviorist exclusion of philosophy, especially in the study of consciousness, is then justified neither by research success nor cogent argument. Philosophy and psychology have good reason to coalesce into an integrated method of consciousness research, since this has already proven its scientific value. Yet here, again, lingering behaviorism obscures a promising research direction.
1.10 Watson's ability as a scientist is questionable, but on one point there is universal agreement: he was a propagandist of genius. And as with so many 20th Century revolutionaries, Watson yearned to found a movement and recruit true believers to a cause. In this context, Watson's rhetoric triumphed: The first page of his Behaviorism begins by distinguishing "introspective or subjective psychology, and behaviorism or objective psychology" (Watson, 1930, p. 1). "Subjective psychology [James being used as a prominent example] claimed that consciousness is the subject matter of psychology. Behaviorism, on the contrary, claims that consciousness is neither a definite nor a usable concept. The behaviorist, who has been trained always as an experimentalist, holds ... that belief in the existence of consciousness goes back to the ancient days of superstition and magic." (Watson, 1930, p. 2; his emphasis; see his Chapter 1 for many pages in a similar vein.)
1.11 The result of these founding assumptions is that experimental psychology, even today, has probably the single most reactionary attitude toward the study of consciousness of any discipline. As a trained experimental psychologist, Baars showed real courage in exploring consciousness when virtually all his colleagues considered it the domain of cranks. In effect, Baars is finally answering Watson's challenge within a field that is still largely conditioned by Watson's assumptions. Above all, Baars' contrastive analysis shows that consciousness IS, contrary to Watson, a "definite" and a "usable" concept.
1.12 This background helps explain why Baars (1994) at times writes about consciousness as if it were a "theoretical entity" (abstract), an "inferred construct" or "explanatory construct" (section 1.3), and otherwise handles consciousness as if it were a sort of discovery arrived at via theoretical reasoning, and subject to all the vicissitudes a purely theoretical entity is heir to. (Note that this stance does NOT presume that consciousness is a robust phenomenon of nature about which we theorize, construct scientific explanations, etc.) Here Baars simply follows the logical positivist and operationalist notions of scientific theory adopted by behaviorism (and continued by the so-called "functionalists" to this day; see Dennett, 1993). But it is also clear that Baars wants to fully acknowledge consciousness in the good, old-fashioned, intuitive, introspective and realist sense: he refers to "consciousness AS SUCH" (section 1.6), "consciousness per se" and "'real' consciousness" (section 2.1). And this is the meaning of consciousness Baars invokes in the first sentence of his introduction: "You, the reader, are now conscious of the words in your visual focus" (section 1.0). Here is consciousness as THE direct fact of human existence and the least dubious reality.
1.13 Now at times Baars does seem to handle these two senses of consciousness by reducing the realist, first-person notion (i.e., consciousness "as it is") to the operational, third-person notion (i.e., consciousness as an "inferred" theoretical entity). Consider, for example, what we might call Baars' (1994) thesis statement: "The question is whether conscious and unconscious processes can be usefully treated as inferred constructs expressed in [information processing] language" (abstract). But I think we can take this way of putting things to be an artifact of the (at least) quasi-behaviorist audience Baars is accustomed to addressing.
1.14 In any case, as a logical matter, showing experimental psychologists how to incorporate consciousness into their existing theoretical apparatus does not imply that this is the ONLY way the scientific investigation of consciousness must proceed. There is, then, no full reduction in the technical sense. Furthermore, a little consideration by those familiar with consciousness research should make it clear that the behaviorist presumptions about the nature of science are simply wrong: science is not co-extensive with experimental method OR the third-person stance. And it is not necessary to argue this point theoretically. We can show IN FACT that this is the case, i.e., show an unquestioned case of a successful scientific investigation of consciousness that demanded neither psychological experiment nor an exclusively third-person approach.
1.15 We need look no further than to Baars' epigraphs for a researcher who satisfies this demand. I trust few readers doubt that James achieved a powerful scientific understanding of consciousness. Baars (1988) himself has demonstrated this in many ways. But James' scientific approach was completely different from that of experimental psychology. As noted above, James used a "mixed" method (Mangan, 1993) drawing primarily on first-person introspection, buttressed by many other considerations from neurophysiology to traditional philosophy. James found the experimental method in psychology all but useless, calling it "microscopic psychology" even when it was still "asking every moment for introspective data" (1890/1918, p. 192).
1.16 We can avoid the appearance of reducing consciousness to a third-person construct by changing a single word in Baars' thesis statement. "Conscious and unconscious processes are corroborated by information processing constructs." Here "corroborated" replaces Baars' "inferred." Our first-person sense of consciousness is certainly not "inferred" by third-person means. Just the opposite, as I think Baars would agree.
1.17 Contrastive analysis is an inductive method, based on our intuitive understanding of what it is like to be conscious. It is only by virtue of this intuitive foundation that we are able to pick out areas of third-person experimental research that look like they point to first-person consciousness. Otherwise we would have no more reason to take, say, the attention literature to be relevant to consciousness than the experimental literature on digestion. Again, contrastive analysis must always turn on an antecedent first-person sense about which experimental findings capture conscious processes. Third-person experimental findings -- selected by first-person consciousness -- show common characteristics (e.g., chunking limits; one consistent interpretation at a time in focal attention), and a systematic structural contrast with otherwise similar processes that do not seem to be conscious (e.g., relatively unlimited processing capacity; simultaneous but contradictory interpretations at the unconscious level). The first-person stance is, then, hardly an "inferred" construct of third-person experiments. And as the result of an inductive process, third-person findings about consciousness are, as a logical matter, restricted accordingly.
1.18 Watson took the study of consciousness to be intractable. Baars' method directly refutes this and all other forms of the "intractability" canard. Besides showing that the scientific analysis of consciousness is not only possible but actual, contrastive analysis shows that the first-person stance is indispensable. It is not simply "complementary" (Velmans, 1994, section 1.8). The first-person stance is the foundation of the scientific investigation of consciousness.
A Welcome Dialogue on Empirical Issues:Reply to commentaries on Baars on contrastive analysis Summary Full Text
Bernard J. Baars
Davis, Allen, and Newman raise significant empirical questions. I agree with Davis that the operational definition of unconsciousness is criterion-dependent, and that the criterion can be set more conservatively than I did here. Contrastive analysis would still work if we compared clearly "conscious" to "much less conscious" phenomena. I agree with Velmans and Mangan that contrastive analysis involves the subject's first-person perspective --- that is why we study consciousness, after all --- but a rigorous physicalist could equally well trace the logic from behaviorally defined operations to the first-person perspective. There is no principled disconnect between these two perspectives on the evidence, and we know from almost two centuries of psychophysical research that there is rarely any mismatch in scientific practice. I am very encouraged by the ease of communication in these commentaries. Bringsjord's challenge seems to involve a difference of views on what are the most interesting questions.