The Neurophilosophy of Consciousness

 

Abstract: The neurophilosophy of consciousness brings neuroscience to bear on philosophical issues concerning phenomenal consciousness, especially issues concerning what makes mental states conscious, what it is that we are conscious of, and the nature of the phenomenal character of conscious states. Here attention is given largely to phenomenal consciousness as it arises in vision. The relevant neuroscience concerns not only neurophysiological and neuroanatomical data, but also computational models of neural networks. The neurophilosophical theories that bring such data to bear on the core philosophical issues of phenomenal conscious construe consciousness largely in terms of representations in neural networks associated with certain processes of attention and memory.

 

Traditional philosophical issues that phenomenal consciousness raises involve the relation of phenomenal consciousness to the rest of the world, especially as that world is conceived of by the natural sciences. Thus much philosophical discussion concerns whether the world as conceived of by physical theory can adequately accommodate phenomenal consciousness or if instead we are left with a dualism that cleaves reality into, for example, a non-physical phenomenal consciousness and a physical everything else. Even among philosophers that agree that phenomenal consciousness is consistent with physicalism, there is much disagreement, for there are several proposals for how best to spell out the consistency of a physicalistic world-view that makes room for phenomenal consciousness. One way of portraying this cluster of issues is in terms of which natural science is best suited to study phenomenal consciousness and how to conceive of the relation between that science and the sciences involving the most basic aspects of reality (the physical sciences). One major view is that psychology is the proper science for understanding phenomenal consciousness and further, that psychological investigation of phenomenal conscious should be regarded as autonomous from sciences such as the neurosciences. In opposition is the view that the proper science is neuroscience and whatever contributions come from psychology are only valid insofar as psychological theories are reducible to neuroscientific theories. Increasingly, proponents of the latter view identify themselves as practitioners of neurophilosophy.

            Neurophilosophy is a sub-genre of naturalized philosophy—philosophy that embraces Quine’s (1969) vision of philosophy as continuous with the natural sciences—wherein the natural science in primary focus is neuroscience. The term “Neurophilosophy” entered philosophical parlance with the publication of Patricia Churchland’s Neurophilosophy (1986). Patricia Churchland and husband Paul Churchland are paradigmatic examples of neurophilosophers. Their professional training is primarily philosophical, their appointments are in philosophy departments, and they publish in philosophy journals. Thus, neuroscience and philosophy do not have equal influence over neurophilosophy.  Instead the primary forces that drive its development as an academic pursuit emanate from conventions of philosophical institutions. Thus, neurophilosophical work on phenomenal consciousness proceeds largely by bringing neuroscientific theory and data to bear on philosophical questions concerning phenomenal consciousness.

Such questions are diverse. However, a useful way to focus the discussion—as well as to understand what has been of primary concern to neurophilosophical theories of phenomenal consciousness—will be to focus on just three questions: the question of state consciousness, the question of transitive consciousness, and the question of phenomenal character. (The terms “transitive consciousness” and “state consciousness” are due to David Rosenthal. For discussion, see Rosenthal 1993.) The question of state consciousness concerns in what consists the difference between mental states that are conscious and mental states that are unconscious. We have conscious mental states, such as my conscious perception of the words I type. Mental states vary with respect to whether they are conscious. Consider, for example, your memory of your mother’s name. You may have had that memory for years but until you read the previous sentence it is unlikely that it was a conscious memory for the entire time between its initial acquisition and its current retrieval. In what does the difference between conscious and unconscious mental states consist? The question of transitive consciousness concerns what it is that we are conscious of. When one has a conscious state, typically, if not always, one is conscious of something, as when I am conscious of a buzzing insect. Things may vary with respect to whether I am conscious of them, as when I am only intermittently conscious of the conversation at a nearby table in a restaurant. What does it mean to be conscious of something? The question of phenomenal character concerns the so-called qualia of conscious states. Conscious states have certain properties—their phenomenal character—properties in virtue of which there is “something it is like” to be in that state. When I have a conscious perception of a cup of coffee there is, presumably, something it is like for me to have that perception and, for all I know, what it is like for you to have a conscious perception of a cup of coffee is quite different. What makes a conscious state have “something it is like” to be in that state?

            Neurophilosophical theories of consciousness bring neuroscience to bear on answering these three questions of consciousness. The question arises, of course, of what motivates neurophilosophy of consciousness. The primary answer is that it has a certain appeal to those with an antecedent belief in physicalism in that it seems especially well suited to bridge the gap between the phenomenal and the physical. Attempting to bridge the gap by reducing the phenomenal all the way down to chemistry or microphysics may strike many as too far a distance to traverse. More plausible is to seek a higher level physical set of phenomena as offered in biology. Of the biological phenomena, the most plausible candidates are neural. The appeal of neurophilosophical approaches to phenomenal consciousness may become more evident upon examination of some sample theories.

            Before examining the neurophilosophical theories, it will be useful to look at a small sample of some of the relevant neuroscience. Vision is one of the most important and best understood senses. Accordingly, most of the fruitful progress in combining philosophy and neuroscience to get a grip on phenomenal consciousness has occurred in the domain of visual consciousness.

 

NEUROSCIENCE AND VISUAL CONSCIOUSNESS

 

            The processing of visual information in the brain can be understood as occurring in a processing hierarchy with the lowest levels in the retina and the highest levels in areas of the cerebral cortex. Processing begins after light is transduced by the rods and cones in the retina and electrochemical signals are passed to the retinal ganglia. From there, information flows through the optic nerve to the lateral geniculate nucleus (LGN) in the subcortex. From the LGN, information is passed to the first stage of cortical processing in the primary visual area of occipital cortex (area V1). From V1, the information is sent to other areas of occipital cortex and is then sent along a “ventral stream” from occipital to infero-temporal cortex as well as along a “dorsal stream” from occipital to posterior parietal cortex (Milner and Goodale1995). Beyond that information is sent to areas of frontal cortex (Olson et al 1999) as well as the hippocampus (Milner and Goodale1995). As will be discussed further later, information does not simply flow from lower levels to higher levels but there are many instances in which it flows from higher levels down to lower levels (Pascual-Leone and Walsh 2001). Further, information is processed in various ways in different regions of the different levels and can be briefly characterized in the following ways. Information at the lowest levels is represented by neural activations that serve as detectors of features localized to specific retinocentrically defined locations of the visual field. Thus, at the lowest levels, neural activations in LGN and V1 constitute egocentric representations of visual features as in, for instance, the detection of an oriented line by a cell with a relatively small retinocentric receptive field. At progressively higher level areas (such as visual areas V2 through V5), locally defined visual features are “grouped” or integrated as in when local information about shading is grouped to give rise to representations of depth. Progressively higher levels of information processing increasingly abstract away from the egocentric information of the lower level representations and give rise to progressively allocentric (“other centered” ) representations as in view-point invariant representations in inferior termporal cortex that underwrite the recognition of objects from multiple angles and other viewing conditions. Thus information represented at progressively higher levels of processing becomes progressively less egocentric and progressively more allocentric with the most allocentric representations being in frontal areas and hippocampus (Mandik 2005).

            The question arises of how best to apply the concepts of consciousness of interest to philosophers--state consciousness, transitive consciousness, and phenomenal character--in the context of a neuroscientific understanding of visual perception. We may make the most progress in this regard by focusing on breakdowns and anomalies of normal vision. We will briefly examine two such cases. The first is blindsight, a condition that results from a certain kind of brain damage. The second is motion induced blindness, a condition that occurs in normal subjects under certain unusual conditions conditions.

            Blindsight is a condition in which lesions to V1 cause subjects to report a loss of consciousness in spite of the retention of visual ability. For so-called blind regions of their visual fields, blindsight subjects are nonetheless better than chance in their responses (such as directed eye movements or forced-choice identifications) to stimulus properties such as luminance onset (Pöppel, E., Held, R., and Frost, D. 1973)  wavelength (Stoerig, P., Cowey, A. 1992) and motion (Weiskrantz, L. 1995). Lack of consciousness is indicated in such studies by, for example, having the subject indicate by pressing one of two keys “whether he had any experience whatever, no matter how slight or effervescent” (Weiskrantz, L. 1996).

            Blindsight subjects’ responses to stimuli in the blind portions of their visual fields give evidence that the stimuli are represented in portions of the brain. However, it is clear that these representational states are not conscious states. Thus, the kind of consciousness that seems most relevant in describing what blindsight patients lack is state consciousness. Further, blindsight patients arguably also lack transitive consciousness with respect to the stimuli in the blind regions of their visual field. One consideration in favor of this view arises when we take the subject’s own reports at face value. They claim not to be conscious of the stimuli in question. It would be difficult to affirm that blindsight subjects do have transitive consciousness of the relevant stimuli without affirming that all instances of representation are instances of transitive consciousness, and thus instances of unconscious consciousness.

            Regarding the question of qualia, of whether there is anything it is like for blindsight subjects to have stimuli presented to the blind regions of their visual fields, I take it that it is quite natural to reason as follows. Since they are not conscious of the stimuli and since the states that represent the stimuli are not conscious states, there must not be anything its like to have stimuli presented to those regions. Of course, the reader may doubt this claim if the reader is not a blindsight subject. It will be useful in this regard to consider a case that readers will be more likely to have first-person access to. For precisely this reason it is instructive to look at the phenomenon of motion induced blindness (Bonneh et al2001).

            Motion induced blindness may be elicited in normal subjects under conditions in which they look at a computer screen that has a triangular pattern of three bright yellow dots on a black background with a pattern of blue dots moving “behind” the yellow dots. As subjects fixate on center of the screen it appears to them that one or more of the yellow dots disappear (although in reality the yellow dots remain on the screen). The effect is quite salient and readers are encouraged to search the internet for “motion induced blindness” and experience the effect for themselves. There are several lines of evidence that even during the “disappearance” the yellow dots continue to be represented in visual areas of the brain. The effect can be influenced by transcranial magnetic stimulation to parietal cortex (a relatively late stage of visual processing in the brain). Additionally, the effect can be shown to involve non-local grouping of the stimulus elements. So, for example, if the yellow dots are replaced with a pair of partially overlapping circles, one yellow and one pink—sometimes an entire circle will disappear leaving the other behind even though some parts of the two different circles are very close in the visual field. As mentioned previously, the brain mechanisms thought to mediate such object groupings are relatively late in the visual processing hierarchy.

            We may turn now to the applications of the concepts of transitive consciousness, state consciousness, and qualia to motion induced blindness. First off, motion induced blindness looks to be a phenomenon involving transitive consciousness since in the one moment the subject is conscious of the yellow dot, in the next they are not conscious of the yellow dot, and along the way they are conscious of a yellow dot seeming to disappear. Second, we can see that motion induced blindness allows for applications of the concept of state consciousness, since studies of motion induced blindness provides evidence of  conscious states that represent the presence of yellow dots as well as unconscious states that represent the presence of yellow dots.

            Let us turn now to ask how the concept of phenomenal character applies in the context of motion induced blindness. The best grip we can get on this question is simply by asking what it is like is like see yellow dots disappear. When there is an unconscious state that represents the yellow dots or no transitive consciousness of yellow dot, there is, with respect to the yellow dot, nothing it is like to see it. Or, more accurately, what this instance of motion induced blindness is like is like not seeing a yellow dot. When the state representing the yellow dot is conscious, what it is like to be in that state is like seeing a yellow dot. One might suppose, then, as will be discussed later, that what it is like to be in the conscious state is determined, at least in part, by the representational content of that state. In this case, it is the content of the representation of a yellow dot.

 

NEUROPHILOSOPHICAL THEORIES OF CONSCIOUSNESS

 

            I will now turn to examine a sample of neurophilosophical theories of consciousness. In keeping with the paradigmatic status of the work of the Churchlands in neurophilosophy, my primary focus will be on Paul Churchland’s neurophilosophical work on consciousness. However, other philosophers have produced neurophilosophical accounts and I will mention their work as well.

            Paul Churchland articulates what he calls the "dynamical profile approach" to understanding consciousness (2002). According to the approach, a conscious state is any cognitive representation that is involved in (1) a moveable attention that can focus on different aspects of perceptual inputs, (2) the application of various conceptual interpretations of those inputs, (3) holding the results of attended and conceptual interpreted inputs in a short-term memory that (4) allows for the representation of temporal sequences.

            Note that these four conditions primarily answer the question of what makes a state a conscious one. Regarding the question of what we are conscious of, Churchland writes that "a conscious representation could have any content or subject matter at all" (p. 72) and he is especially critical of theories of consciousness that impose restrictions on the contents of conscious representations along the line of requiring them to be self-representational or meta-representational (pp. 72 – 74).

            Much of Churchland’s discussion of the dynamical profile account of consciousness concerns how all of the four conditions may be implemented in recurrent neural networks. A recurrent neural network may be best understood in terms of contrast with feed-forward neural networks, but we should first give a general characterization of neural networks.  Neural networks are collections of interconnected neurons. These networks have one or more input neurons and one or more output neurons. They may additionally have neurons that are neither input nor output neurons and are called "interneurons" or "hidden-layer" neurons. Neurons have, at any given time, one of several states of activation. In the case of input neurons, the state of activation is a function of a stimulus. In the case of interneurons and output neurons, their state of activation is a function of the states of activation of other neurons that connect to them. The amount of influence the activation of one neuron can exert on another neuron is determined by the "weight" of the connection between them. Learning in neural networks is typically thought to involve changes to the weights of the connections between neurons (though it may also involve the addition of new connections and the "pruning" of old ones). In feed-forward networks, the flow of information is strictly from input to output (via interneurons if any are present). In recurrent networks there are feedback (or "recurrent") connections as well as feed-forward connections.

            Let us turn now to Churchland's account of how the four elements of the dynamical profile of conscious states might be realized in recursive neural networks. It helps to begin with Churchland’s notion of the conceptual interpretation of sensory inputs and we do well to begin with what Churchland thinks a concept is. Consider a connectionist network with one or more hidden layers that is trained to categorize input types. Suppose that its inputs are a retinal array to which we present grayscale images of human faces. Suppose that its outputs are two units, one indicating that the face is a male and the other indicating that the face is female. After training the configuration of weights will be such that diverse patterns of activation in the input layer provoke the correct response of “male” to the diversity of male faces and “female” for female faces. For each unit in the hidden layer, we can represent its state of activation along one of several dimensions that define activation space. A pattern of hidden layer activation will be represented as a single point in this space. This space will have two regions: one for males and one for females. Regions in the center of each of the two spaces will constitute “attractors” that define what, for the network, constitutes prototypical female faces and prototypical male faces, respectively.

            The addition of recurrent connections allows for information from higher layers to influence the responses of lower layers. As Churchland puts the point:

This information can and does serve to 'prime' or 'prejudice' that neuronal population's collective activity in the direction of one or other of its learned perceptual categories. The network's cognitive 'attention' is now preferentially focused on one of its learned categories at the expense of the others. (p.75 )

Churchland is not explicit about what this might mean in terms of the example of a face categorization network, but I suppose what this might mean is that if the previous face was a prototypical female, then the network might be more likely to classify an ambiguous stimulus as female. We can construe this as exogenous cueing of attention. Churchland goes on to further describe shifts of attention in recurrent networks that we might regard as endogenous. "Such a network has an ongoing control of its topical selections from, and its conceptual interpretations of, its unfolding perceptual inputs." (p.76).

            Recurrent connections allow for both a kind of short term memory and the representation of events spread out over time. In a feedforward network, a single stimulus event gives rise to a single hidden layer response then a single output response. With recurrence however, even after the stimulus event has faded, activity in lower layers can be sustained by information coming back down from higher layers and that activity can itself reactivate higher layers. Also, what response a given stimulus yields depends in part on what previous stimuli were. Thus recurrent connections implement a memory. Decreasing connection weights shortens the time it takes for this memory to decay. The ability to hold on to information over time allows for the representation of events spread out over time, according to Churchland, and the representation in question will not be a single point in activation space but a trajectory through it.

            Churchland (2002) does not go into much neuroanatomical or neurophysiolgical detail, but adverts, though tentatively, to the account in Churchland (1995) wherein he endorses Llinas' view whereby consciousness involves recurrent connections between the thalamus (a bilateral structure at the rostral tip of the brainstem) and cortex.

            The neurophilosophical account of consciousness by Prinz (2000, 2004) is relatively similar and fills in a lot of neuroanatomy and neurophysiology that Churchland leaves out. Prinz characterizes the processing hierarchy we discussed earlier and then notes that the contents of consciousness seem to match it with representations at the intermediate level of processing (areas v2-v5), meaning that the contents of conscious states do not abstract entirely from points of view as do the highest level of the processing hierarchy but neither are they the same as the representations at the lowest level. However, Prinz argues that intermediate representations are alone insufficient for consciousness. They must additionally be targeted by attention. Prinz thinks attention is required because of considerations having to due with the pathology of attention known as “neglect.”  Prinz cites Bisiach’s (1992) study of neglect patients who were able to demonstrate certain kinds of unconscious recognition. Prinz infers from such results that not only did  high level stuff areas in the visual hierarchy get activated (they are necessary for the kinds of recognition in question) but also that intermediate levels had to have been activated. (Prinz seems to be assuming that information can only get to higher levels of cortical processing by way of intermediate level but one wonders if perhaps the intermediate level of was bypassed via a sub-cortical route.)

            Given the large role that Prinz assigns to attention in his theory of consciousness, the question naturally arises as to what Prinz thinks attention is and what it does. Prinz endorses the account of attention by Olshausen, Anderson, and van Essen (1994) whereby attention involves the modulation of the flow of information between different parts of the brain. Further, Prinz endorses the speculation that the attention crucial in making intermediate level representations conscious involves a mechanism whereby information flows from intermediate areas, through high level visual areas (infero temporal cortex) to working memory areas in lateral prefrontal cortex. Pieces of information in working memory “allow the brain to recreate an intermediate-level representation by sending information back from working memory areas into the intermediate areas.” (2004, p. 210). Prinz (2000) summarizes, emphasizing attention’s role, as follows:

When we see a visual stimulus, it is propagated unconsciously through the levels of our visual system. When signals arrive at the high level, interpretation is attempted. If the high level arrives at an interpretation, it sends an efferent signal back into the intermediate level with the aid of attention. Aspects of the intermediate-level representation that are most relevant to interpretation are neurally marked in some way, while others are either unmarked or suppressed. When no interpretation is achieved (as with fragmented images or cases of agnosia), attentional mechanisms might be deployed somewhat differently. They might ‘‘search’’ or ‘‘scan’’ the intermediate level, attempting to find groupings that will lead to an interpretation. Both the interpretation-driven enhancement process and the interpretation-seeking search process might bring the attended portions of the intermediate level into awareness. This proposal can be summarized by saying that visual awareness derives from Attended Intermediate-level Representations (AIRs). (p. 249)

            Prinz's account of attention's role in consciousness seems a lot like Churchland’s conceptual interpretation, short term memory, and of course, attention requirements on consciousness. Tye raises objections to the sort of view advocated by Churchland and Prinz. Tye is critical of accounts of consciousness that build in constitutive roles for attention. Tye’s claim is based on introspective grounds (1995, p. 6). The thought here is that one might have a pain for a length of time but not be attending it the entire time. Tye insists that there is still something it is like to have an unattended pain.  Tye infers from these sorts of considerations that the neural correlate of visual consciousness is lower in the processing hierarchy than an attention-based theory would locate it. Tye thus locates the neural correlates of conscious states in “the grouped array” located in the occipital lobe and, regarding the phenomenon of blindsight, rejects “the hypothesis that blindsight is due to an impairment in the linkage between the spatial-attention system and the grouped array” (Tye 1995 p. 215-216) Tye accounts for the  retained visual abilities of blind sight subjects (p. 217) in terms of a subcortical pathway “tecto-pulvinar pathway” from retina to superior coliculus that continues through the pulvinar to various parts of the cortex, including both the parietal love and area v4. Thus Tye seems to think consciousness is in V1. Prinz 2000 argues against this, citing evidence against locating consciousness in V1 (see Crick & Koch, 1995, and Koch & Braun, 1996, for reviews). Prinz writes:

As Crick and Koch emphasize, V1 also seems to lack information that is available to consciousness. First, our experience of colors can remain constant across dramatic changes in wavelengths (Land, 1964). Zeki (1983) has shown that such color constancy is not registered V1. Second, V1 does not seem responsive to illusory contours across gaps in a visual array (von der Heydt, Peterhans, & Baumgartner, 1984). If V1 were the locale of consciousness, we would not experience the lines in a Kanizsa triangle.” (pp. 245-246).

           

Turning from disagreements to agreements, we may note that Churchland, Prinz, and Tye agree that conscious states are representational states. They also agree that what will differentiate a conscious representation from an unconscius representation will involve relations that the representation bears to representations higher in the processing hierarchy. For both Churchland and Prinz, this will involve actual interactions, and further these interactions will constitute relations that involve representations in processes of attention, conceptual interpretation and short term memory. Tye disagrees on the necessity of actually interacting with concepts or attention. His account is dispositional meaning that the representations need only be poised for uptake by higher levels of the hierarchy.

            Turning to the question of  transitive consciousness, we see both agreements and disagreements between the three authors.  For Churchland , Tye, and Prinz, they all agree that what one is conscious of is the representational content of conscious states. In all cases what the subject is conscious of is what the representational contents of the conscious states are. However, these theorists differ somewhat in what they think the contents can be. Churchland has the least restrictive view: any content can be the content of a conscious state. Prinz’s is more restrictive: the contents are not going to include high level invariant contents.  Tye’s is the most restrictive: the contents will only be first order and non-conceptual. Tye thinks that they are non-conceptual since he thinks that creatures without concepts—perhaps non-human animals and human infants—can have states for which there is something it is like to have them even though they possess no concepts. Tye says little about what concepts are and for this, among other reasons, it is difficult to evaluate his view. The reason Tye thinks the contents of consciousness are first-order is because he believes in the pre-theoretic obviousness of the transparency thesis whereby when one has a conscious experience, all that one is conscious of is what the experience is an experience of. Thus if one has a conscious experience of a blue square one is only aware of what the mental state represents: the blue square. One is not, Tye insists, able to be conscious of the state itself. So, for example, if the state itself is a pattern of activity in one’s nervous system, one will not be able to be conscious of this pattern of activity, but only be able to be conscious of external world properties that the pattern represents. Mandik (2005, 2006) argues that Churchland’s (1979) thesis of the direct introspection of brain states  provides the resources to argue against the kinds of restrictions on content that Tye makes.

I will not spell out the full argument here, just indicate the gist of it. Conceptual content can influence what it is like to have a particular experience. What it is to look at a ladybug and conceive of it as an example of Hippodamia convergens is, intuitively, quite different from what it would be like to conceive of it as one’s reincarnated great-great-grandmother. Thus if a person had the conceptual knowledge that consciously perceiving motion involved activity in area MT, and acquired the skill of being able to automatically and without conscious inference apply that conceptual knowledge to experience, then that person would be able to be conscious of the vehicular properties of that experience.

            I turn now to what neurophilosophical accounts have to say about phenomenal character. I focus, in particular, on the suggestion that phenomenal character is to be identified with the representational content of conscious states and will discuss this in terms of Churchland’s suggestion of how qualia should be understood in terms of neural state spaces.

Our experience of color provides the most often discussed example of phenomenal character by philosophers and Churchland is no exception. When Churchland discusses color qualia, he articulates a reductive account of them in terms of Land’s theory that human perceptual discrimination of reflectance is due to the sensory reception of three kinds of electromagnetic wavelengths by three different kinds of cones in the retina. In keeping with the kinds of state-space interpretations of neural activity that Chruchland is fond of, he explicates color qualia in terms of points in three dimensional spaces, the three dimensions of which correspond to the three kinds of cells responsive to electro-magnetic wavelengths. Each color sensation is identical to a neural representation of a color (a neural representation of a spectral reflectance). Each sensation can thus be construed as a point in this three dimensional activation space and the perceived similarity between colors and the subjective similarities between corresponding color qualia are definable in terms of proximity between points within the three-dimensional activation space. “Evidently, we can reconceive [sic] the cube [depicting the three dimensions of coding frequencies for reflectance in color state space] as an internal ‘qualia cube’”(1989, p. 105).  Churchland thinks this approach generalizes to other sensory qualia, such as gustatory, olfactory, and auditory qualia (ibid, p. 105-106). Bringing this view in line with the thesis of the direct introspection of brain states, Churchland writes:

The “ineffable” pink of one’s current visual sensation may be richly and precisely expressible as a 95Hz/80Hz/80Hz “chord” in the relevant triune cortical system. The “unconveyable” taste sensation produced by the fabled Australian health tonic Vegamite [sic.] might be quite poignantly conveyed as a 85/80/90/15 “chord” in one’s four-channeled gustatory system (a dark corner of taste-space that is best avoided). And the “indescribable” olfactory sensation produced by a newly opened rose might be quite accurately described as a 95/35/10/80/60/55 “chord” in some six dimensional system within one’s olfactory bulb.

            This more penetrating conceptual framework might even displace the commonsense framework as the vehicle of intersubjective description and spontaneous introspection. Just as a musician can learn to recognize the constitution of heard musical chords, after internalizing the general theory of their internal structure, so may we learn to recognize, introspectively, the n-dimensional constitution of our subjective sensory qualia, after having internalized the general theory of their internal structure (ibid, p. 106).

Three particular and related features of Churchland’s view of qualia are of special note. The first is that qualia are construed in representational terms. The second follows from the first, namely, that qualia so construed are not intrinsic properties of sensations and thus overturns a relatively traditional view of qualia. The third is that it allows for intersubjective apprehensions of qualia. To see these points more clearly it will be useful to briefly examine the traditional account of qualia noting the role of supposedly intrinsic properties in the account.

            It is difficult to say uncontroversial things about qualia; however, there are several points of agreement among many of those philosophers that believe that mental states have such properties. These philosophers describe qualia as (i) intrinsic properties of conscious states that (ii) are directly and fully knowable only by that subject and (iii) account for “what it is like” for a subject to be in that state. More briefly, qualia are (i) intrinsic, (ii) subjective, and (iii) there is “something it is like” to have (states with) them. Less briefly, we can start with (iii) and work our way to (i) as follows. When I have a conscious perception of a cup of coffee there is, presumably, something it is like for me to have that perception and, for all I know, what it is like for you to have a conscious perception of a cup of coffee is quite different. Further, for all that you can tell me about your experience, there is much that cannot be conveyed and thus is subjective, that is, directly and fully knowable only by you alone. The supposition that qualia are intrinsic properties of conscious states serves as a possible, though questionable, explanation of their subjectivity. (See Mandik 2001 for a neurophilosphical account in which subjectivity is consistent with qualia being extrinsic.) The inference from subjectivity to the intrinsic nature of qualia may be articulated as follows. If something is defined by the relations that it enters into then it is fully describable by the relations it enters into and if it is not fully describable by the relations it enters into it must not be defined by the relations it enters into.

To construe qualia in terms of representational content, however, is to construe them as no longer intrinsic, since typical accounts will spell out representational content in terms of either (1) causal relations sensory states bear to states of the external world, or (2) causal relations that they bear to other inner states, or (3) some combination of the two sorts of relations. In neural terms, a pattern of activation in a neural network is the bearer of representational content in virtue of either (1) the distal or proximal stimuli that elicit the activation, or (2) other patterns of activation that influence it via, e.g., recurrent connections or (3) some combination of the two.

While it is relatively clear how Churchland’s view is supposed to rule out the view of qualia as being intrinsic, it is not so clear that it is equally able to rule out their being subjective. The above quoted passage contains Churchland’s view that properties of neural states previously inexpressible could, if one acquired the relevant neuroscientific concepts and the skill to apply them introspectively, become expressible. However, this view seems to be in tension with the earlier mentioned view that concepts influence phenomenal character. The phenomenal character of an experience prior to the acquisition and introspective application of a concept will not, then, be the same as the phenomenal character of an experience after the acquisition and introspective application of that concept. Thus, even within a general neurophilosophical view of consciousness, there may remain certain representational contents of neural states that are directly and fully knowable only by the subject who has them. Neurophilosophy, then, may be fully compatible with the subjectivity of phenomenal consciousness.

References

Bisiach, E. 1992: Understanding consciousness: Clues from unilateral neglect and related disorders. In A. D. Milner and M. D. Rugg (eds.) The Neuropsychology of Consciousness, 113-139. London: Academic Press.

Bonneh, Y., A. Cooperman, and D. Sagi. 2001: Motion induced blindness in normal observers. Nature 411(6839), 798-801.

Churchland, P. M. 1979: Scientific Realism and the Plasticity of Mind, Cambridge: Cambridge University Press.

Churchland, P. S. 1986: Neurophilosophy, Cambridge, MA: MIT Press.

Churchland, P. M. 1989: A Neurocomputational Perspective, Cambridge, MA: MIT Press.

Churchland, P. M. 2002: Catching consciousness in a recurrent net. In A.Brook, and D. Ross (eds.),  Daniel Dennett: Contemporary Philosophy in Focus, 64-80. Cambridge: Cambridge University Press.

Crick, F., and C. Koch. 1995: Are we aware of activity in primary visual cortex? Nature 375, 121–123.

Koch, C., and J. Braun. 1996: Towards a neuronal correlate of visual awareness. Current Opinion in Neurobiology 6, 158–164.

Land, E. H. 1964: The retinex. Scientific American 52, 247–264.

Mandik, P. 2001: Mental representation and the subjectivity of consciousness. Philosophical Psychology 14 (2), 179-202.

Mandik, P. 2005:  Phenomenal consciousness and the allocentric-egocentric interface in R. Buccheri et al. (eds.); Endophysics, Time, Quantum and the Subjective. World Scientific Publishing Co.

Mandik, P. 2006: The introspectability of brain states as such. In B. Keeley, (ed.) Paul M. Churchland: Contemporary Philosophy in Focus. Cambridge: Cambridge University Press.

Milner, A. and M. Goodale 1995: The Visual Brain in Action. New York: Oxford University Press.

Olshausen, B. A., Anderson, C.H., & van Essen, D.C. (1994). A neurobiological model of visual attention and invariant pattern recognition based task. Journal of Neuroscience, 14, 6171-6186.

Olson, C., Gettner, S., and Tremblay, L. (1999) Representation of allocentric space in the monkey frontal lobe, in N. Burgess, K. Jeffery, and J. O'Keefe (eds.), The Hippocampal and Parietal Foundations of Spatial Cognition, Oxford University Press, New York, pp. 359-380.

Pascual-Leone, A., and Walsh, V. (2001) Fast backprojections from the motion to the primary visual area necessary for visual awareness, Science 292, 510-512.

Pöppel, E., Held, R., and Frost, D. (1973) Residual visual functions after brain wounds involving the central visual pathways in man, Nature 243, 295-296.

Prinz, J. 2000 A Neurofunctional Theory of Visual Consciousness Consciousness and Cognition 9, 243–259.

Prinz, J. 2004 Gut Reactions New York: Oxford University Press.

Quine, W. (1969). “Epistemology Naturalized,” in Ontological Relativity and Other Essays (New York: Columbia University Press).

Rosenthal, D. 1993. "State Consciousness and Transitive Consciousness", Consciousness and Cognition, 2, 4 (December 1993), pp. 355-363

Stoerig, P., Cowey, A. (1992) Wavelength discrimination in blindsight, Brain 115, 425-444.

Tye, M. (1995). Ten Problems of Consciousness: A Representational Theory of the Phenomenal Mind.  Cambridge, MA: MIT Press.

von der Heydt, R., Peterhans, E., & Baumgartner, G. (1984). Illusory contours and cortical neuron responses. Science, 224, 1260–1262.

Weiskrantz, L. (1995) Blindsight: not an island unto itself, Curr. directions Psychol. Sci. 4, 146-151.

Weiskrantz, L. (1996) Blindsight revisited, Curr. Opin. Neurobiol, 6(2), 215-220.

Zeki, S. (1983). Colour coding in the cerebral cortex: The reaction of cells in monkey visual cortex to wavelengths and colour. Neuroscience, 9, 741–756.

 

PETE MANDIK

Biographical sketch:

Pete Mandik is Cognitive Science Laboratory Coordinator, Associate Professor, and Chairman of the Philosophy Department of William Paterson University, New Jersey, USA. He is a junior member and co-director of the McDonnell Project in Philosophy and the Neurosciences, Simon Fraser University, Canada. He is co-editor of Philosophy and the Neurosciences: A Reader (Basil Blackwell, 2001).