Popping the Thought Balloon
DRAFT of (contact author for most recent version)
Many recovering dualists find that the old Cartesian demons are hard to exorcise. Dual substance abuse manifests itself not only as metaphysical dualism, but as a pervasive epistemological framework that creates an unhealthy codependent relationship between scientific realism and phenomenology. Daniel Dennett has led philosophers to recognize many of the symptoms of creeping crypto Cartesianism. In this paper, I try to take Dennett to the limit: Descartes lives on, I argue, in the very heart of cognitive science, in the concept of representation. I outline a five-step program for overcoming this lingering, fundamental, allegiance to Cartesianism, and discuss Dennett's own progress along this path.
Acknowledgments: Many thanks to Dan Dennett, Don Ross, Miller Brown, and Richard Lee for their insight and suggestions, and to Kate Weingartner for her help with the bibliography.
I. Ain't Misbelieving
Non sunt multiplicanda entia praeter necessitam. The patron saint of Anglo-American philosophy might well be William of Ockham, who with his legendary razor undercut the ever-present impulse to reify dispositions, processes, capacities, and correlations, transforming them into discrete enduring things, black boxes to be presupposed, discussed, and finally taken for granted. But for a really close shave, one should choose Dennett, who outockhams Ockham by not just holding the line on new entities, but by resolutely cutting the extant ontological inventory. Over the years many a familiar entity has suffered redennettia ad absurdum, of the following rough form:
1. Many people believe in the existence of x.
2. X has (or should have) properties P..., If it is indeed the x that many people believe in.
3. Nothing with properties P... does/could exist.
4. Therefore, x does not exist, or better,
5. Therefore, x exists but has few or none of the properties commonly believed of it. (Instead, it has the properties Q....)
For his reductio Dennett draws from a repertoire of rhetorical and argumentative strategies:
3A. Intuition pumps: You really can imagine the world without x
if you only try a little harder. (And here's a story to help....)
3B. Analogies: x is very like the utterly disreputable y, which scientists/philosophers/everyone have long cast off or would cast off instantly if its existence were seriously proposed.
3C. Skeptical disjunctions: The properties ascribed to x are initially vague. The attempt to resolve which of P... really do apply either leads to a contradiction among P... or a genuine open question which is irresolvable in principle.
3D. Sorites and slippery slopes. The borders of xness can be established only through a foolish consistency of microcriteria.
3E. Empirical challenges: x, if it exists, would cause certain observed effects. These have been disconfirmed in the lab.
This is a powerful toolbox, and the strategies above lend themselves well to organic combinations. Their deployment places Dennett solidly in the esteemed tradition of constructive skepticism, from Hume to Ryle to Quine. Armed with Dennett's various razors, it is surprising, then, that Dennett frequently ends up with conclusion 5 above, rather than the more restrictive 4. I suspect this is the product of the faculty of Yankee Cognition, or the Mainer Cortex, or maybe just plain Niceness: Ockhamite New England ontological conservatism and a powerful (equally New England) impulse to preserve appearances. The opponent should be edified about his or her confusions, not skewered on them.
Preserving appearances is, however, exactly the right initial impulse or "stance" (if I may use the term) in the study of consciousness. It is precisely appearances that need to be explained, and so discussion must begin with what seems to be the case. For the really conservative, stubbornly retrograde philosophers -- a surly bunch who shall go nameless -- discussion must also end with initial appearances. For them, the ways things seem are immutable, and anyone who attempts to revise initial appearances is guilty of absurdity ab initio.
Dennett, in contrast, holds out the possibility that explanation and rational reconstruction of some appearances can render them as mere apparent appearances, appearing otherwise after the explanation is grasped by the appearee. On the surface, the idea that we can be wrong about what seems to be the case to us seems paradoxical, or it entails an infinite regress of seemings to seem. What seems to me may not be what's real, but that it seems to me is a real mental event which is constituted by the seeming itself -- "correction" of this experience by "reality" is impossible. The air of paradox, however, is merely what Dennett calls (in conversation) a "deepity," a claim that appears profound but is in fact a superficial equivocation. (His example of a deepity is "Love is just a four-letter word." It kinda makes you think.) The deepity of unseeming seemings dissolves when one considers the unstated temporal dimension of all judgments, including judgments of appearance and reality. There's real fruit and wax fruit. A wax apple may seem real at time t1, and then will seem artificial at time t2, about the time you attempt to sink your teeth into it. The discovery of its waxiness alters the way it seems, an alteration of many of its sensory properties. But it will always be true that it seemed real to you at t1. That time-indexed seeming is indeed immutable, like any past event, but it would be a confusion of the deepity type to claim that the way an apple seems is immutable in general. Every object of perception is subject to change in its appearance, and this is no less true when the "object" in question is a mental episode (like a seeming). The sadness we feel at the end of a two hankie movie may seem one way today (stirring, noble, magnanimous) and another way tomorrow (sentimental, puerile, foolish).
The deepity about appearances, about states of consciousness as they seem to us and as seemings in themselves, and the ever-shifting current, retrospective, and prospective ways these seemings seem, this deepity would be a relatively minor point were it not subtly enshrined at the foundation of modern philosophy, in Descartes. Descartes raises the famous question, am I dreaming right now? (It's a Dennett sort of question: see strategy 3C above.) Due to a cognitive accident about dreams, namely, our chronic loss of insight into our own state while dreaming, this question is hard to answer. Dreams seem real as they are dreamed, as does waking reality while it is experienced. Descartes raises the bar a bit by posing the example of the mundane dream ("sitting in front of the fire," etc.). So, at any time, I cannot tell (in principle) whether what seems to be the case really is the case.
Sophomores in Philosophy 101, for whom the natural ontological attitude comes naturally, can be relied on to object strenuously: "Of course I can tell the difference between dreaming and waking," they proclaim. Their account is usually reminiscent of distinctions between dreaming and waking given by Descartes himself in the Sixth Meditation:
Case closed. Between the First and the Sixth Meditations Descartes has reassured himself about the reliability of his reason, and thus can appeal to the revisable retrospective judgments of reality and their continuous provisional updating, the regular pragmatics of cognitive life in sophomores and everyone else. But in spite of our everyday confidence, philosophy has overlooked Descartes' pragmatic Sixth while remaining permanently dazed by his First: "[T]here are no certain indications by which we may clearly distinguish wakefulness from sleep.... And my astonishment is such that it is almost capable of persuading me that I now dream." (p. 146) The dream argument, as has often been noted, seems to work too well, succeeding out of proportion with its place as an opening gambit in the Method of Doubt. But why is that? Descartes' persistent deepity lies in the urgency with which he presents the cognitive risk entailed by his skeptical arguments. The skeptical specter of occurrent error is identified with the risk of real error, the kind which emerges in subsequent correction, with its real effects on the progress of our lives. The Evil Demon amplifies the same deepity. In the demon's thrall, we are deluded about everything, and all the time -- a terrible fate, but one whose terrors are only hypothetical, since systematic demonic delusion is without practical consequences. In practice, an uncontradicted delusion is as good as a truth.
The fact that we philosophers share with Descartes the view that his skepticism reveals a genuine epistemic problem indicates that we share a pervasive assumption (an assumption not shared by our wise sophomores.) To put this in the most neutral language I can, let us call this the "doctrine of registration." The doctrine holds that minds model the world, and that from moment to moment the mental model is more-or-less in registration with the world. In veridical waking perception, the registration is "tight," while in psychosis or dreaming the model ranges far afield. The doctrine of registration assumes several components and specific relations among them: a model, a medium from which the model is formed, a separate world to be modeled, and a mapping relationship between them, which somehow makes sense of both veridical and false modeling. That relationship is our old friend "representation" or "intentionality." It is usually the baby that is preserved as the bathwater of Cartesian dualism is discarded. Because Descartes fills the mind with mental representations that point with intangible but determinate arrows to intentional objects, there is a possibility of widespread occurrent error, error undermining whole continents of cognition. Worse, the error could be covert. We could be unwitting dupes and stooges. In short, with a good dose of Cartesian skepticism, we could experience every intellectual and emotional consequence of error, without ever actually experiencing the pragmatic collision with a contrary world. This is error lite; Descartes' freighted weighting of error lite is a splendid deepity and the concomitant creation of the modern mind is an astonishing result.
Had Descartes sprung his deepity on the world in a different time, it might have been less effective. (Indeed, both Augustine and Montaigne noted the dream problem, but without its modern foundational implications.) He moved at the exact moment when, as Donne put it, "New philosophy calls all in doubt." New cosmology and new physics gave European intelligentsia a galloping case of what we would now call mkodern science. In a sense, Descartes' worst fantasies were unfolding: What everyone thought to be the case was turning out to be systematically and pervasively false. The forces and laws that governed the physical world were discovered to be quite different from the folk-inspired and Aristotelian theories that had seemed so obvious. And so it goes on to this day: The world-model in one's mind is at best a radical translation of the distal energies of reality.
To borrow a metaphor from Dennett, what Descartes bequeathed the world was the thought balloon, a free-floating package of ideas that is always distinct from the world of matter ( Dennett, 1996 ). The thought balloon of cartoonists, as Dennett notes, typically captures most of the Cartesian legacy. It depicts representation, the mental model of the world. It incidentally depicts unity and sequence, the single stream of consciousness. It implies interiority, popping as it does from the head. It often implies a linguistic basis. It assumes an inner speaker or illustrator, and an inner witness. And its cloudiness suggests its emergence from a nonphysical dimension.
Substance dualism, however, was merely the ontological icing on this epistemic cake. The problems of substance dualism have long been conspicuous and as a result it has been the first Cartesian construct to be deleted from the mix. But that's the easy part. As Dennett has steadily illuminated, what remains after the rejection of dualism is "Cartesian materialism" (including some exotic variants: David Chalmer's quinteseentialism; Roger Penrose's quantum dualism; the mysterian philosophies of Thomas Nagel or Colin McGinn). The Cartesian materialist thinks of the brain as a res cogitans. But the res is now a squishy gray gland, not obviously cogitans at all. It is spatially squishy, with elaborate and obscure goings-on all over the place, and it is temporally squishy, with the goings-on going on asynchronously. Where, then, is the thought balloon to be tethered? The Cartesian materialist retreats in space to the Pineal frontier, and the surmise that amid the squishy stuff there is a special clearinghouse where the full-blown cogitations gather, and retreats in time to the idea that content crosses a threshold into consciousness. (For example, Baars, 1996 )
Dennett summarizes the assumptions of Cartesian materialism in the metaphor of the Cartesian Theater. After many applications of Dennett's wrecking ball, what remains is the Multiple Drafts Model. It is telling, I think, to examine both what the MDM denies as well as what it tacitly accepts. In Consciousness Explained (CE) (1991), "the novel feature" of the model is this:
MDM explicitly denies spatiotemporal pinealism. However, there remains a Cartesian residue in the idea of an "observation" made by "a specialized, localized part of the brain." In an earlier statement, Dennett and Kinsbourne (1992: 234) are more explicit about this local "observation":
Several early respondents to the Multiple Drafts model, myself included, worried that these distributed, yet discrete, microtakings had the effect of replacing the Cartesian theater with the Cartesian cineplex, a profusion of minicinemas featuring arty films like "I am curious -- Yellow" in cine V4, or "North by Northwest" in cine V5 (See the commentaries following Dennett and Kinsbourne 1992). Dennett and Kinsbourne immediately rejected the metaphorical implication that little homunculi took in these shows, but the minicinema objection also highlighted a different problem: If each microtaking is a discrete, determinate neural event, and consciousness was an organized symphony of microtakings, then it might have quite determinate spatial and temporal boundaries after all. Even though there would be no one stream of consciousness, no cardinal microtakings, there would still be a time and place when a content would dribble into the watershed of consciousness. To further complicate things, just about everyone (except me) wanted to insist that many microtakings were unconscious (or preconscious), saddling Dennett and Kinsbourne with the chore of distinguishing when the microtakings emerged from prehistory into consciousness proper. Very theatrical, very Cartesian (Lloyd 1992).
Dennett and Kinsbourne responded to this worry in the midst of a summary of their overall position:
The last sentence deploys one of Dennett's sharper razors (3D in the toolbox list). Sharper than intended: If indeed we want to eschew arbitrary boundaries, the cheerful posit of internal "microtakings" is in exactly the same Dennettian hot water. Let's turn up the magnification just a bit. To microtake or simply to judge is, we are told, the job of a specialized part of the brain. For example, in area V4, say, we get the microjudgment we could roughly label "Yellow ho!" Now, does that judgment form just when the V4 yellow neurons fire? How many neurons? Must the spike be fully formed and moving down the axon? But why not consider the microtaking to begin with the presynaptic surge of neurotransmitters that is sufficient to generate the spike train? And why not extend it to all the downstream/inboard effects of yellow-detection, right on out to the motor response? To insist that any of these questions has a determinate answer is to seek a place "where it all comes together" for the microtaking. It's just like -- let's see -- the British Empire learning of the end of the War of 1812 (CE 146f.). As a distributed entity, the Empire's recognition of world events cannot be precisely dated. But everything material is extended, and no mechanical process is instantaneous. This is true "all the way down." Is an ion channel officially open when its proteins flex, or when the first ion enters, or when it leaves? As for the British Empire, so for this microsystem. Just as there is no nonarbitrary, privileged seat of consciousness overall, when we look more closely, there is no nonarbitrary, privileged seat of the microtakings from which consciousness is built.
But these microtakings are representations, content-bearers, the foundation of cognition; their existence is the central assumption of cognitive science. If they lack identity conditions, then there's trouble ahead. I will hazard two bold generalizations. The first is philosophical: The panoply of anti-realist intuition pumps, examples, and arguments deployed against the Cartesian theater and against a host of intentional entities (especially belief) will also afflict the concept of representation overall, whether applied at the micro or macro scale. The Intentional Stance might as well be called the Representational Stance.
Second, we are beginning to see that the empirical project of isolating microtakings -- the component tasks of complex behavior -- and localizing them in discrete regions of the brain is breaking down.
On this second point, we may make a case study of the current state of the art in functional brain imagery, particularly Positron Emission Tomography (PET). The avowed goal of this highly successful research program is the isolation of component tasks of complex cognition, and PET experimental design and interpretation all contribute to this end. Over the last few years, the results of hundreds of experiments have been archived in a database known as Brainmap (ric.uthscsa.edu/projects/brainmap.html). One can ask of each function, each "microtaking," chosen for study, whether it indeed is isolated in a specific component of the brain. To begin with some overview, Brainmap archives 733 distinct experiments (PET, MRI, and EEG), with a total count of local maxima of activation of 7508. That is a mean of 10.24 activation peaks per experiment. It is a rare experiment where all these peaks are located in a single region, suggesting distribution of function. But this in itself is not definitive, since the average experiment might just as well be picking out a dedicated subnetwork of 10 (plus or minus) components. A dedicated processor need not be packed into a single anatomical box.
A more decisive analysis would work through a list of components, asking of each whether it is a locus of activation for specific functions. Perhaps the narrowest anatomical specification of the brain accessible to PET discrimination is the cortical Brodmann area. Brodmann areas are distinct both in their geography and in their cytoarchitecture, two factors which indicated to Brodmann and generations to follow that each of these numbered areas were functionally distinct. Well, are they? As part of a larger project I have tabulated all of the reported areas of activation in thirty-five PET experiments. Thirty-eight Brodmann areas are involved. Table 1 compiles the result.
TABLE 1 LEGEND:
The case for distribution suggested by Table 1 is even stronger if one factors in the many steps of PET study design that favor localist interpretation, the assignment of "microtakings" to regions of the brain. Foremost among these interpretive filters is the "subtraction method." Each image is in fact a difference image, the result of a subtraction of a control condition from a test condition. Often the controls are components of the task. For example, to locate semantic processing the experimenters might use a control scan of subjects reading pseudowords, to isolate just the distinctive components of the task in question. Even after this selective pre-screening of the data, however, the table shows distinct multifunctionality for most of the areas. Last but not least, the experiments indexed here are but a tiny slice of all the potential functions of the mind.
II. Getting (phenomenally) real
In homage to Professor Dennett I have turned Dennett's razors against the revered foundation of cognitive science, representation, suggesting that it is a Cartesian legacy founded in Descartes' deepity and vulnerable to Dennettian intuition pumps. From the toolbox, I have used a Dennettian analogy between the big fish of intentionality, conscious states, and the small fry, Dennett's "micro-takings" (Strategy 3B, above); A slippery slope argument about the boundaries of even the tiniest representational state (3D); And an empirical challenge (3E). But it is also in the spirit of Dennett to pick up the pieces, surveying the prospects and future of the field, and from there forge a new world view (as is often the burden of the intuition pumps, 3A).
How could we even imagine a post-cognitive post-representational view of the mind or brain? How can we have thoughts without a thought balloon to contain them? Towards this end, I'd like to replace the thought balloon with a trial balloon, a tentative effort to continue along the path Dennett has hewn. How do we turn from Descartes all the way? This is what we must abandon, somehow: On the one hand, the world at large is described by science, and concerning it (brains included) we are realists. On the other hand, conscious experience is described by phenomenology, to which is traditionally assigned nasty properties like idiosyncrasy and inescapable subjectivity. The entities of phenomenology do not match the entities of science, and as a result it is common to suppose that "phenomena" are somehow inside our heads. So our scientific realism is married to "phenomenological phenomenalism." As such, we suppose that any examination of phenomenological entities is a special exercise in introspection. However, this inward turn mistakes a fundamental fact about experience (and its phenomenological description), namely, that its domain is also the world, that same real world that science images. Phenomenology simply describes it in very different terms, subject to different epistemic standards. Descartes' legacy emerges primarily in the inward turn for phenomenology (reified, by the way, in the Cartesian theater) and secondarily in the failure to reconcile phenomenal language with scientific realism. (The impulse to discard either science or phenomenology is accordingly a "reactive Cartesianism," no less presupposing Cartesian metatheory than dualism.) We need, therefore, a more inclusive ontology, which somehow accords equal standing to both electrons and thoughts. I will call this as-yet-undefined ontology "Phenomenal Realism" (PR).
It's going to be a brain-bender, so in the best Dennett tradition we must begin with an intuition pump. After scouring the history of modern philosophy, one comes across the following example:
Return to the object that would be an equally good sail mender or cherry pitter, and suppose now we find it in the company of cherries. Cherries and the object pair up in this in this context, in a special way, expressed in a simple counterfactual: Had cherries been different in some respects - bigger, tougher, cubical -- the object, the cherry pitter, would have been different. We can, accordingly, speak of a "functional complex": the cherry pitter and the cherries. Though both pitter and cherries are separate entities, the idea of the functional complex underlines a dense network of causal and counterfactual relations between the two. To predict the fate of either the cherries or the pitters, their participation in their functional complex is relevant. Moreover, properties emergent from the functional complex may be referred to its components: In a world of pitters, cherries have the property of being pittable. And, among cherries, pitters exhibit the capacity to pit. (Among sails, they exhibit the capacity to mend.)
The idea of a functional complexes, exotic as it is, is simply illustrative. We are concerned to dissolve the Cartesian mind-world divide, and to that end let us ponder phenomenal complexes, the basic entities of Phenomenal Realism. In deference to another famous philosopher, the argument for Phenomenal Realism will be transcendental. That is, it develops from the premise that phenomenal properties are worldly, and asks, what must their ontology be, in order to found this worldliness? We approach via several successive transcendental approximations:
1. Level the playing field. The "realism" shared by Phenomenal Realism and Scientific Realism must be the same realism. Phenomenal properties (P properties) must meet the standards for reality applied to scientific properties (S properties). But how can this be? S properties are the good old primary qualities, defined in part by their existence independent of observers. A rock has mass regardless of who, if anyone, notices. But P properties do not exist in the absence of observers. Hot sauce is hot only if there is some observer to whom its spiciness is noticeable. Does observer-dependence impugn the reality of P properties? It depends on the dependency. If the observation is sufficient for the reality of the P property, then illusory and hallucinatory properties join the real; no ontology can tolerate that. On the other hand, if observer-dependence is merely necessary for P property realization, there is no problem insofar as everything has necessary conditions. What, then, will be necessary and sufficient for P property realization? The first approximate answer is simply that P properties are complexes, somehow incorporating both the object exhibiting the P property and the (potential) observer of that property. The whole complex, then, is real in the ordinary sturdy sense.
Real observers and a real object of observation together compose a phenomenal complex. But must the observer really observe and must the object really be observed? It seems to me that the heat of hot sauce is a disposition, and as such it need not be manifest in particular observation. That means that as long as there is at least one observer, somewhere, to whom the hot sauce would be hot, then the sauce itself has the P property -- even if that observer is forever light years away. (This is analogous to the cherries having the property of being pittable. Only in worlds with cherry pitters of some sort can cherries be pitted. But a pittable cherry need not be pitted.) This further entails that there are many P properties as yet unknown. These would be properties we could notice, should they cross our sensory path. Or, they could be properties noticeable to other sorts of observers -- Martians who sense gamma rays, for example.
Nonetheless it often happens, as a contingent fact, that P properties are observed. We do notice them; and here too the analogy with S properties is exact. Both kinds of properties can be noticed. Both often are noticed. But neither must be noticed. There are, therefore, two roles for the observer to play in the ontology of P properties. First, essentially, an observer with the capacity to observe the P property must exist. Second, contingently, it may happen that an observer really does observe a P property, just as it may happen that an observer really does observe an S property. The deepity is the conflation of the essential condition, the capacity, with the contingent realization of the capacity. The confusion arises naturally because it is the experience -- the realization or actualization of the capacity -- that we are most interested in. (What is it like to be a bat? Vs. What sorts of properties might a bat be able to notice, if there were bats?)
So far, the leveling of the playing field between S properties and P properties has exactly paralleled Dennett's treatment of qualia in CE and elsewhere (especially 1988 ). In CE, he begins with the observation that colors are in no way the intrinsic, unitary properties of objects that they seem to be, but "in reality" are both relational and disjunctive. But he initially resists (as have I) the impulse to reel these unruly secondary properties into the mind. Instead, he explains how it came to be that sensory systems could ever develop an interest in "gerrymandered" properties like color. The answer, suggested to him by Akins 1989 , is that many species have coevolved mutually advantageous relations of color display and color detection. While it is not clear whether an evolutionary story is an essential condition for the existence of phenomenal property types, the main tenet of Dennett's doctrine of qualia grants the reality of gerrymandered qualitative properties, as in Phenomenal Realism. That is, P properties are dispositional properties of the world that are present when there are potential observers (somewhere, as opposed to observer-dependent properties that presuppose actual observation). An emerald can be green despite being forever buried two miles deep; but it can be owned only if someone actively instantiates the appropriate economic relations toward the gem itself.
Dennett summarizes, with reference to "Otto," his fictional qualophiliac antagonist:
Scientific realism offers us two events, red-light-at-location-L1-at time-t1 and green-light-at-L2-at-t2. Both must have some effect on the observer's brain. But it requires a special assumption to conclude that each event in the world gives rise to a corresponding event in the brain -- that there is a specific brain-event which is the microtaking with the content "green light" and that it is taken in every case where a green light is detected. Dennett's discussion admits only the "official" stimuli (denizens of science) into the outbound description of the case. What is seen therefore must be composed of internal states, illusory seemings. It is as if the S properties preempt the P properties. In Dennett's interpretation of color phi, the S properties determine a set of S representations, and the P properties are woven around them.
Phenomenal realism, in contrast, would begin with a phenomenal complex consisting of a light (red or green) and the perceiver's brain state, noting that there is no registration of green flashes as such, or simpliciter. That is, a green flash following a red flash gives rise to a different phenomenal complex than a green flash alone, owing to the fact that the initial red flash has specific effects on the observer, and those effects alter the effect of the green flash. Specifically, the observer in the color phi experiment "registers" both the red and green stimuli as a single stimulus, a moving light that changes color in mid-trajectory. The appropriate comparison case in PR is one in which the stimulus is a genuinely moving light that changes color in mid-trajectory. The observer would form exactly the same percept following this stimulus as in the illusory, color phi, condition. From the point of view of Phenomenal Realism, the two indiscernible stimuli both exhibit the same phenomenal property, in spite of their (scientifically) real difference. This phenomenal conflation is the breakout from the Cartesian representationalist scenario: Under PR, there is no color-phi illusion. That is, there is no inner shadow (of a moving color-changing spot) to be contrasted with the outer "reality." Instead, there is a disjunction of outboard conditions which is accurately (but ambiguously) observed.
In discussing another example, Dennett makes what I regard as exactly the right observation:
As it might be in the perception of motion, so also in many other phenomenally real cases. For example, a cherry can enter into a phenomenal complex as I notice it. That complex consists of the cherry and the inflections of the state of my brain that depend on the cherry. Phenomenal complexes are clearly personal and changeable, due to the Heraclitan brainstorm of neural inflection. They can be changed at will, by simply attending to other aspects and relations within the phenomenal field. It is natural to try to extract stable elements from phenomenal complexes. The standard method is to develop special new properties that are relatively invariant across observers and times of observation. These are the variables and measurements of science. In other words, scientific realism is a special case of phenomenal realism. In the limiting case, science discovers common factors at the objective pole, enabling one to ignore the subject pole.
To wrap up the first approximation, note that to see a P property as a P property, a higher-order observer must notice the entire phenomenal complex, both the sauce and the taster to whom the sauce would be hot. This will be a complicated epistemic undertaking. In short, the first order observer experiences the world; The second-order observer experiences a phenomenal world, which he or she can distinguish from the scientific world. Both kinds of observation might occur in a single observer, at different times. Nonetheless, we must carefully distinguish these two forms of experience.
On to the second approximation:
2. Observers are brains. "Observation" and "observers" cannot be reassuring to a recovering Cartesian. The next approximation is to put material brains in for the very vague "observers." This approximation of PR thus holds that P properties are properties that are detectable by brains. So also S properties, although the detection might well be mediated by inference or instrument. Because the brain is itself so complex, the inflection appended to the cherry can be nuanced and detailed. It can embrace everything from my awareness of the cherry's color to my remembrance of cherries past, their similarities and differences, the role of cherries in literature, their role in exciting a food allergy, and so on. Each of these, I would argue, are inflections in the occurrent experience of particular cherries, as opposed to distinct further thoughts. They are a phenomenal, albeit non-sensory, halo that alters the cherries in our experience. But the halo is real: Connectionists have shown that distributed representations are capable of encoding all these cherry nuances and cherry contexts. The halo of inflection I just described is implemented in a distributed neural representation. It is a real component of the phenomenal complex that binds the cherry to the cherry experience as a single complex entity ( Lloyd, 1995 1996 ).
But the wonderful brain is just a place-holder for undecifered complexity. Because we have not yet said what it is about the brain that makes it an observer, the door is not yet closed on representational models of consciousness. We have still not cut the Cartesian umbilical cord. On to the third approximation:
3. Brains are detectors. The fundamental divide between S properties and P properties concerns the independence of S properties. Since no event is either uncaused or without consequence, this is a special sense of independence. I suggest that the most minimal sense of independence enjoyed by S properties is that they can exist undetected. This means not just that they happen to escape detection (which is also true for P properties) but that they can exist happily in a universe where no appropriate detector exists. That suggests, in turn, that the minimal condition for expelling a property from the paradise of scientific realism is its dependence on a detector.
What might a detector-dependent property be? Let us look again at the simpler S properties. An S property, as part of the map of the scientific image of reality, owes its robust reality to its place in the system of regularities that could be described (but not created) by scientific theory. Its role in the lawful description of nature explains why it need not be paired with a detector to be real -- it already has real work to do, regardless of whether notice accrues to the work. (Its straightahead engagement in the causal fray is also what makes it so detectable.) A P property, in contrast, does not have a role in the lawful description of reality, except insofar as it is detected. A P property might, for example, be a complex disjunction of S properties. Without a detector to make something of it, the disjunction, although each disjunct is perfectly real, does not find expression in any law of nature or scientific theory. The detector changes that, because the event of detection (the change of state of the detector itself) is a unitary event and as a result specific lawlike regularities may radiate from there. Detection, in other words, gathers up the disjuncts and makes something of them. But without a detector, there is no gathering up, no bundling of the disjuncts, no property, no there there. The kinds of detectors that resonate to P properties, accordingly, are more complex than S detectors, since the properties they detect are complexes. And, as in the earlier approximations, once the world has a P detector in it, the P property can be harnessed. Then, once the detector and the detectee both exist, the phenomenal property exists too.
The idea of detectors and detection, in short, preserves the core intuitions of Phenomenal Realism: P and S properties are both real, but S properties can endure in worlds without detectors specific to them, while P properties cannot. P properties are constituted by devices that detect them (even though those devices might never see action).
4. Detection is not representation. But what is detection? In "Real Patterns," Dennett invokes a mathematical conception of randomness and its complement, pattern:
Furthermore, there is no guarantee that the gerrymandered P property really does support a compressed representation. That is, it could be random. In that case, a detector would be a template for the random pattern. Still, it detects. Thus, compressed description cannot be an essential condition for detection. In short, we should resist the temptation to say that the detector encodes the stimulus it detects.
These claims greatly enlarge the class of detectors, especially detectors of P properties. The broken glass on the sidewalk, for example, could arise via a disjoint set of possible causes, any one of which may have been the real cause. That disjoint set of causal conditions thus becomes a P property. But, appearances notwithstanding, this is a welcome result. Recall that the stripped down detectors described here may signal the presence of a P property to us, but that interpretation is lost on the detector itself. To detect a P property as a P property requires detection of an entire phenomenal complex. That is, to see the bottle as an indicator of its possible causal conditions requires an explicit apprehension of at least some of those causal conditions. That, however, is just what the phenomenology of the case suggests: The broken glass, and every other object in the world, is shadowed by a story. In the act of seeing we read that story, with all its ambiguities and ellipses. An important part of the constitution of phenomenal objects is exactly their open-ended origins and uncertain destinies. Our awareness of these aspects of objects and events, although it is not sensory awareness, is distinctly part of our ongoing experience of the world.
Since two components comprise the phenomenal complex,
and since both must be real for the P property to exist,
to perceive a P property as a P property demands
the detection of the whole complex by a "meta
detector." As a small consolation to the Cartesian,
consider the continuum of P property meta-detectors, lest
we underestimate their formidable cognitive and pragmatic
abilities. These are no mere thermostats. To do their
job, they must not only be accurate but thorough, because
to ascertain the existence of a P property one must
search hither and yon for its appropriate detector (since
the mere existence of that detector suffices to
constitute the P property). In a very simple world,
consisting, say, of two keys and two locks, such a search
might be quite mechanical (although even here a real
detector will have to solve most of the hardest questions
posed in AI to do its job). But in our world of infinite
joints and vast search spaces, the establishment that
some property is a P property depends on a massive
background of implicit know-how, aided by myriad
heuristics. You'd have to have a big brain to pull it
off. The bigger the brain, the harder to deny it
We turn instead to the causal antecedents of plus-states or "experiences." The cryptos could claim that plus-states are sui generis, utterly disconnected from any detectable properties, but this would be a weird claim to make. Surely everyone must grant that experience, for all its magic, must somehow fit with some aspects of the observed or detected world. So, whatever lends the plus to D+ must itself correlate with some other property. And everyone also agrees that this property is material.
With this concession, the cryptos yield a critical toehold. For now there is a distinction between a D+ experiencer and a mere detector that can be expressed in terms of a property the D+ device is sensitive to. This is tantamount to enlarging the sphere of properties for the detector to detect. D+ turns out to be a detector after all -- at least part of the distinction is in the complexity of the properties it can detect.
The argument from here is inductive. Every time the cryptocartesian insists that something is missing, the phenomenal realist can answer by translating the interior, Cartesian, position into a realist analogue involving a new detectable property. That property can be phenomenal, in the sense discussed above.
This brings us to the second part of the crypto claim. The Cartesians not only point to something left out, the plus-stuff, but also make the claim that a materialist theory of consciousness cannot account for this residue. Phenomenal realism accounts for the reality of phenomenal properties, and explains how minds/brains/detectors could be sensitive to phenomenal properties as well as scientific properties. A detector is a functional type, however, so its full explanation would also account for its implementation. Here we drop down to level of structure and material. But this is exactly what good old scientific explanation handles best. In short, PR rebuts the deepity that materialist theories of consciousness "leave out the phenomenal." PR embraces the phenomenal in the only sense a realism could, by reconstruing it as a complex of properties that can be detected by highly specialized devices. PR tells us what these phenomenal detectors have to be able to do, and it expands easily to accommodate the most elaborate demands of the cryptocartesian. Scientific realism handles the rest, the implementation-level description of the phenomenal detectors. The cryptos can offer only two kinds of "something else," a missing phenomenal property or an intrinsic state of the conscious system, and the dynamic merger of phenomenal realism and scientific realism are prepared for either one.
But why carve nature at these joints? With the phenomenal complex bundled in this way, there is no additional need to refer or project the brain-based inflections onto the world. That reference or projection is superfluous, a Cartesian extravagance. Thus ends the thought balloon. But it is not the end of thoughts. Rather, the human world is reconceived as a thoughtful world, in which brains and other entities engage in a fabulous ongoing dance.
IV. Non sunt multiplicandi mundi praeter necessitam.
To summarize: Cartesianism split the world in two, creating an inboard model of an outside world. Ever since Descartes, science and philosophy have preserved this fundamental divide. One of Daniel Dennett's enormous contributions to philosophy has been his agile skepticism toward all the shoots and tendrils of the Cartesian mind-world picture. The endless creativity of his persuasions intimate the resolution of the Cartesian rift, and have offered me the incentive to take Dennett to the limit, scuttling even the concept of representation.
If you really want consciousness explained, mental content isthe wrong place to look. Phenomenal Realism dumps all of the contents of consciousness into the world. Not into an "outer" world, nor into an "inner" one. Just into the one world, in which the brain is just one entity among others. The brain is special only by virtue of the subtlety and idiosyncrasy of its spatiotemporal integrations. To sort the world into S and P properties is subtle indeed, and it takes a big brain to handle the huge set of stimulus energies our sensory neurons transduce.
With content in sight and out of mind, what remains of consciousness itself is simply the complex pattern of neural activity, ever shifting according to its own dynamics and the pushes and pulls of the proximal environment. Our state of mind is simply what we are (or, more precisely, what are brains are). Dennett is right to banish qualia, and should likewise purge "micro-takings" and "drafts" and every other intentional or representation item as well from the account of consciousness. But at the same time thw world of Phenomenal Realism is far more lush than the world of science. (The Phenomenal Realist, accordingly, welcomes Ross's Rainforest Realism, this volume.) This tradeoff could finally cure the hangover from Descates deepity.
But backsliding is a continual risk. I'll conclude by offering dual substance abusers -- and who among us is really immune? -- a Five Step program:
1. Renounce dual substance abuse. Instead, acknowledge
the higher explanatory power of monism.
Dennett's philosophy elegantly moves us through the first four steps. Here, I've outlined a possible fifth, possibly final step. But really the essential commitment is in the first step. One must overcome the temptation of Cartesian duality, whether in its full-blown metaphysical form or in all of its stand-ins. One must keep one's gaze steadily on the one world, and resist the multiplication of worlds with the same zeal that one resists the multiplication of entities. As Ockham might put it, in response to the Cartesian tradition, do not multiply worlds beyond necessity. There is one world -- we live there, entirely within it. We create the problems of intentionality and consciousness at the moment we create an ersatz shadow world, the "mental model." Can we live without this crutch? We'll see, one day at a time.
Akins, K. (1989). On Piranhas, Narcissism and Mental Representation. (un pub)
Andreasen, N. (1998). Remembering the past: two facets of episodic memory explored with positron emission tomography. American Journal of Psychiatry, 152, 1576-1585.
Baars, B. (1996). &127;&127;&127;&127;&127;In the Theater of Consciousness. Oxford, UK: Oxford University Press.
Bottini, G., Paulesu, E., Sterzi, R., Warburton, E., Wise, R.&127;&127;&127;&127;, Vallar, G., Frackowiak, R., & Frith, C. (1995). Modulation of conscious experience by peripheral sensory stimuli. Nature, 376, 778-781.
Dennett, D. (1987). Beyond Belief. In Intentional Stance (pp. 117-202). Cambridge, MA: MIT Press
Dennett, D. (1988). Quining Qualia. In A. Marcel & E. Bisiach (Eds.). Consciousness in Contemporary Science (pp. 42-77). New York: Oxford University Press
Dennett, D. (1991a). Real Patterns. Journal of Philosophy, 88, 27-51.
Dennett, D. (1991b). Consciousness Explained. Boston: Little, Brown.
Dennett, D. (1993). Back from the Drawing Board. In B. Dahlbom (Ed.). Dennett and his Critics (pp. 203-235). Oxford, UK: Blackwell
Dennett, D. (1996). Consciousness: More Like Fame than Television. In C. Maar, E. Po:ppel & T. Christaller (Eds.). Die Technik auf dem Weg zure Seele Rowohlt
Dennett, D. & Kinsbourne, M. (1992). Time and the Observer: The where and when of consciousness in the brain. Behavioral and Brain Sciences, 15(2), 183-248.
Dretske, F. (1981). Knowledge and the Flow of Information. Cambridge, MA:: MIT Press.
Fodor, J. (1998). Concepts: Where Cognitive Science Went Wrong. Oxford, UK: Oxford University Press.
Fox, P., Fox, J., Raichle, M., & Burde, R. (1985). The role of cerebral cortex in the generation of voluntary saccades: a positron emission tomographic study. Journal of Neurophysiology, 54(2), 348-369.
Fox, P., Burton, H., & Raichle, M. (1987). Mapping human somatosensory cortex with positron emission tomography. Journal of Neurosurgery, 67, 34-43.
Fox, P., Ingham, R., Ingham, J., Hirsch, T., Downs, J., Martin, C., Jerabek, P., Glass, T., & Lancaster, J. (1996). A PET study of the neural systems of stuttering. Nature, 382, 158-162.
Grasby, P., Frith, C., Friston, K., Bench, C., Frackowiak, R., & Dolan, R. (1993). Functional mapping of brain areas implicated in auditory-verbal memory function. Brain, 116, 1-20.
Haldane, E. & Ross, G.(trans.) (1969). Philosophical Works of Descartes, Vol. 1. Cambridge, UK: Cambridge University Press.
Hsieh, J. (1998). Urge to scratch represented in the human cerebral cortex during itch. Journal of Neurophysiology, 72, 3004-3008.
Jenkins, I., Bain, P., Colebatch, J., Thompson, P., Findley, L., Frackowiak, R., Marsden, C., & Brooks, D. (1993). A positron emission tomography study of essential tremor: Evidence for overactivity of cerebellar connections. Annals of Neurology, 34, 82-90.
Jueptner M, (1998). Localization of a cerebellar timing process using PET. Neurology, 45, 1540-1545.
Lloyd, D. (1989). Simple Minds. Cambridge, MA:: Bradford Books/MIT Press.
Lloyd, D. (1992). Toward an Identity Theory of Consciousness. Behavioral and Brain Sciences, 15, 215-216.
Lloyd, D. (1995). Consciousness: A Connectionist Manifesto. Minds and Machines, 5, 161-185.
Lloyd, D. (1996). Consciousness, Connectionism, and Cognitive Neuroscience: A Meeting of the Minds. Philosophical Psychology, 9, 61-81.
Pardo, J., Raichle, M., & Fox, P. (1991a). Localization of a human system for sustained attention by positron emission tomography. Nature, 349, 61-63.
Pardo, J., Raichle, M., & Fox, P. (1991b). Localization of a human system for sustained attention by positron emission tomography. Nature, 349, 61-63.
Pardo, J., Pardo, P., & Raichle, M. (1993). Neural correlates of self-induced dysphoria. American Journal of Psychiatry, 150, 713-719.
Paus, T., Petrides, M., Evans, A., & Meyer, E. (1998). Role of the human anterior cingulate cortex in the control of oculomotor, manual and speech responses: a positron emission tomography study. Journal of Neurophysiology, 70, 453-469.
Reiman, E. (1998). Neuroanatomical correlates of anticipatory anxiety. Science, 243, 1071-1074.
Seitz, R., Roland, P., Bohm, C., Greitz, T., & Stone-Elander, S. (1991). Somatosensory discrimination of shape: tactile exploration and cerebral activation. European Journal of Neuroscience, 3, 481-492.
Seitz, R. & Roland, P. (1992). Learning of sequential finger movements in man: a combined kinematic and positron emission tomography (PET) study. European Journal of Neuroscience, 4, 154-165.
Tempel, L. (1998). Abnormal cortical responses in patients with writer's cramp. Neurology, 43, 2252-2257.
Weiller, C., Isensee, C., Rijntjes, M., Huber, W., Mueller, S., Bier, D., Dutschka, K., Woods, R., Noth, J., & Diener, H. (1995). Recovery from Wernicke's aphasia: a positron emission tomographic study. Annals of Neurology, 37, 723-732.
Wessel, K., Zeffiro, T., Lou, J., Toro, C., & Hallett, M. (1995). Regional cerebral blood flow during a self-paced sequential finger opposition task in patients with cerebellar degeneration. Brain, 118, 379-393.