Phenomenal transparency and cognitive self-reference1 · PHENOMENAL TRANSPARENCY AND COGNITIVE© 2003 SELF-REFERENCE 353 Phenomenology and the Cognitive Sciences 2: 353–393, 2003

353PHENOMENAL TRANSPARENCY AND COGNITIVE SELF-REFERENCE

Phenomenology and the Cognitive Sciences 2: 353–393, 2003. © 2003 Kluwer Academic Publishers. Printed in the Netherlands.

Phenomenal transparency and cognitive self-reference1

THOMAS METZINGERPhilosophisches Seminar, Johannes Gutenberg-Universität Mainz, 55099 Mainz, Germany(E-mail: [email protected]; http://www.philosophie.uni-mainz.de/metzinger/)

Received 1 August 2003; received in revised version 2 September 2003

Abstract. A representationalist analysis of strong first-person phenomena is developed (Baker1998), and it is argued that conscious, cognitive self-reference can be naturalized under thisrepresentationalist analysis. According to this view, the phenomenal first-person perspectiveis a condition of possibility for the emergence of a cognitive first-person perspective. Cogni-tive self-reference always is reference to the phenomenal content of a transparent self-model.The concepts of phenomenal transparency and introspection are clarified. More generally, Isuggest that the concepts of “phenomenal opacity” and “phenomenal transparency” are in-teresting instruments for analyzing conscious, self-representational content, and that their rel-evance in understanding reflexive, i.e., cognitive subjectivity may have been overlooked inthe past.

Key words: consciousness, epistemic transparency, phenomenal transparency, representa-tion, self-consciousness, self-model

Introduction

In this paper I will argue that conscious, cognitive self-reference can benaturalized under a representationalist analysis. In order to sketch such arepresentationalist analysis I will introduce two new concepts, the concept of“phenomenal transparency” and the notion of “attentional availability forintrospection.” I also apply these new tools to an interesting challenge fornaturalism posed by Lynne Baker (1998). In particular, I will point to poten-tial equivocations between phenomenological and epistemic notions of “trans-parency,” arguing that certain Cartesian thoughts of the form [I am certain thatI* exist] can be reduced to a specific form of phenomenal mental content, andhence do not per se support the conclusion that they present us with genuineself-knowledge. I then briefly review some pathological cases in which psy-chiatric patients actually seem to hold and consistently express the firm be-

354 THOMAS METZINGER

lief that they do not exist. One goal of this paper is to show how central, butfrequently overlooked phenomenological constraints for representationalisttheories of self-consciousness are, first, that phenomenally opaque states doexist, and second, that they are interestingly related to phenomenally trans-parent states (namely within one and the same “vehicle”). My more generalaim in this paper is to draw attention to the fact that the concepts of “phenom-enal opacity” and “phenomenal transparency” are interesting instruments foranalyzing conscious, self-representational content, and that their relevance inunderstanding reflexive, i.e., cognitive subjectivity may have been overlookedin the past.

Phenomenal transparency

The concept

The locus classicus for the notion of phenomenal transparency is usually givenas G.E. Moore’s paper “The refutation of idealism”:

[. . .] the fact that when we refer to introspection and try to discover what the sensation ofblue is, it is very easy to suppose that we have before us only a single term. The term “blue”is easy enough to distinguish, but the other element which I have called “consciousness”– that which a sensation of blue has in common with a sensation of green – is extremelydifficult to fix. [. . .] And in general, that which makes the sensation of blue a mental factseems to escape us; it seems, if I may use a metaphor, to be transparent – we look throughit and see nothing but the blue; we may be convinced that there is something, but what itis no philosopher, I think, has yet clearly recognized. (Moore 1903, p. 446)

Today, a broad standard definition of phenomenal transparency, on which mostphilosophers would probably agree, is that it essentially consists in only thecontent properties of a conscious mental representation being available forintrospection, but not its non-intentional or “vehicle-properties.” Introspec-tively, we can access its content, but not the carrier of this content. Typically,it will be assumed that transparency in this sense is a property of all phenom-enal states.

• DEF 1: Phenomenal states are transparent in that only their content-prop-erties are introspectively accessible to the subject of experience.

Below, I will argue that this definition is unsatisfactory, because it violatesimportant phenomenological constraints. It is phenomenologically implausi-ble, because introspective unavailability of the carrier is not a necessary con-dition for phenomenality: non-intentional and vehicle properties frequently


are accessible for introspection. It may, therefore, be interesting to rememberthat Moore in his original paper pursued the same philosophical intuition:

[. . .] that the moment we try to fix our attention upon consciousness and to see what,distinctly, it is, it seems to vanish: it seems as if we had before us a mere emptiness. Whenwe try to introspect the sensation of blue, all we can see is the blue: the other element isas if it were diaphanous. Yet it can be distinguished if we look attentively enough, and ifwe know that there is something to look for. (Moore 1903, p. 450)

In § 275 of the Philosophische Untersuchungen Wittgenstein pointed to thenaïve realism inevitably brought about by transparent, phenomenal experi-ence.2 Interestingly, many authors today have returned to the notion of trans-parency, employing it as a useful conceptual instrument. Robert van Gulick,in an important early paper, has developed a functionalist analysis of trans-parency in terms of the speed, reliability, and global interdependence of phe-nomenal representations in combination with an accompanying lack of accessto earlier processing stages.3 For qualia, Sydney Shoemaker has pointed outthat we have no introspective access to the nonintentional features of ourexperience that encode this content.4 Gilbert Harman has defined the trans-parency of experience as an unawareness of intrinsic non-intentional features.5

Also, Michael Tye now uses the concept at many places in many of his writ-ings,6 making the strong and interesting claim that phenomenal character isactually identical with intentional contents.

Let me now introduce my own working definition of phenomenal trans-parency, the definition to which I want to refer hereafter. Transparency in thissense is a property of active mental representations already satisfying theminimally sufficient constraints for conscious experience to occur. Some ofthese constraints might even be necessary conditions. For instance, phenom-enally transparent representations are always activated within a virtual win-dow of presence and are functionally integrated into a unified global modelof the world. Conscious content is always bound into an experiential “Now,”and functionally it is intimately connected with those mechanisms constitut-ing short-term memory and the emergence of a psychological “moment” (seePöppel 1988 for an accessible overview). In addition, in non-pathologicalconfigurations, the content of consciousness forms a single and integratedmodel of reality; phenomenologically, being conscious simply is to be in a world(e.g., Yates 1985; for an excellent recent collection of texts, see Cleeremans2003). Rather than discussing these constraints and the logical relations amongthem here, I will simply assume that such a set exists (for a brief list, seeMetzinger 2000, fig. 20.1; for a more extensive discussion see Metzinger 2003,chapter 3).


The second defining characteristic of phenomenal transparency postulatesthat, on the functional level of analysis, what makes mental representationstransparent is the attentional unavailability of earlier processing stages in thebrain for introspection. Transparency results from a structural/architectonicproperty of the neural information-processing going on in our brains. Thenotion of “attentional availability for introspection” is here used with refer-ence to a special case, namely that of introspective attention (further analyzedbelow).

What is attention? In short, attention is a process of subsymbolic resourceallocation taking place within a representational system exhibiting phenom-enal states. It is a form of non-conceptual metarepresentation operating oncertain parts of the currently active, internal model of reality, the consciousmodel of the world. By guiding our attention towards a perceived object, weachieve a selection and an enhancement. By turning towards the phenom-enal representation of this object, we automatically intensify the informa-tion processing in the brain, which underlies it. Simultaneously we increaseour degree of alertness and orient towards the object in question. However,we do so without mentally forming a concept for this object. Typically theobject becomes more salient in this act, it becomes richer in detail and morecoherent by being segregated from a background at the same time – a new andbetter representation is generated. Within neuroscience, a popular assumptionis that – strictly speaking – the object is first of all constituted through atten-tion, because a number of single feature detectors are now functionally inte-grated into a larger whole, a dynamically bound assembly. In this way, attentionis a constructive form of representing another representation (or a setthereof) with the help of non-conceptual mental means, a subsymbolic kindof metarepresentation. Object perception never is a passively synthetic proc-ess, because many target stimuli can only be discovered after their elemen-tary features have already bound into a whole. And this is another centralfunction of attention: it identifies a certain characteristic combination of fea-tures within an ambiguous perceptual context. For example, visuo-spatial at-tention helps to segment a phenomenal scene and thereby contributes to areduction of ambiguity.

Within the representational architecture of the human mind, the guiding ofattention is a supramodal capacity. For instance, attention can be shifted aroundindependently of eye-movements, i.e. it can move within the visual field whilethe position of the eyes remain stable. Husserl frequently spoke of the Blickstrahl,the ray of the glance. Another beautiful and classical phenomenological meta-phor is that of describing visuo-spatial attention as a “cone of light.” This meta-phor is consistent with a functional analysis of attentional metarepresentation,


because objects in the cone of light of attention are processed in a better, faster,and deeper fashion. On the level of neuroscience we find that, for instance,spatial attention functionally penetrates early stages of visual informationprocessing, by selectively inhibiting certain paths of processing, and enhanc-ing others. Although there presently exists no comprehensive neuroscientifictheory of attention, it is quite clear that it is a limited-capacity process, whichcan directly modulate the response profile of individual neurons and can bedescribed neither as a property of any narrowly circumscribed anatomicalregion nor as a property of the brain as a whole. Our attention system is con-stituted by a large number of single and interacting networks, which can bespecialized in a highly diverging manner (see Gazzaniga 1995, part V; Posner1995).

Attention can also be directed inwards, for instance at the process generat-ing a mental self-representation, while continuously embedding it into ourconscious model of reality. In this context, the following fact is of particularphilosophical interest: The more earlier the processing stages, the more ear-lier aspects of the internal construction process leading to the final, explicitand disambiguated phenomenal content, are available for introspective atten-tion, the more will the system be able to recognize these phenomenal statesas internal, self-generated constructs. Full transparency means full attentionalunavailability of earlier processing stages. Degrees of opacity come as degreesof attentional availability.

• DEF 2: For every phenomenal state, the degree of phenomenal transpar-ency is inversely proportional to the introspective degree of attentional avail-ability of earlier processing stages.

This definition diverges from earlier notions of phenomenal transparency inallowing us to describe three important facts about phenomenal conscious-ness, which philosophers have frequently overlooked. First, cognitive avail-ability of the fact that currently active phenomenal contents are the finalproducts of internal representational processes is not enough to dissolve orweaken phenomenal transparency. The naïve realism of our experience re-mains. One cannot “think oneself out of” one’s phenomenal model of realitywith the help of purely cognitive operations alone. To simply have a mentallyrepresented concept of the book you are holding in your hand, perhaps asbeing only a special form of representational contents, does not change theuntranscendably realistic character of your phenomenal experience at all – atleast not in a way that would be relevant in the current context. However, thereseems to be a relevant difference between cognitive and attentional process-


ing, between conceptual and non-conceptual metarepresentation of first-or-der phenomenal states. Only if you could actually attend to the constructionprocess “itself” would you experience a shift in subjective experience, namelyby adding new and non-conceptual content to your current model of reality.

Second, this definition departs from the classical vehicle-content distinc-tion. The standard way of defining transparency would be to say that onlycontent properties of the phenomenal representata are introspectively avail-able to the system, and not vehicle properties. The vehicle-content distinctionis a highly useful conceptual instrument, but it contains subtle residues ofCartesian dualism in that it always tempts us to reify the vehicle and the con-tent by conceiving of them as ontologically distinct, independent entities. Amore empirically plausible model of representational content will have todescribe it as an aspect of an ongoing process and not as some kind of ab-stract object. What we need is embodied content, as it were – or rather, not“a” content, but an ongoing and physically realized process of containing (see,e.g., Churchland 1998, [unpublished manuscript]; Clark 1997; O’Brien andOpie 2001).

Third, describing phenomenal transparency in terms of the attentional avail-ability of earlier processing stages has the advantage of being able to developmany different, fine-grained notions of degrees of transparency and opac-ity. For different phenomenal state-classes resulting from different types ofprocessing, it may also be possible to describe not only variable degrees, butdistinct kinds of transparency and opacity. This allows for a much more real-istic description of certain phenomenological features pertaining to differentclasses of conscious states.

The phenomenology of transparency

Transparency is a special form of darkness. With regard to the phenomenol-ogy of visual experience transparency means that we are not able to see some-thing, because it is transparent. We don’t see the window, but only the birdflying by. The negative fact that we don’t see the medium, the window, is it-self not explicitly represented in the seeing process itself.

Phenomenal transparency in general, however, means that something par-ticular is not accessible for subjective experience, namely the representationalcharacter of the contents of conscious experience.7 This analysis refers to allsensory modalities and to our integrated phenomenal model of the world as awhole in particular – but also to large parts of our self-model (Metzinger 2003).The instruments of representation themselves cannot be represented as such,


and hence the system making the experience, on this level and by conceptualnecessity, is entangled in a naive realism. This happens, because, necessarily,it now has to experience itself as being in direct contact with the current con-tents of its own consciousness. What precisely is it, which the system cannotexperience? What is inaccessible to conscious experience is the simple factof this experience taking place in a medium. Therefore, transparency of phe-nomenal content leads to a further characteristic of conscious experience,namely the subjective impression of immediacy. Many bad philosophical ar-guments concerning direct acquaintance, infallible first-person knowledge anddirect reference are based on an equivocation between epistemic and phenom-enal immediacy: from the fact that the conscious experience, e.g., of the colorof an object, carries the characteristics of phenomenal immediacy and directgivenness it does not follow that any kind of non-mediated or direct kind ofknowledge is involved. Of course, there may be coincidences of phenomenalimmediacy and epistemic immediacy. But phenomenal content – as such – isnot epistemic content, and it is a widely held and plausible assumption that itlocally supervenes on brain properties.8 For every veridical perception therewill be a hallucinatory state, indistinguishable from the first-person perspec-tive. What is common between the two is their phenomenal content (I like tocall it the “lowest common denominator,” or, more traditionally, “the highestcommon factor”; see Metzinger 2004). Phenomenal content can be dissoci-ated from intentional content: a brain in a vat could possess states subjectivelyrepresenting object colors as immediately and directly given. Any fully trans-parent phenomenal representation is characterized by the vehicle-generatingmechanisms, which have led to its activation, plus the fact of a concrete in-ternal state now being in existence and carrying its content, not being intro-spectively available anymore. The phenomenology of transparency, therefore,is the phenomenology of naive realism.

Of course, opaque phenomenal representations do exist as well. Here aresome preliminary examples of opaque state-classes: Most notably consciouslyexperienced thoughts, but also some types of emotions, pseudo-hallucinationsor lucid dreams are subjectively experienced as representational processes.Such processes sometimes appear to us as deliberately initiated cognitive orrepresentational processes, and sometimes as spontaneously occurring, lim-ited or even global phenomenal simulations, frequently not under the experi-ential subject’s control. Lucidity, becoming aware that one is dreaming in adream, is the standard example of a situation in which all of the consciousmodel of reality is suddenly experienced as a model (see Metzinger 2003,section 7.2.5). In such cases we know that they do not present realities to us,but only possibilities: the fact that they are representational processes, the


content of which may or may not properly depict an external reality, is glo-bally available for attention, cognition and behavioral control (for the notionof “global availability,” see Baars 1988; Chalmers 1997; Metzinger 2000).Many authors describe phenomenal transparency as an all-or-nothing phenom-enon. To do phenomenological justice to conscious experience, however, de-mands a more differentiated description. As I will show towards the end of thispaper, perhaps unexpected and far-reaching consequences for the logic andepistemological status of self-consciousness follow from such a description.

Therefore, let us take a second and closer look at some phenomenologicalexamples in which the transparency-constraint is satisfied to different degrees.Sensory experience is the paradigmatic example of fully transparent phenom-enal content. There are, however, examples of sensory opacity, for instance,during extremely short transition phases in bistable phenomena, e.g., if a con-sciously experienced Necker cube switches from one interpretation to the nextand back, or during the phenomenon of binocular rivalry (see, e.g., Leopoldand Logothetis 1999). On the next level of sensory awareness there is thephenomenon of simple, sensory pseudo-hallucinations. If a subject in a labo-ratory experiment and under the influence of a hallucinogenic psychoactivesubstance (say, LSD or 2-CB) observes abstract geometrical patterns on thewall, breathing, and slowly evolving into deeper and ever deeper forms ofineffable beauty, then she will frequently be aware of the representationalcharacter of her visual experience in that sub-region of her phenomenal space.Typically, the subject will immediately have doubts about the veridicality ofher experiential state, cognitively “bracket” it and take back the “existenceassumption,” something which effortlessly goes along with visual experiencein standard situations. My claim is that what this subject becomes aware ofare earlier processing stages in her visual system. Visual pseudo-hallucina-tions – the breathing patterns on the wall – are such earlier processing stages.Today, we have the first mathematical models describing the self-organiz-ing dynamics characterizing such unstable states of the visual cortex (e.g.,Bressloff, Cowan, Golubitsky, Thomas, and Wiener 2001). These models offerprecise predictions about the phenomenological form constants characteriz-ing such simple, context-invariant and opaque sensory states, and they per-mit certain conclusions about the functional architecture of the visual system.They also give us a first idea of the type of self-organizational dynamics, theactual earlier processing stages, which enable the non-propositional, sub-sym-bolic form of metarepresentation, which then leads the subject to the experi-ence of opaquely hallucinating. Of course, complex hallucinations, which arefully transparent and in which the experiential subject gets lost in an alterna-tive model of reality, do exist as well (see, e.g., Siegel and West 1975).


More importantly, however, the paradigmatic examples for fully opaquestate-classes are deliberately initiated processes of conscious thought. In thesecases we really experience ourselves as deliberately constructing and operat-ing with abstract representations, ones that we have generated ourselves andwhich can, at any time, turn out to be false. We are cognitive and epistemicagents at the same time, as thinking subjects actively trying to achieve anexpansion of knowledge. In particular, we are introspectively aware of ear-lier processing stages, namely the formation of thoughts. Cognitive referenceis phenomenally opaque.

It must be noted that there are also forms of thought, which are localizedon the other end of the spectrum between phenomenal transparency and phe-nomenal opacity. If we slide into manifest daydreams, we frequently experi-ence cognitive processing not as cognitive processing any more. A further pointto be noted when discussing the phenomenology of transparency and opac-ity: not only can individual phenomenal contents exhibit a variable degree oftransparency, the same is true of global phenomenal world-models as well.Right after a traffic accident the whole world can appear as “unreal” or as“dreamlike” to us. The same phenomenon is known in stress situations andin transitory phases during certain psychiatric syndromes (“derealization”).However, the best and most basic example for an almost fully opaque, globalphenomenal state is the lucid dream (see LaBerge and Gackenbach 2000;Metzinger 1999, 2003).

Returning from global phenomenal state-classes to particular examples, itis of great interest to investigate the transparency and opacity of emotions.As opposed to sensory and cognitive states, emotions are neither a paradig-matic example of transparency nor of opacity. Our emotional states frequentlyseem to be directly given forms of subjective self-representation. Their con-tent is something we do not doubt, but just take as reliable, immediately giveninformation about our own current state and about our relation to other hu-man beings. However, sometimes we suddenly become aware that our emotionalresponse might actually be inappropriate to our current social environment. Takejealousy as an example: we may suddenly realize the representational char-acter of our own jealousy, if we become aware of the fact that all this actuallymight be a misrepresentation – for instance, of those persons in our socialenvironment about whom we are jealous. Then – just as in the case of thehallucination recognized as a hallucination – what previously was experiencedas an objective, external property of another person suddenly becomes a prob-lem one may potentially have oneself, as the content of one’s own mental state.In emotions we frequently oscillate between certainty and uncertainty, be-tween the immediacy of an obvious perception and doubts. This simple


phenomenological observation points to another important characteristic ofopaque phenomenal representations. They make the possibility that they ac-tually might be misrepresentations globally available for cognition, attentionand behavioral control. They allow a representational system to experienceitself as a representational system.

Summing up, my three central claims about phenomenal transparency are:

• Transparency is not a necessary condition for phenomenality, becauseopaque forms of phenomenal content do exist.

– Examples in the phenomenal world-model (of external reality): Pseudo-hallucinations, lucid dreaming.

– Examples in the phenomenal self-model: consciously experienced cog-nitive reference/self-reference.

• Transparency is a gradual property of phenomenal states, because thedistribution of transparency/opacity in phenomenal space possesses vari-ance.

– Examples in the phenomenal world-model: Stress-situations, transitorystates in psychiatric disorders.

– Examples in the phenomenal self-model: Emotions, transitions betweenreflexive thought and daydreaming.

• Any convincing theory of phenomenal transparency/opacity must be ableto explain how we could, phylo- as well as ontogenetically, acquire this prop-erty in a series of steps.

– Two potential strategies: Differences in length between periods of evo-lutionary optimization for sensory, emotional, and cognitive processing;limited temporal resolution of “introspective” metarepresentation in thebrain.

Three classical equivocations

Let me now proceed to describe the three most important equivocations orpotential misunderstandings of the notion of “phenomenal transparency,” asintroduced in DEF 2. Such misunderstandings do exist, they are quite com-mon, but a clarification can also be used to further enrich the target concept.There are three different, but well-established usages of “transparency,” twoin philosophy and one in communication theory.


First, transparency is not an epistemological notion, but a phenomenologicalconcept. In particular, it has nothing to do with the Cartesian notion of epistemictransparency, the philosophical intuition that in principle I cannot be wrongabout the content of my own consciousness, that the notion of an unnoticederror in introspectively accessing the content of your own mind is incoher-ent. Descartes famously expressed this idea in the last paragraph of his sec-ond Meditation, and the advance of clinical neuropsychology today makes thisclassical philosophical assumption about the human mind untenable. The studyof anosognosia, neglect and many other disorders has demonstrated that un-noticed (and unnoticeable) errors about the contents of one’s own conscious-ness can occur at any time. The modern concept of phenomenal transparency,however, is systematically related to the Cartesian project in so far as it presentsan important building block in a theory that attempts to make the overwhelmingintuitive force behind this false assumption intelligible: Any conscious sys-tem operating under a phenomenally transparent self-model will by necessityinstantiate the phenomenal property of selfhood in a way that is untranscendablefor this system itself (see note 10, and Metzinger 1999, 2000, 2003). If oneapplies the notion of phenomenal transparency to the conscious-model a sys-tem has of itself as a whole, one can see how such a system will – due tophenomenal immediacy – inevitably feel infinitely close to itself, and willbecome a naïve realist about itself at the same time. The Cartesian claim aboutthe epistemic transparency of self-consciousness can itself not be epistemicallyjustified, but it has the great advantage of correctly describing the phenom-enology of certainty going along with the target phenomenon, the self-certaintyof the subject.

Second, transparency is here conceived of as a property of phenomenalrepresentations in a sub-symbolic medium, i.e., of non-linguistic entities un-der an empirically plausible theory of mental representation, and not as aproperty of a context. The second potential equivocation is the extensionality-equivocation: “Phenomenal transparency” is used not as a notion belongingto the field of formal semantics, but rather as a new concept in philosophicalneurophenomenology. Transparency as a property of contexts is somethingentirely different. Extensional (i.e., referentially transparent) contexts areconstituted by sentences, which are characterized by the intersubstitutivity ofcoreferential expressions salva veritate and by an implication towards theexistence of the entities mentioned by them. Intensional (i.e., referentiallyopaque) contexts are constituted by sentences characterized by an inter-substitutivity of expressions with identical meaning salva veritate. Such con-texts do not preserve the same truth value if co-referential expressions at theindividual variable x or at the place held by the predicate letter F are substi-


tuted for each other. Opaque contexts are, for instance, constituted by refer-ence to propositional attitudes, to temporal and modal expressions, or by in-direct speech. Whoever concludes from the fact that a certain expression xcannot be substituted by a co-referential expression y, salva veritate, withinan intensional context, that x and y do not refer to the same aspect of reality,commits what philosophers call the “intensional fallacy.”

Transparency as a property of contexts is not what I am talking about here.Phenomenal transparency can exist in non-linguistic creatures, lacking anyform of cognitive reference. Phenomenally transparent representations couldsupervene on a brain in a vat, whereas referentially transparent ones arguablycould not. However, certain complex phenomenal contents can potentiallyconstitute certain types of linguistic contexts, and in particular opaque phe-nomenal contents contributing to higher-order forms of mental self-represen-tation can do so (see below). It is interesting to note, how, once again, there isa connection between referential and phenomenal transparency: in both caseswe have something like an implication towards the existence of the entitiesrepresented. Sentences constituting extensional contexts imply the existenceof the entities mentioned within them. Fully transparent phenomenal repre-sentations force a conscious system to functionally become a naive realist withregard to their contents: whatever is transparently represented is experiencedas real and as undoubtedly existing by this system.

There is a third common use of the notion “transparency,” which shouldnot be confused with the notion as here intended, but which at the same timeexhibits a third interesting semantic parallel. It too is a well-established con-cept. Transparency can be conceived as a property of media. For instance, intechnical systems of telecommunication, transparency can be the property ofa channel or of a system for the transmission of information in general. Togive an example, transparency in this sense can be a feature of a server in theInternet. Defining characteristics for this notion of transparency are, first, thatit accepts unmodified user information as input; second, that it delivers userinformation that is unmodified with respect to form and informational con-tent on the output side; and third, that although user information may well beinternally changed and re-processed in many different ways, it is alwaysretransformed into the original format before reaching the output stage with-out causal interaction with the user. E-mail is the obvious example. A singlemessage that you send to your friend may be chopped up in many differentportions, each of these taking many different roads and “hops” through thenet before being reassembled on the other side. The user has no access to thesubpersonal mechanisms underlying successful personal-level communica-


tion. Obviously, phenomenal transparency in the sense here intended is not aproperty of technical systems. However, it is interesting to again note theparallel that emerges if we look upon the neural correlate of consciousnessor the conscious model of reality as a medium: This medium is transparentinsofar as the subpersonal processing mechanisms contributing to its cur-rently active content are attentionally unavailable to high-level introspec-tive processing.

What we now need is a clearer understanding of the notion of internalityunderlying this concept of introspective processing. What, exactly, does itmean to say that for any phenomenal state, the degree of phenomenal trans-parency is inversely proportional to the introspective degree of attentionalavailability of earlier processing stages?

Introspection

There are many different notions of internality relevant to the investigationof phenomenal and cognitive mental content: temporal internality (constitut-ing the experiential “Now”), functional internality (as in the closing of sen-sorimotor loops, constituting complex phenomenal contents like experientialembodiment or situatedness), or representational internality (as is relevant forself-consciousness). In our present context, it is of particular relevance todistinguish between introspection as operating on first-order states the phe-nomenal content of which only supervenes on physically internal system statesand introspective operations which additionally represent their intentionalcontent as actually being an aspect of the system itself.

Simultaneously, we have to apply the distinction between attention and cog-nition. Attention is a selection process, which episodically increases the ca-pacity for information-processing in a certain partition of representationalspace. Functionally speaking, attention is internal resource allocation. Atten-tion, as it were, is a representational type of zooming in, serving a local eleva-tion of resolution and richness in detail within an overall representation. Itgenerates non-conceptual content, for which frequently we possess no trans-temporal identity criteria and which therefore is ineffable (Metzinger andWalde 2000; Raffman 1995). Cognition, on the other hand, enables recogni-tion, supports linguistic concept formation and generates a form of contentwhich at least emulates compositionality, syntacticity, systematicity, etc. It willbe helpful to distinguish four different notions of introspection, as there aretwo types of internal metarepresentation, a subsymbolic, attentional kind(which only ”highlights” its object, but does not form a mental concept), and


a cognitive type (which forms or applies an enduring mental ”category” orprototype-centred region of state-space for its object).

• DEF 3: Introspection1 (“external attention”). Introspection

1 is sub-sym-

bolic metarepresentation operating on a pre-existing, coherent world-model.This type of introspection is a phenomenal process of attentionally repre-senting certain aspects of an internal system state (this is why it can be con-ceptually described as a form of introspection), the intentional content ofwhich is constituted by a part of the world depicted as external.

The accompanying phenomenology is what we ordinarily describe as atten-tion or the subjective experience of attending to some object in our environ-ment. Introspection1 corresponds to the folk-psychological notion of attention.

• DEF 4: Introspection2 (“consciously experienced cognitive reference”).

This second concept refers to a conceptual (or quasi-conceptual) form ofmetarepresentation, operating on a pre-existing, coherent model of the world.This kind of introspection is brought about by a process of phenomenallyrepresenting a cognitive reference to certain aspects of an internal systemstate, the intentional content of which is constituted by a part of the worlddepicted as external.

Phenomenologically, this kind of introspection is constituted by all experi-ences of attending to an object in our environment, while simultaneouslyrecognizing it or forming a new mental concept of it: it is the conscious expe-rience of ongoing cognitive reference. A good example is what Fred Dretske(1969) called “epistemic seeing.”

• DEF 5: Introspection3 (“inward attention” and “inner perception”). Thisis a subsymbolic metarepresentation operating on a pre-existing, coherentself-model, i.e., a dynamic, multimodal representation of the system as awhole (for the notion of a “self-model” see Metzinger 1999, 2000, 2003).This type of introspective experience is generated by processes of phenom-enal representation, which direct attention towards certain aspects of aninternal system state, the intentional content of which is being constitutedby a part of the world depicted as internal.

The phenomenology of this class of states is what in everyday life we call“inward-directed attention.” On the level of philosophical theory it is this kindof phenomenally experienced introspection that underlies classical theories


of inner perception, e.g., in John Locke or Franz Brentano (see Güzeldere 1995for a recent critical discussion).

• DEF 6: Introspection4 (“consciously experienced cognitive self-refer-

ence”). This type of introspection is a conceptual (or quasi-conceptual) kindof metarepresentation, again operating on a pre-existing, coherent self-model. Phenomenal representational processes of this type generate con-ceptual forms of self-knowledge, by directing cognitive processes towardscertain aspects of internal system states, the intentional content of which isbeing constituted by a part of the world depicted as internal.

The general phenomenology associated with this type of representationalactivity includes all situations in which we consciously think about ourselvesas ourselves (i.e., when we think what some philosophers call I*-thoughts;for an example see Baker 1998, and the next section). On a theoretical level,this last type of introspective experience clearly constitutes the case in whichphilosophers of mind have traditionally been most interested: the phenomenonof cognitive self-reference as exhibited in reflexive self-consciousness.

Obviously the first two notions of introspection viz. introspective availabil-ity are rather trivial, because they define the internality of potential objects ofintrospection entirely by means of a simple physical concept of internality. Inthe present context, internality as phenomenally experienced is of a higherrelevance. However, it may be useful to remember the principle of local su-pervenience for phenomenal content, which is highly plausible on empiricalgrounds: For all forms of mental representation referred to in DEF 3–6 it istrue that their phenomenal content is fixed as soon as all internal and contem-poraneous physical properties of the respective system are fixed.

Cartesian thoughts and cognitive self-reference

What is a cognitive first-person perspective as opposed to a phenomenal first-person perspective, what is a cognitive subject as opposed to a phenomenalself? As I will argue in this section, the former is a special case of the latter:The possession of a phenomenal first-person perspective is one, and arguablythe most relevant, necessary precondition for a cognitive first-person perspec-tive to come into existence. My claim is that, all other constraints for perspectivalphenomenality satisfied, a conscious cognitive subject is generated as soonas a globally available representation of the system as currently generatingand operating with the help of quasi-linguistic, opaque mental representations


is integrated into the already existing transparent self-model. “Opaque” and“transparent” are now used in accordance with DEF 2. It is vital to arrive at aconvincing analysis of this specific form of phenomenal content, because thiswill provide us with an understanding of what it means to possess a cognitivefirst-person perspective, as opposed to a merely phenomenal first-person per-spective. A phenomenal first-person perspective could already be realized byan animal or conceivably by a primitive artificial system operating under afully transparent mechanism of attentional processing as described in the pre-vious section (for the concept of a “phenomenal model of the intentionality-relation,” see Metzinger 2000, 2003, chapter 6.5; Metzinger and Gallese 2003).Here I will try to illustrate the issue by drawing on an example from the re-cent philosophical discussion.

In an important and helpful paper Lynne Baker (1998) has pointed out howa convincing conception of the cognitive first-person perspective will consti-tute a test for any robust version of naturalism. Baker differentiates betweentwo classes of first-person phenomena. Weak first-person phenomena are thosewhich, for instance, animals can be conceived of as exhibiting, while operat-ing under an egocentric world-model forming the center of their own universeand the origin of their own perspective. As Baker points out, all sentient be-ings are conscious subjects of experience, but not all of them have first-personconcepts of themselves. For Baker, only those who do are fully self-consciousin an interesting sense (Baker 1998, p. 328).9 Given the conceptual tools in-troduced above it is clear what such a description amounts to: Such animalswould use an integrated, global and transparent model of the world func-tionally centered by a transparent self-model in order to regulate their ownbehavior. As Baker correctly points out, such organisms could be said to besolving problems by employing perspectival attitudes, while not yet possess-ing a concept of themselves as a subject.

First-person phenomena in a stronger and more interesting sense, however,are not only characterized by the necessary condition of possessing a self-world-boundary and being able to differentiate between the first and thirdperson, but also the capacity to possess this distinction on a conceptual leveland actually using it. In the terminology so far introduced, this means that theexistence of a pre-attentive self-world-boundary and the difference betweenfirst and third person attributions is cognitively available, in terms of DEF 4and DEF 6. As Baker points out it is not only necessary to have thoughts, whichcan be expressed using ”I.” What is necessary is the possession of a conceptof oneself as the thinker of these thoughts, as the owner of a subjective pointof view. In short, what is needed is not only reference from the first-person


point of view, but the capacity of mentally ”ascribing” this act of reference tooneself while it is taking place. Here is how Baker makes the point:

A conscious being who exhibits strong first-person phenomena not only is able to recognizeherself from a first-person point of view (. . .), but also is able to think of herself as herself.For strong first-person phenomena, it is not enough to distinguish between first-person andthird-person; also one must be able to conceptualize the distinction, to conceive of oneselfas oneself. To be able to conceive of oneself as oneself is to be able to conceive of oneselfindependently of a name, or description or third-person demonstrative. It is to be able toconceptualize the distinction between oneself and everything else there is. It is not just tohave thoughts expressible by means of ”I,” but also to conceive of oneself as the bearer ofthose thoughts. (. . .) But merely having a perspective, or a subjective point of view, is notenough for strong first-phenomena. Rather one must also be able to conceive of oneself ashaving a perspective, or a subjective point of view. (Baker 1998, 329 pp.)

This conceptual distinction is important for cognitive science in general, butalso for the philosophical notion of a true cognitive subject. As Baker notes,this capacity may also be important for so-called “Theory-of-Mind-Tasks”:Only if one is able to think about oneself as being the thinker and the subjectof first-person thoughts, is one also able to form the concept of other subjectsof first-person thoughts. The capacity at issue would also enable one to con-ceive of one’s own desires and wishes as one’s own – for instance, in terms ofsecond-order volitional acts in the sense of Harry Frankfurt (1971).

In the light of the theoretical model here proposed it is quite clear what Bakeris actually demanding. The representational architecture needed for cognitivesubjectivity is not only a self-representation including certain cognitive ac-tivities as one’s own, plus a mental representation of oneself being the initia-tor of these internal events, i.e., a cognitive agent. What is really needed is arepresentational structure which makes this very fact globally available forhigher-order cognition.

It is important to note that the concept of a phenomenal self-model (PSM;elsewhere proposed, see above, and Metzinger 2000, 2003) achieves preciselythis capacity. If a system integrates its own operations with opaque mentalrepresentations, i.e., with mental simulations of propositional structures thatcould be true or false, into its already existing transparent self-model whilesimultaneously attributing the causal role of generating these representationalstates to itself, the system as a whole – all other necessary constraints forphenomenal representation satisfied – will not only possess the consciousexperience of cognitive agency, of being the thinker and the subject of its ownthought. It will also make this new phenomenal content available for higher-order cognitive processing: It now possesses the additional capacity to forma concept of itself as a cognitive agent.


Baker suggests, in addition:

If I attribute first-person reference to myself, my sentence cannot be adequately paraphrasedby any sentence that fails to attribute first-person reference to me: The attribution of first-person reference to oneself seems to be ineliminable. (Baker 1998, p. 331)

I disagree. However, my aim is not to directly refuse the semantic irreduc-ibility of de-se-attitudes, but to make a contribution towards a deeper under-standing of the phenomenal content mediating first-person self-reference. Iwant to offer an alternative. First, no such things as selves exist in the world;all that exists are conscious systems operating under transparent self-models.Second, we need to differentiate between the contents of linguistic and ofmental representations. Linguistic self-reference always refers to the phenom-enal content of the self-model, typically to its transparent portion (as the PSMis what functionally enables linguistic self-reference in the first place). Be-cause this transparent portion cannot be experienced as a form of representa-tional content, the fact that a linguistic content refers to a mental content cannotbe consciously experienced. This fact, however, must be reflected in our theoryof self-consciousness. One way of paraphrasing this sentence according to thecurrent model would be: ”This system currently uses a sentence in a publiclanguage to refer to a certain capacity, namely, the capacity to access the con-tent of certain opaque, cognitive simulations integrated into an already exist-ing transparent self-model by higher-order cognitive operations, which thenin turn can be integrated into this self-model.” Let me explain.

According to Baker, possession of a first-person perspective in the moreinteresting sense consists in possessing a certain capacity. It is the capacityfor forming I*-sentences or I*-thoughts (the asterisk is here used followingthe common notation introduced by Hector-Neri Castañeda 1966).10 As Bakercorrectly adds, a complete mastery of ”I” includes this capacity. She writes:“The first-person perspective is a necessary condition for any form of self-consciousness, and a sufficient condition for one form of self-consciousnessas well” (Baker 1998, p. 333). Baker then goes on to illustrate her point, us-ing a particularly beautiful example:

For example, there is no third-person way to express the Cartesian thought, ”I am certainthat I* exist.” The certainty that Descartes claimed was certainty that he* existed, notcertainty that Descartes existed. (Baker 1998, p. 336)

What really was the reason for Descartes’ certainty about his own existence?He possessed a transparent self-model that turned him into an “autoepistemicallyclosed” being: the representational nature of a considerable portion of his self-model was in principle unavailable to him, i.e., he could not gain introspective

3


knowledge about it. Autoepistemic closure, as represented by the phenomenaltransparency constraint being maximally satisfied for large partitions of thehuman self-model, must by necessity lead to the prereflexive phenomenal prop-erty of “selfhood,” which simply is the certainty about one’s own existence.11

Any representational system satisfying the constraints for conscious experiencewhile operating under a transparent self-model has to be certain of its own ex-istence. This is a conceptual necessity for the class of representational systemsdescribed, in all possible worlds. The linguistic expression ”I*” invariably re-fers to this content of the conscious, current transparent model of the self.

Baker is of course right in pointing out that Descartes’ claim did not referto any third-person representation of the person Descartes. What he referredto was a representation of Descartes ”in the mode of selfhood,” under a ”phe-nomenal EGO-mode of presentation” as it were (for this notion, see Newen1997).12 This mode is the mode of internal, transparent self-modeling. Thecontent of this transparent self-model, in particular the phenomenologicalcharacteristic of this content as being given in a seemingly direct and imme-diate manner, is cognitively available. If so, then, obviously, higher-ordercognitive content referring to this very fact can also be formed. If this con-tent, in turn, is embedded into the already existing phenomenal self-represen-tation, then the relevant conscious experience of having this thought willautomatically ensue. An additional and non-conceptual sense of ownershipwill emerge, as a new form of active, higher-order phenomenal content. Thisthought is Baker’s I*-thought. Its content is:

[I am certain that I* exist]

From now on, I will refer to active phenomenal content by using square brack-ets, and to linguistic expression referring to active phenomenal content byusing pointed brackets. <I> refers to the fact that this current conscious thoughtis a component of the self-model. This fact is available to cognitive introspec-tion operating on the self-model, in terms of DEF 6. It is phenomenally rep-resented as the experience of being the thinker of this thought (more below).<I*>, on the other hand, refers to the transparent partition of this self-model.The certainty predicated of the phenomenal self, and in particular the exist-ence assumption invariably going along with it, refers to a globally availablemental representation of the phenomenal quality of ”infinite proximity,” theindubitable (i.e., phenomenally necessary) experience of selfhood, of directgivenness-to-oneself caused by the transparency of the self-model.13

The sentence <I am certain that I* exist> therefore refers to a complex, in-tegrated and globally available form of phenomenal mental content, namely


to the currently active PSM of the respective person. Given the necessary out-put device, it could be uttered by a brain in a vat, as this content locally su-pervenes. The content of this self-model is: [I am certain that I* exist]. Fromnow on I will only be concerned with the phenomenal mental content and notwith the linguistic expression that, as Lynne Baker puts it, indicates the exist-ence of this content. However, please note the fascinating architecture of thisstructure: We are confronted with a form of metarepresentation crossing thelinguistic-phenomenal-divide between propositional and non-conceptual con-tent, a metarepresentation that is not experienced as such by the representingsystem. An internal, locally supervening form of phenomenal content becomesthe content of a linguistic representation in public space, with the speaker beingunaware of this very fact due to the transparency of the first-order mentalcontent. Let us take a closer look.

The issue I want to draw attention to is that the content of this self-modelpossesses a phenomenally transparent and a phenomenally opaque componentin terms of DEF 2. The degree of attentional availability of earlier processingstages varies. The transparent component of the self-model (e.g., the contentof the body-model) is experienced as directly and immediately given. Theopaque component is often, but not necessarily experienced as deliberatelyconstructed, for instance as one’s own thought. Both forms of content aresubsymbolic content. The opaque component often, if not consisting of picto-rial imagery etc., approximates constituent-structure, propositional modularity,etc. to a sufficient degree to be able to be interpreted as a quasi-linguistic formof mental content on the level of external autophenomenological reports.However, let us stay with the phenomenal content itself and begin with thetransparent component of Baker’s Cartesian thought.

[I*]

is the content of the transparent self-model. This content is activated underthe ”principle of autoepistemic closure” already mentioned above. Any con-scious system operating under a transparent self-model will by necessity in-stantiate a phenomenal self to which, linguistically, it must refer using <I*>.It is autoepistemically closed against earlier processing stages of its own self-model, because it cannot introspectively

3 discover its representational nature.

In other words, in Baker’s example the I*-thought is being constituted by thecontent of the transparent self-model, by a subset of its content properties.Please note how a natural background assumption is that the system alreadysatisfies all other constraints sufficient for at least a minimal degree of con-scious experience.


[I* exist]

denotes the decisive property of the transparent self-model. What are the gen-eral constitutive conditions for the phenomenal experience of ”existence”? Letus define a minimal notion of self-consciousness: the content of the self-modelhas to be embedded into the currently active world-model; it has to be acti-vated within a virtual window of presence; and it has to be transparent. There-fore not only the fact that the world-model is a model, but also the fact thatthe temporal internality of the contents of the window of presence is an inter-nal construct is not introspectively

3 available to the subject. Then there is

the special case of phenomenal self-presence, the subjective experience ofAnwesenheit: the fact that the self-model is a model and that the temporalinternality of its content as integrated into the current window of presence isan internal construct, again, is not available to the system as a whole. Thisproperty is a functional property (attentional unavailability of earlier process-ing stages, i.e. of certain vehicle properties), not a content property.

[I]

<I> refers to the speaker of a sentence, [I] is the phenomenal content of theopaque component of the current self-model, the thinker of this thought. Thefact that the current cognitive content is only a mental representation is intro-spectively available for attention as well as for cognition. In addition, we alsohave the structural characteristic of cognitive agency: I experience myself asthe thinker of the I*-thought. The opaque component of the self-model repre-sents the respective subject as a being, which deliberately generates the men-tal content in question within itself. Therefore, [I] internally models the systemas a whole, the content of the opaque component, which has already beenembedded into the continuously active background of the transparent self-model. The content of [I] is the thinker, currently representing herself as op-erating with mental representations. Please note that the content here referredto, as all phenomenal content, can not count as epistemically justified con-tent.

[am certain that]

is the subsymbolic equivalent of what would be an ”attitude specificator”(believing that p, wanting that p, being certain that p, etc.) of a propositionalattitude. It is the relation between self and cognitive content as it is currentlyphenomenally represented. What are constitutive conditions for phenomenal


certainty? Let me name the two defining characteristics: the object-compo-nent of the phenomenal first-person perspective is transparent and the respec-tive person is therefore, on the level of phenomenal experience, forced intoan (epistemically unjustified) existence assumption with respect to the inten-tional content of the object-component. The same is true of the subject-compo-nent. The second defining characteristic is the transparency of the self-model,yielding a phenomenal self depicted as being certain. Please note how a phe-nomenal first-person perspective now reveals itself as the ongoing consciousrepresentation of dynamic subject-object relations: To see an object, to feel apain, to selectively ”make a thought your own,” to choose a potential action-goal14 – or, to be certain of oneself, as currently existing.15

Being phenomenally certain about oneself, again, is a special case. It ap-pears if the object-component of the phenomenal first-person perspective isformed by a transparent self-representation; for instance by a phenomenalmodel of the respective person as a now existing subject. Please note how theobject-component of a phenomenal first-person perspective will typically betransparent, because this precisely is what makes it an object – something thatcannot be experienced as a representation any more. So we are confronted witha consciously experienced form of second-order self-representation. It is im-portant to note that the vehicles of representation are not distinct entities.

On the representational level of description, strong first-person phenom-ena and Cartesian I*-thoughts can now be analyzed as the integration of anopaque self-model into a preexisting, transparent self-model. What we areunable to consciously experience during cognitive self-reference is the factthat, even in this situation, we are referring to the content of a representationthat is ”in ourselves” (in terms of locally supervening on brain properties). Thisis a necessary consequence of the transparency-constraint for the subsymbolicself-model. Cognitive self-reference, therefore, on the phenomenal level isnecessarily experienced as direct and immediate, because it is not mediatedthrough any sensory channel (it takes place in a supramodal format) and be-cause of the fact that it is a second-order process of phenomenal representa-tion is not introspectively available (the naïve-realistic self-misunderstanding).Therefore, the possibility that the first-order component might be a misrepre-sentation is not available to us on the level of attentional introspection

3 ac-

cording to DEF 5.Let us now return to the level of linguistic self-reference. The overall sen-

tence <I am certain that I* exist> indicates that a certain form of phenomenalcontent is currently active. From a third-person perspective, it can be analyzedas follows:


<The speaker of this sentence currently activates a transparent PSM into which second-order, opaque self-representations have been embedded. These representations are char-acterized by three properties:

– First, they possess a quasi-symbolic format (e.g., through a connectionist emulationof constituent-structure, propositional modularity, etc.);

– second, their content is exclusively formed by operations on the transparent partitionsof the currently active PSM;

– third, the resulting relation between the system as a whole and the transparent contentis phenomenally modeled as a relation of certainty.>

Please note how it is the internal model of the relation between system andcontent, which, in this case, generates the cognitive first-person perspective.This cognitive first-person perspective, however, can always be reduced to aspecial case of the phenomenal first-person perspective and, as such, all cor-responding belief states about what actually is being represented – i.e., allbeliefs about the intentional or epistemic content of strong first-person phe-nomenal and Cartesian I*-thoughts – are not epistemically justified. In par-ticular, this is true of the belief that it is a self that carries out the respectiveact of cognitive self-reference.

Consistent misidentification and existence denial:Two short neurophenomenological case-studies

In this section I will briefly point to some empirical material, which demon-strates how conscious persons can actually (a) consistently misidentify them-selves with other persons, or (b) claim – with absolute certainty – that theydo not exist. Why is this relevant? If Baker’s Cartesian thought was a genu-ine form of epistemic, intentional content, it would seem to be a logical con-tradiction that such cases do exist. How could any rational subject know aboutits own non-existence? If, however, the relevant form of mental content is onlyphenomenal content, which locally supervenes on its minimally sufficientneural correlate, then one could at least conceive of such possibilities as re-sulting from severe brain dysfunction. We could – at least in principle – un-derstand how a rational subject could experience itself as another person oras non-existent.

The phenomenal experience of selfhood and cognitive self-reference canvary along an extremely large number of dimensions. Such variations consti-tute important phenomenological constraints to be satisfied by conceptualanalyses of the specific forms of mental content active in conscious, cogni-tive self-reference. For example, there are simple, but unnoticed losses of


content (as in blindness denial or Anton’s syndrome; see Anton 1898, 1899;Benson and Greenberg 1969). There are also various typologies of phenom-enal disintegration as in schizophrenia, Dissociative Identity Disorder (DID)and in various depersonalization disorders. These may be accompanied bymultiplications of the phenomenal self within one and the same physical sys-tem and major redistributions of the phenomenal property of “mineness”, i.e.the experiential sense of ownership. Finally, there are at least four differentdelusions of misidentification (DM;16 namely Capgras syndrome, Frégoli syn-drome, intermetamorphosis, reverse intermetamorphosis and reduplicativeparamnesia). These are, for example, of importance for philosophical theo-ries of self-consciousness and self-reference, because they violate the Witt-genstein/Shoemaker principle of immunity to error through misidentification(Shoemaker 1968; Wittgenstein 1953, p. 67). It is therefore important to testnew conceptual tools at least against some examples of the enormous phen-omenological richness of our target phenomenon.

The relevance of case-studies from cognitive neuropsychiatry consists inthe fact that they allow for “reverse engineering”: If we can develop an em-pirically plausible representationalist analysis of identity disorders, we willautomatically arrive at a better understanding of what precisely it means toconsciously be someone in standard situations. What are necessary and suffi-cient conditions for the cognitive forms of self-modeling underlying linguis-tic self-reference? As the history of science has shown, a closer inspection ofborderline cases of complex phenomena reveals implicit assumptions, helpsto dissolve intuitive fallacies and makes conceptual deficits of existing theo-ries clearly visible.

In the present context, at least two types of identity disorders are of directphilosophical relevance: A specific form of DM, and the Cotard delusion. Letus begin with reverse intermetamorphosis, usually defined as “the belief thatthere has been a physical and psychological change of oneself into anotherperson” (Breen, Caine, Coltheart, Hendy, and Roberts 2000, p. 75). As usual,I will argue that this kind of disorder rests on a deviant form of phenomenalself-modeling, and that in principle it could also occur in a non-linguistic/non-cognitive creature not able to form anything like “beliefs” in the more nar-row philosophical sense. Here is a brief excerpt from a recent case-study onpatient Roslyn Z, conducted by Nora Breen and colleagues:

RZ, a 40 year old woman, had the delusional belief that she was a man. This had been astable delusion for two months prior to our assessment with RZ. During most of that twomonths she believed that she was her father, but occasionally she would state that she washer grandfather. At the time we saw RZ, she had taken on the persona of her father. Shewould only respond to her father’s name, and she signed his name when asked to sign any


forms. She consistently gave her father’s history when questioned about her personal his-tory. For example, she said she was in her 60s. (. . .) The following excerpts are from aninterview with RZ. Throughout the interview RZ’s mother, Lil, was sitting beside her.

Examiner: Could you tell me your name?RZ: Douglas.Examiner: And your surname?RZ: B_____.Examiner: And how old are you?RZ: I don’t remember.Examiner: Roughly how old are you?RZ: Sixty-something.Examiner: Sixty-something. And are you married?RZ: No.Examiner. No. Have you been married?RZ: Yes.Examiner: What was your partner’s name?RZ: I don’t remember. Lil.Examiner: Lil. And you have children?RZ: Four.Examiner: And what are their names?RZ: Roslyn, Beverly, Sharon, Greg.(. . .)RZ standing in front of a mirror looking at her own reflection.Examiner: When you look in the mirror there, who do you see?RZ: Dougie B_____ (her father’s name).Examiner: What does the reflection look like?RZ: His hair is a mess, he has a beard and a moustache and his eyes are all droopy.Examiner: So is that a man or a woman?RZ: A man.Examiner: How old is Dougie?RZ: Sixty-something.Examiner: And does that reflection you are looking at now look like a sixty-somethingperson?RZ: Yes.Examiner: It looks that old does it?RZ: Yes.Examiner: Do you think that sixty-something year old man would have grey hair?RZ: Well, I haven’t worried a lot over the years so my hair didn’t go grey.Examiner: So it’s not grey?RZ: No. It’s brown.(Breen, Caine, Coltheart, Hendy, and Roberts 2000, pp. 94f., 98f.).

Reverse intermetamorphosis is of philosophical relevance because it chal-lenges the Wittgenstein/Shoemaker-principle of immunity to error throughmisidentification (Shoemaker 1968; Wittgenstein 1953, p. 67). I am not go-


ing to enter into a philosophical discussion of these cases here. I only want toillustrate what the notion of phenomenal transparency comes to when appliedto self-consciousness, and to the resulting self-ascription of personal-level,global properties of the cognitive system as a whole. Very obviously there arecases of phenomenal self-representation, of the phenomenal representation ofone’s own personal identity in particular, which are misrepresentations. Suchpatients, I would submit, linguistically (or cognitively) still refer to the con-tent of a transparent self-model – just that they consistently use an internalmodel of another person as their first-order self-representation. This delusionalself-model is the target of cognitive self-reference, and the fact that it is onlya model is neither cognitively nor attentionally available to the system as awhole. Misidentification is a symptom rather than a syndrome comprising astable collection of symptoms, and this particular variety of self-mis-identification is closely associated with severe psychotic states (Förstl,Almeida, Owen, Burns and Howard 1991, p. 908).

Let us take a look at a second type of identity disorder, because it too maybe of particular theoretical relevance. This disorder is what I would like tocall existence denial. In the year 1880 French psychiatrist Jules Cotard intro-duced the term délire de negation to refer to a specific kind of “nihilistic”delusion, the central defining characteristic of which consists in the fact thatpatients deny their own existence, and frequently even that of the externalworld (see Cotard 1880, for a more detailed account see Cotard 1882). From1879 onwards this condition was referred to as the “Cotard Syndrome” in thescientific literature (Séglas 1879; for a concise review of the literature seeEnoch and Trethowan 1991, pp. 163ff). Although there still is considerablediscussion about the notion of a delusion as such (see, e.g., Young 1999) andabout the conceptual status of “pathological” belief systems (see Coltheart andDavies 2000, Marshall and Halligan 1996) in general, most researchers tendto agree that the Cotard syndrome is likely a distinct theoretical entity.17 How-ever, I will here simply treat it as a neurophenomenological state-class char-acterized by a specific form of deviant self-modeling, without entering intoany further empirical speculations. As a delusion, the Cotard syndrome cer-tainly is quite dramatic – for instance, because it locally violates the globallogical coherence of the patients “web of belief” (see Young 1999, pp. 582ff)and simply ends biographical coherence, while on the other hand exhibitingonly a more or less modularized damage to the cognitive model of reality towhich the patient has conscious access. As in many delusions, it is the ratherisolated nature of a specific belief content that raises serious doubts about hisstatus as a rational subject. However, as a closer look at the data will reveal,a Cotard patient may count as a rational subject, in developing the only pos-


sible conclusion from a dramatic shift in his sub-cognitive PSM. A promisingattempt towards a testable and conceptually convincing hypothesis may there-fore start from the assumption that the Cotard disorder is simply a modularized,cognitive-level reaction to very uncommon perceptual experience (Young andLeafhead 1996). It must be clearly noted, though, that the phenomenology offirmly believing in one’s own non-existence may also turn out to be too intri-cate and complex to be tractable under classical belief-desire-approaches –for instance, because it decisively involves pathology in non-propositional lev-els of phenomenal self-representation.

In pure and extreme versions of the Cotard delusion patients may explic-itly state not only that they are dead, but also that they don’t exist at all. Inother words, something that seems an a priori impossibility on logical grounds– a conscious subject truthfully denying its own existence – turns out to be aphenomenological reality. And phenomenology has to be taken serious. Thereare actual, nomologically possible representational configurations in thehuman brain, which lead truthful subjects into logically incoherent auto-phenomenological reports. In the previous section, when discussing cogni-tive subjectivity as a challenge to naturalism, I offered a representationalistanalysis of the Cartesian thought:

[I am certain, that I* exist].

In extreme forms of the Cotard delusion, we are faced with a delusional be-lief that can be expressed as follows:

[I am certain, that I* do not exist.]

Weaker forms are delusional beliefs that can be expressed as follows:

[I am certain, that I* am dead.]

What is the phenomenological landscape of the Cotard delusion? A recentanalysis of 100 cases (Berrios and Luque 1995, see Fig. 2, p. 186) points outthat severe depression was reported in 89% of the subjects, with the most fre-quent forms of “nihilistic delusion” concerning the body (86%) and existenceas such (69%). Other very common features of the reported content of the PSMin Cotard patients are anxiety (65%), guilt (63%), hypochondrial delusions(58%) and, even more surprising, delusions of immortality (55%; see Berriosand Luque 1995, p. 187). In many cases, certain elements of the bodily self-model seem to have disappeared, or at least to be attentionally unavailable.


For instance, one 59-year-old patient would say “I have no blood” (Enoch andTrethowan 1991, case 5, p. 172), or another one would say “I used to have aheart. I have something which beats in its place” (p. 173) while a further casestudy (Ahleid 1968, quoted after Enoch and Trethowan 1991, p. 165) reportsa patient asking “to be buried because he said he was “a corpse which alreadystinks”. A month later he said that he had no flesh and no legs or trunk. Theseideas were unshakeable, so that the clinical picture remained unchanged formonths.” Such early stages will frequently proceed to states in which the bodyas a whole is denied, and the patient feels like a “living corpse,” e.g., whensaying “I am no longer alive”; “I am dead” (Enoch and Trethowan 1991, case1, p. 168) or stating “ “I have no body, I am dead” (Enoch and Trethowan 1991,case 2, p. 168) or, like Young and Leafhead’s patient WI, simply being con-vinced that he was dead for some months after a motorcycle accident (involv-ing contusions in the right hemisphere temporo-parietal areas and bilateralfrontal damage), with this belief then gradually resolving over time (Youngand Leafhead 1996, p. 154).

What seems unique about the Cotard delusion is the belief that one is dead.The Cotard patient, on the level of his cognitive self-model, acquires a new,introspectively

4 available content – and this content is specific in being highly

irrational and functionally immune to rational revision. The difference in thephenomenal content in the transparent and sub-cognitive layers of his self-model can be aptly described by employing a traditional conceptual distinction(one that plays a major role in the philosophy of Husserl and Merleau-Ponty),which is only available in the German language, but not in Greek (e.g., Homeruses only “soma” referring to corpses), Latin (with corpus only referring tothe body-as-thing, being the etymological root of the English term “corpse”)or other modern languages like Italian, French, Spanish or English: The Cotardpatient has a bodily self-model only as a Körper, but not as a Leib (see alsoAhleid 1968, and Enoch and Trethowan 1991, p. 179p). What is the differ-ence? A Leib is a lived body, one that is connected to a soul, or, in more mod-ern parlance, the body as subject, as the locus of an individual first-personperspective. The body as inanimate object, on the other hand, is what the PSMin the Cotard configuration depicts. The Cotard-patient only has access to aKörper-model, but not to a Leib-model.

It is interesting to note how one can arrive at a better understanding of theunderlying neurophenomenological state-class by simply following the tra-ditional line of thought. What kind of loss could make a Leib-model a repre-sentation of something inanimate, of something that is no longer tied to theinner logic of life? If the logic of survival is made globally available by con-scious emotions (cf. Damasio 1999), then a complete loss of the emotional


self-model should have precisely this effect. Philosophically speaking thiswould mean that what the Cotard patient claiming to be a dead corpse is truth-fully referring to is the transparent content of his self-model, predominantlyconcerning the spatial, proprioceptive and emotional layers. This contentportrays a moving res extensa, from the inside, closely resembling a livinghuman person, but, as a matter of phenomenal fact, not tied to the logic ofsurvival any more. In traditional terminology, the patient has a belief de se.As the first-order, non-conceptual self-representational content grounding thisbelief, in the most literal sense possible, simply no longer contains the infor-mation that actual elementary bioregulation is still going on, as emotions arecompletely flattened out, the patient forms a hypothesis. This hypothesis, givenhis current internal sources of information, is absolutely coherent: He mustbe a dead object resembling a human being. The existence of this object, al-though experienced as the origin of a phenomenal first-person perspective doesnot affectively matter to the patient in any way. As Philip Gerrans (2000, p.112) writes: “The Cotard delusion, in its extreme form, is a rationalization ofa feeling of disembodiment based on global suppression of affect resultingfrom extreme depression.” Many Cotard patients make utterances like “I haveno feelings” (Enoch and Trethowan 1991, case 4, p. 171) or, like Young andLeafhead’s patient KH, state that they are dead as a result of “feeling nothinginside” (Young and Leafhead 1996, p. 160). If it is the conscious self-modelwhich, as previously claimed, mediates embodiment not only on the phenom-enal, but also on the functional level, then these patients suffer from functionaldeficits, because they are, due to a severe impairment in their PSM, emotion-ally disembodied. It is a specific subset of system-related information, whichcan not be made globally available any more. This fact triggers further changeson the cognitive level of self-modeling.

As Gerrans (1999, p. 590) has importantly pointed out, on the level ofrepresentational content the Cotard delusion may be miscategorized if de-scribed as a DM. At least in weaker forms, the Cotard patient truthfully re-ports about the content of a very unusual PSM. This unusual PSM results froma globalized loss of affect, mirroring the global distribution of the neurochemi-cal substrate that causes the actual deficit, which in turn is then cognitivelyinterpreted. The issue is to first explain how such a PSM could actually comeabout. Classical belief-desire-psychology and a traditional philosophical analy-sis in terms of propositional attitudes may, however, not be very helpful inbridging the necessary levels of description in order to arrive at a fuller un-derstanding of the target phenomenon. For instance, it seems plausible that anon-linguistic creature like a monkey could suffer from most of the Cotardphenomenology, without being able to utter incoherent autophenomenological


reports. An animal could feel dead and emotionally disembodied, withoutpossessing the capacity to self-ascribe this very fact to itself, linguistically oron the level of cognitive self-reference. What is special about the human caseis that an isolated and functionally rigid new element on the level of the cog-nitive self-model is formed. As soon as we arrive at a convincing representa-tional and functional analysis of the human Cotard patient’s self-model, wecan proceed to the philosophical issue of whether it is possible to exhibit strongfirst-person phenomena in Lynne Baker’s (1998) sense, while simultaneouslyentertaining the belief that one actually doesn’t exist at all.

The Cotard delusion may be analysed as a combination of loss of a wholelayer of non-conceptual, transparent content and a corresponding appearanceof new, quasi-conceptual and opaque content in the patient’s PSM. Is there amore specific way of describing the causal role of the information no longeravailable on the level of the patient’s phenomenal model of reality? What trig-gers the massive restructuring of his “web of belief”? A second, and morespecific hypothesis may be the “emotional disembodiment”-conjecture: ThePSM of the Cotard patient is emptied of all emotional content, making it, inparticular, impossible to consciously experience familiarity with himself.What the Cotard patient loses may be precisely the phenomenal quality of“prereflexive self-intimacy.” The Cotard patient is not infinitely close to her-self, but infinitely distant. If, in addition, it is true that emotional content,generally speaking, represents the logic of survival to an organism, then self-representational emotional content, in particular, will represent the internallogic of autonomic self-regulation, of its own life-process to this organism.For the Cotard patient, the logic of survival has been suspended: His life-proc-ess – although functionally unfolding as still continuously realized in thephysical body – is not owned any more, by being represented under a PSM.His life-process is not only not his own any more, it may not even be part ofhis phenomenal reality any more – depending on the severity of the degree ofpsychotic depression. The fact that autonomous self-regulation, a continuousbodily process aiming at self-preservation is going on, but is not a globallyavailable fact anymore, means this fact is not a part of the patient’s realityanymore. The dynamic process of self-modeling is a process of self-contain-ing. A Cotard patient may therefore be described as a living, self-represent-ing system that can no longer self-contain the fact that it actually is a livingsystem.

However, there is more to the Cotard delusion than emotional disembodi-ment, leading to a persistent false belief via the principle of phenomenal self-reference (i.e., all cognitive self-reference ultimately can only refer to thecontent of the currently conscious self-model). Claiming to be dead – in terms


of a dead body - is not the same as claiming to be non-existent. As Enoch andTrethowan write:

Subsequently the subject may proceed to deny her very existence, even dispensing alto-gether with the use of the personal pronoun “I”. One patient even called herself “MadamZero” in order to emphasize her non-existence. One of Anderson’s patients said, referringto herself, “It’s no use. Wrap it up and throw “it” in the dustbin”.18 (. . .) If the delusionbecomes completely encapsulated, the subject may even be able to assume a jovial moodand to engage in a philosophical discussion about her own existence or non-existence.(Enoch and Trethowan 1991, p. 173; 175).

In this phenomenal state-class the phenomenal property of selfhood is notinstantiated any more, while a coherent model of reality as such is still inexistence. However, a Cotard patient may express his dramatic and general-ized emotional experience of unfamiliarity even by denying the existence ofreality as a whole. The phenomenon of explicit existence denial can not beignored, because on the level of explicitly negated content, it is the secondmost common representational feature of the Cotard delusion, to be found in69% of the cases (Berrios and Luque 1995, p. 187). Therefore, the pivotalquestion is: What kind of neurophenomenological configuration could lead ahuman being to (a) deny his or her own existence, and (b) stop using the per-sonal pronoun “I” (or “I*” for that matter)?

Let us begin by considering the first issue. The conscious representationof existence is coded via phenomenal transparency. Phenomenally, we are be-ings experiencing the content of a certain active representation as real, if andonly if, earlier processing stages of this representation are attentionally una-vailable to us. This leads to the prediction that if a human being’s self-modelbecame fully opaque, then this person would experience herself as non-exist-ent – the phenomenal property of selfhood would not be instantiated any more.The subject-component of such a being’s phenomenal model of the intention-ality-relation would be fully opaque, and only continue to function as the originof its first-person perspective for functional reasons, due to the ongoing sourceof continuous input making it the functional centre of its representational space.As there would be no phenomenal self as subject any more, such a systemwould not have a PSM in the true sense of the term any more, and its phe-nomenal model of the intentionality-relation, strictly speaking, would instan-tiate only a functional, but no longer a phenomenal first-person perspective.The second logical possibility is that in extreme Cotard configurations exist-ence is denied because there is no PSM at all in existence any more. This isempirically implausible, because Cotard patients exhibit a high degree ofsensorimotor integration, of coherent speech, etc. – they certainly are notcomatose or in deep sleep.


The third possibility, then, is that a transparent, conscious self-model is inplace, but it is not a subject-model any more, only an object-model. Some-thing still exists, something that looks like the model of a person, but some-thing that is utterly unfamiliar, not alive, and not a phenomenal self in the actof living, perceiving, attending and thinking. The PSM has lost the emotionallayer. In such a case, there would not be a model of a subject-object-relation,but only one of an object-object-relation. It would not constitute a phenom-enal first-person-perspective, but rather a first-object-perspective. The “firstobject,” for purely functional reasons, persists as the invariant centre of real-ity, because it is tied to an invariant source of internally generated input. Phe-nomenally, this functional centre is the place at which things happen, but allof them – as well as the centre itself – are not owned. The phenomenal prop-erty of “mineness,” the non-conceptual sense of ownership has disappearedfrom the patient’s reality. As Philip Gerrans puts it:

In this type of case the patient conceives of herself as nothing more than a locus, not ofexperience – because, due to the complete suppression of affect, her perceptions andcognitions are not annexed to her body – but of the registration of the passage of events.(Gerrans 2000, p. 118)

Let us now proceed to the second issue: Why would such a system stop usingthe pronoun “I” when referring to itself? It is plausible to assume that thesystem still operates under a functionally centred model of reality. Motorcontrol, attentional processing and cognitive availability are in place, and inprinciple well integrated. Sensorimotor integration is successfully achieved.Phenomenologically however, there seems to be a subjectless model of real-ity. On the representational level of analysis we find that there is no globallyavailable representation of a self as subject. Maybe we can give a new answerto the old philosophical question of what the personal pronoun “refers” to, incases of consciously experienced cognitive self-reference structurally resem-bling the one discussed above. <I*> inevitably refers to a specific form ofphenomenal mental content: to the transparent, sub-cognitive partition of thecurrently active PSM depicting the speaker as subject. If this partition is lostor becomes opaque, then speakers will stop using <I*>.

As we have seen, there is a naturalist version of Descartes’ argument con-cerning the certainty of his own existence: as long as a coherent self-model isactive and globally available, we have to assume that some kind of physicalsystem exists which generates the model. This is, from a third-person perspec-tive, simply “built in” as a central theoretical background assumption of natu-ralism. Interestingly, there is another “built in”-aspect concerning the functionalarchitecture underlying the phenomenology of selfhood itself: From a first-


person perspective, this assumption has been hardwired into our brains by thesheer transparency of the PSM. It is the “existential quantifier” which, as itwere, is displayed in a non-propositional format by the most invariant and fullytransparent partitions of the self-model. It results from the attentional unavail-ability of earlier processing stages and is reflected in the naïve realism of non-cognitive self-consciousness.

Please remember that one of the central claims guiding the current investi-gation is that no such things as selves exist in the world. All that, in an ontologi-cal sense, does exist are certain classes of information-processing systemsoperating under transparent self-models. Therefore all selves are either hal-lucinated (phenomenologically), or elements of inaccurate, reificatory phen-omenological descriptions. What is untranscendable, in standard situations (foran exception, see the Cotard delusion sketched above), is the experience ofactually existing right now. However, all other content properties realized bythe currently active self-model may in principle be entirely phenomenal prop-erties and may not carry any information about the real status of the system.

Conclusion

One interesting consequence of this short representationalist analysis of strongfirst-person phenomena in Baker’s sense is that the phenomenal first-personperspective is a condition of possibility for the emergence of a cognitive first-person perspective. Cognitive self-reference always is reference to the phenom-enal content of a transparent self-model. More precisely, it is a second-ordervariant of phenomenal self-modeling, which, however, is mediated by one andthe same integrated vehicle of representation. The capacity to mentally con-ceive of oneself as oneself* consists in being able to activate a dynamic, “hy-brid” self-model: Phenomenally opaque, quasi-symbolic and second-orderrepresentations of a pre-existing phenomenally transparent self-model arebeing activated and continuously re-embedded into it. This process is the pro-cess of introspection

4, as introduced in DEF 6.

What, then, can be said about the concept of a conscious cognitive subject?The transparent, sub-symbolic self-model is a tool used by an information-processing system to make a certain type of information cognitively available.It does so by enabling opaque mental self-simulations. Such self-simulationscan be phenomenally expressed as conscious I*-thoughts or, linguistically, asI*-sentences, i.e., they can lead to the cognitive or linguistic self-attributionof a prereflexive first-person perspective. In the act of self-attribution, a phe-nomenal first-person perspective is transformed into a cognitive first-person


perspective. Cognitive self-reference is a process of phenomenally modelingcertain aspects of the content of a pre-existing transparent self-model, which,from a third-person perspective, can be interpreted as the capacity of conceiv-ing of oneself as oneself*.

Baker is perfectly right in pointing out that the postulation of a mentalsymbol functionally playing the role of <I*> only renames the problem with-out solving it. She thinks what is truly necessary is the possession of a par-ticular capacity, the capacity of conceiving of oneself as oneself*, a capacitybeyond having self-localizing beliefs à la Perry (1979). The capacity she islooking for is the capacity of operating under a phenomenally transparent self-model that is globally available, not only for self-directed attention and selec-tive action control, but for mental concept formation as well. It is the capacityto apply introspection

4 to a pre-existing transparent self-model in the brain.

In human beings, natural evolution on our planet has already brought aboutthis capacity – as can be seen from the writings of René Descartes. Cognitiveagency, conceptually mediated self-consciousness, and strong first-personphenomena in Baker’s sense do not constitute an insurmountable obstacle fornaturalism. It is possible to accommodate these phenomena in an empiricallyplausible theory of mental representation. The deeper philosophical problem,however, seems to be that the purported, tacitly assumed epistemic content ofcognitive self-reference can now, in principle, always be reduced to a purelyphenomenal kind of mental content. Cognitive self-consciousness and the formof self-reference discussed here need an independent form of epistemic justi-fication.

Let me point out one important caveat. As Baker points out, any robustreductive naturalism has to offer an analysis for the constitutive conditions ofthe above-mentioned capacity in non-intentional and non-semantic concepts.I have offered an analysis in terms of phenomenal concepts. However, thedistinction between intentional and phenomenal content is not exclusive.Phenomenal content can be analyzed as a particularly strong kind of inten-tional content, one that satisfies certain additional constraints (I have tried todevelop a set of such constraints in Metzinger 2003, chapter 3).19 As such it isopen to naturalization: If it is true that phenomenal content locally supervenes,then there may be a nomologically coextensive physical property for everyphenomenal property, i.e. empirical research programs might discover domain-specific identities for every individual form of phenomenal content. Species-specific, local reduction becomes a possibility. This would also be the casefor the phenomenal content of cognitive self-reference.

On the other hand, if, as many philosophers today (including the author ofthe present paper) believe, phenomenal content just is a special variant of


intentional content, the model here sketched will eventually have to be sup-plemented by a full-blown naturalist theory of intentional content, a theorythat additionally allows us to understand the deeper relation between phenom-enal and intentional content. We are far from having such a theory. Today, itseems safe to say only two things in this regard: Phenomenal content as suchis not epistemically justified content, and in human beings it locally super-venes on brain properties. In principle, the distinction between intentional andphenomenal content collapses for self-consciousness: A brain in a vat couldgenerate the conscious experience of thinking [I am certain that I* exist].Cotard patients can generate the conscious experience of thinking [I am cer-tain that I* do not exist]. It is therefore plausible to assume that a minimallysufficient neural correlate for the phenomenal content characterizing each ofthese two specific conscious experiences does exist. The phenomenal contentwill supervene on this physical correlate, being fully determined by it in everysingle case.

The phenomenal experience of certainty is not the same as certainty. Butwill there not be many cases in which the phenomenal content of cognitiveself-reference is reliably correlated with intentional, epistemically justifiablecontent? It is interesting to note how even under naturalistic background-as-sumptions some kind of representational system has to exist, because trans-parent self-models are conceptually introduced as being realized by physicalvehicles. If this background assumption possesses an independent epistemicjustification, every conscious system is clearly justified in the I*-thought thatsome kind of physical and representational system exists, which carries outthe present act of mental self-reference. What is not justified is the claim thatthis system has to be a self.

Notes

1. The author wishes to thank Ulrike Haas-Spohn, Albert Newen, Bettina Walde, and twoanonymous reviewers for their constructive criticism.

2. § 275. Schau auf das Blau des Himmels, und sag zu Dir selbst “Wie blau der Himmelist!” – Wenn Du es spontan tust – nicht mit philosophischen Absichten – so kommt esDir nicht in den Sinn, dieser Farbeneindruck gehöre nur dir. Und du hast kein Bedenken,diesen Ausruf an einen Andern zu richten. Und wenn Du bei den Worten auf etwas zeigst,so ist es der Himmel. Ich meine: Du hast nicht das Gefühl des In-dich-selber-Zeigens,das oft das “Benennen der Empfindung” begleitet, wenn man über die “private Sprache”nachdenkt. Du denkst auch nicht, du solltest eigentlich nicht mit der Hand, sondernnur mit der Aufmerksamkeit auf die Farbe zeigen. (Überlege, was es heißt, “Mit derAufmerksamkeit auf etwas zeigen”.) (Wittgenstein 1952, English version 1953). – Lookat the blue of the sky and say to yourself “How blue the sky is!” – When you do it spon-


taneously – without philosophical intentions – the idea never crosses your mind that thisimpression of colour belongs only to you. And you have no hesitation in exclaiming thatto someone else. And if you point at anything as you say the words you point at the sky.I am saying: you have not the feeling of pointing-into-yourself, which often accompa-nies “naming the sensation” when one is thinking about “private language”. Nor do youthink that really you ought not to point to the colour with your hand, but with your at-tention. (Consider what it means “to point to something with the attention”.) Quoted afterWittgenstein 1958.

3. Van Gulick frequently speaks about “subjectively experienced transparency” as “semantictransparency”, but in most cases it is quite clear that he refers to the phenomenologicalaspect. For instance, he writes: “How can the functionalist account for subjectively ex-perienced transparency? Possessing information always involves a capacity for appro-priate behavior. Being informed about the content of a representation is being able torelate it to other representations and items in the world appropriate to its content. As longas understanding or being informed is analyzed as a behavioral capacity, even if therelevant behavior is all internal, the functionalist can hope to fit it within his theory.Thus the functionalist should resist any view of phenomenal transparency as a formof nonbehavioral self-luminous understanding. He can undercut the intuitive appeal ofthat view, by explaining the subjective experience of understanding in terms of smoothand seemingly automatic transitions. The internal component of understanding need in-volve nothing beyond an ability to interrelate many diverse representations with greatspeed. [. . .] How this is done is not something to which I have linguistic or introspec-tive access, but there must be powerful processors to produce these seemingly instanta-neous transitions.” (van Gulick 1988, p. 178p)

4. “[. . .] The only thing that seems to answer the description “attending introspectively toone’s visual experience” is attending to how things appear to one visually; and offhandthis seem to tell one what the representational content of one’s experience is withouttelling one anything about what the nonintentional features of one’s experience are thatencode this content. One may be inclined to say that one is revelling in the qualitative orphenomenal character of one’s experience when one “drinks in” the blue of a summersky or the red of a ripe tomato. But neither the blue nor the red is an object of introspec-tive awareness; these are experienced, perceptually rather than introspectively, as locatedoutside one, in the sky or in the tomato, not as features of one’s experience. G.E. Mooreonce complained that the sensation of blue is “as if it were diaphanous”; if one tries tointrospect it one sees right through it, and sees only the blue. In a similar vein one mightsay that qualia, if there are such, are diaphanous; if one tries to attend to them, all onefinds is the representative content of the experience.” (Shoemaker 1990, 1996, 100f)

5. “[. . .] in the case of her visual experience of a tree, I want to say she is not aware of, asit were, the mental paint by virtue of which her experience is ad experience of seeing atree. She is aware only of the intentional or relational features of her experience, not ofits intrinsic non-intentional features.[. . .] When you see a tree, you do not experienceany features as intrinsic features of your experience. Look at a tree and try to turn yourattention to intrinsic features of your visual experience. I predict you will find that theonly features there to turn your attention to will be features of the presented tree, in-cluding relational features of the tree ‘from here’”. (Harman 1990, 1997, p. 667)

6. “Generalizing, introspection of your perceptual experiences seems to reveal only aspectsof what you experience, further aspects of the scenes, as represented. Why? The answer,


I suggest, is that your perceptual experiences have no introspectible features over andabove those implicated in their intentional contents. So the phenomenal character of suchexperiences – itself is something that is introspectively accessible, assuming the appro-priate concepts are possessed and there is no cognitive malfunction – is identical with,or contained within, their intentional contents.” (Tye 1995, p. 136; see also p. 30p, 220p;for further uses see Tye 1991, p. 119, 1998, p. 468p, 2000, chapter 3.

7. I do not have space here to explicitly argue for the representationalist approach in thetheory of consciousness. Some of my background assumptions are sketched in Metzinger2003, p. 111ff.

8. For every kind of phenomenal content in humans there will be at least one minimallysufficient neural correlate (see Chalmers 2000).

9. Here is how Lynne Baker characterizes weak first-person phenomena: “Two points shouldbe noted about weak first-person phenomena: (i) They are exhibited by sentient organ-isms, who solve problems by means of perspectival attitudes; these attitudes then ex-plain the problem-solving behavior. (This point is independent of any theory of how, orwhether, attitudes are explicitly represented in the brain.) (ii) No first-person concept isneeded to bind belief-desire-behavior to a single organism. (. . .) Although such an ani-mal has beliefs and desires, he has no conception of belief of desire, nor of himself asthe subject or bearer of beliefs and desires.” (Baker 1998, p. 328).

10. “An ability to conceive of oneself as oneself* in the sense just described is both neces-sary and sufficient for the strong grade of first-person phenomena: An individual who isthe locus of strong first-person phenomena can conceive of herself as a bearer of first-person thoughts. She manifests an ability not only to make first-person reference, butalso to attribute to herself first-person reference.” (Baker 1998, p. 331).

11. The transparency of the self-model is a special form of inner darkness. It consists in thefact that the representational character of the contents of self-consciousness is not ac-cessible to subjective experience. This statement refers to all the different sensorymodalities, constituting the internal sources of information out of which the PSM isgenerated: Visceral sensitivity, kinesthetic and proprioceptive, tactile and haptic or evenvestibular stimuli. However, it is also true of self-directed visual, auditory, tactile, ol-factory or gustatory experience, of self-perception through sources of information. Inparticular, it is true of the integrated phenomenal model of the organism as a whole: Theinstrument of self-representation is not globally available as such and this is the reasonwhy the experiencing system, by necessity, becomes entangled in a naive realism withregard to the contents of its own mental self-representation. For many forms of self-rep-resentation, earlier processing stages are not attentionally available. Through a contin-gent property of its representational architecture, it simply has to experience itself asbeing in direct contact with the contents of its self-consciousness. The fact, which is in-accessible to the system using its on-board representational resources, is that its self-experience takes place within a medium. Completely transparent self-representation ischaracterized by the fact that the mechanisms which have led to its activation and theadditional fact, that a concrete internal state exists, which functions as the carrier of theircontent, cannot be recognized anymore. Therefore, the phenomenology of transparentself-modeling is the phenomenology of selfhood. It is the phenomenology of a systemcaught in a naive-realistic self-misunderstanding.

Phenomenal selfhood results from autoepistemic closure, it is a lack of information.The prereflexive, preattentive experience of being someone directly results from the con-


tents of the currently active self-model being transparent. Any system acting under a trans-parent self-model will, if all other necessary conditions for the emergence of phenomenalexperience in the domain constituted by this class of systems are realized, by necessityexperience itself as being in direct and immediate contact with itself. The phenomenalproperty of selfhood is constituted by transparent, non-epistemic self-representation, andit is on this level of representationalist analysis that the refutation of the correspondingphenomenological fallacy becomes truly radical, because it has a straightforward ontologi-cal interpretation: no such things as selves exist in the world. Under a general principle ofontological parsimony it is not necessary (or rational) to assume the existence of selves,because as theoretical entities they fulfill no indispensable explanatory function. Whatexists, are information-processing systems engaged in the transparent process of phenom-enal self-modeling. All that can be explained by the phenomenological notion of a “self”can also be explained using the representationalist notion of a transparent self-model.

It is important to note that, in humans, there are sections of the PSM, which are al-ways opaque. Whereas higher-order forms of self-modeling like self-directed attentionmay still be fully caught in a naive realistic self-misunderstanding – you simply “lookat” or “feel into” that part of yourself, with which you believe to be in direct contactanyway – self-directed cognition is typically characterized by the phenomenology ofinsight. You simply seem to know that all this is an ongoing process of self-representa-tion, while at the same time you are certain that you are the initiator of this processyourself. This, of course, is Descartes’ classical intuition, and the interesting question isto what extent it is simply another version of the phenomenological fallacy – the unjus-tified use of an existential quantifier within a psychological operator.

12. “Mode of presentation” is here used in a standard epistemological/semantic way, as theway in which a fact can be given. The idea is that the same kind of fact, say, about a cer-tain brain state holding, can simultaneously be presented and known from the outside, viathird-person, neuroscientific theory formation, and from the inside, “under a first-personperspective” – which one may then call an indexical EGO-mode of presentation.

13. Let me illustrate this point. Baker convincingly criticizes Flanagan (1992, 194 pp.) bypointing out that one has to analyze “thinking about one’s own model of one’s own self”as thinking about one’s model of oneself as oneself and not just as thinking of a modelof someone-who-is-in-fact-oneself (Baker 1998, p. 342). The current proposal achievesthis by introducing the transparency-constraint.

14. For more on goal-representation, see Metzinger and Gallese 2003.15. This certainty can be lost for contingent reasons, for instance in patients suffering from

the Cotard delusion. It may be interesting to note that this form of phenomenal contentis not a necessary feature of all conscious experience: There is phenomenal conscious-ness in combination with the absolutely certain belief of already being dead, or of evennot having existed at all. Some of the patients just mentioned will even stop using thepronoun “I”.

16. In using the abbreviation “DM” instead of “DMS” (for “delusional misidentificationsyndromes”) I follow a recent terminological modification introduced by Nora Breenand colleagues in order to avoid the inaccuracy of assuming a whole cluster of symp-toms as defining characteristics for the different categories (Cf. Breen, Caine, Coltheart,Hendy and Roberts 2000, p. 75, n. 1).

17. Young and Leafhead (1996, p. 154) argue that there is no specific symptom shown byevery one of Cotard’s pure cases, whereas Berrios and Luque (1995) offer a statistical


analysis of historical clinical usage of the term, concluding that a pure Cotard syndrome(represented by “Cotard type 1” patients) does exist, and that its nosological origin “isin the delusional and not in the affective disorders;. . .” (p. 187). I will not take a posi-tion on this issue here.

18. Enoch and Trethowan refer to Anderson 1968, p. 136 when making this claim. How-ever, the quoted example cannot be found at this location in Anderson’s textbook.

19. I want to thank one anonymous reviewer of this journal for his particularly helpful criti-cism at this point.

References

Ahleid, A. 1968. Considerazioni sull’esperienza nichilistica e sulla syndrome die Cotard nellepsicosi organiche e sintomatiche. Il Lavoro neuropsichiatrico 43: 927–945.

Anderson, E. W. 1964. Psychiatry. London: Baillière, Tindall and Cox Ltd.Anton, G. 1898. Ueber die Herderkrankungen des Gehirns, welche vom Patienten selbst nicht

wahrgenommen werden. Wiener klinische Wochenschrift 11 (March 10): 227–229.Anton, G. 1899. Über die Selbstwahrnehmungen der Herderkrankungen des Gehirns durch

den Kranken bei Rindenblindheit und Rindentaubheit. Archiv für Psychiatrie undNervenkrankheiten 32: 86–127.

Baars, B. J. 1988. A Cognitive Theory of Consciousness. Cambridge: Cambridge UniversityPress.

Baker, L. 1998. The first-person perspective: A test for naturalism. American PhilosophicalQuarterly 3: 327–346.

Benson, D. F. and Greenberg, J. P. 1969. Visual form agnosia. Archives of Neurology 20: 82–89.

Berrios, G.A . and Luque, R. 1995. Cotard’s syndrome. Analysis of 100 cases. Acta Psy-chiatrica Scandinavia 91: 185–189.

Breen, N., Caine, D., Coltheart, M., Hendy, J., and Roberts, C. 2000. Towards an understandingof delusions of misidentification: Four case studies. In: M. Coltheart and M. Davies (eds),Pathologies of Belief. Oxford: Blackwell.

Bressloff, P. C., Cowan, J. D., Golubitsky, M., Thomas, P. J., and Wiener, M. C. 2001. Geo-metric visual hallucinations, Euclidean symmetry and the functional architecture of stri-ate cortex. Philosophical Transactions of the Royal Society London B 356: 299–330.

Campbell, J. 1999. Schizophrenia, the space of reasons, and thinking as a motor process. TheMonist 82: 609–625.

Castañeda, H.N. 1966. “He”: A study on the logic of self-consciousness. Ratio 8: 130–157.Chalmers, D. J. 1997. Availability: The cognitive basis of experience? In: N. Block, O.

Flanagan, and G. Güzeldere (eds), The Nature of Consciousness: Philosophical Debates.Cambridge, MA: MIT Press.

Chalmers, D. J. 2000. What is a neural correlate of consciousness? In: T. Metzinger (ed),Neural Correlates of Consciousness – Empirical and Conceptual Questions. Cambridge,MA: MIT Press.

Churchland, P. M. 1998. Conceptual similarity across sensory and neural diversity: The Fodor/Lepore challenge answered. Journal of Philosophy 65: 5–32.

Churchland, P. M. (unpublished manuscript). Neurosemantics: On the Mapping of Minds andthe Portrayal of Worlds.


Clark, A. 1997. Being There – Putting Brain, Body, and World Together Again. Cambridge,MA: MIT Press.

Cleeremans, A. (ed) 2003. The Unity of Consciousness: Binding, Integration, and Dissocia-tion. Oxford: Oxford University Press.

Coltheart, M. and Davies, M. (eds) 2000. Pathologies of Belief. Oxford and Malden, MA:Blackwell.

Cotard, J. 1880. Du délire hypocondriaque dans une form grave de la mélancolie anxieuse.Annales Médico-Psychologiques 38: 168–170.

Cotard, J. 1882. Du délire des négations. Archives de Neurologie 4: 152–170/282–295.Damasio, A. 1999. The Feeling of What Happens: Body and Emotion in the Making of Con-

sciousness. Harcourt Brace & Company.Dretske, F. 1969. Seeing and Knowing. Chicago: University of Chicago Press.Enoch, M. D. and Trethowan, W. H. 1991. Uncommon Psychiatric Syndromes. Oxford:

Butterworth-Heinemann.Förstl, H. Almeida, O. P., Owen, A., Burns, A., and Howard, R. 1991. Psychiatric, neurologi-

cal and medical aspects of misidentifications syndromes: a review of 260 cases. Psycho-logical Medicine 21: 905–910.

Frankfurt, H. 1971. Freedom of the will and the concept of a person. Journal of Philosophy68: 5–20.

Gazzaniga, M. (ed) 1995. The Cognitive Neurosciences. Cambridge, MA: MIT Press.Gerrans, P. 1999. Delusional misidentification as subpersonal disintegration. The Monist 82:

590–608.Gerrans, P. 2000. Refining the explanation of Cotard’s delusion. In: M. Coltheart and M.

Davies (eds), Pathologies of Belief. Oxford and Malden, MA: Blackwell.Güzeldere, G. 1995. Is consciousness the perception of what passes on one’s own mind? In:

T. Metzinger (ed), Conscious Experience. Thorverton, UK: Imprint Academic & Paderborn:mentis.

Halligan, P. W. and Marshall, J. C. (eds). 1996. Method in Madness: Case studies in Cogni-tive Neuropsychiatry. Hove, UK: Psychology Press.

Harman, G. 1990. The intrinsic quality of experience. In: J. Tomberlin (ed), PhilosophicalPerspectives, Vol. 4: Action Theory and Philosophy of Mind. Atascadero, CA: RidgeviewPublishing. Reprinted in and quoted after N. Block, O. Flanagan, and G. Güzeldere (eds),The Nature of Consciousness: Philosophical Debates. Cambridge, MA: MIT Press.

LaBerge, S. and Gackenbach, J. 2000. Lucid dreaming. In: E. Cardeña, S. J. Lynne and S.Krippner (eds), Varieties of Anomalous Experience: Examining the Scientific Evidence.Washington, D.C. American Psychological Association.

Leopold, D. A. and Logothetis, N. K. 1999. Multistable phenomena: Changing views in per-ception. Trends in Cognitive Sciences 3: 254–264.

Metzinger, T. 1999. Subjekt und Selbstmodell. Paderborn: mentis.Metzinger, T. 2000. The subjectivity of subjective experience: A representationalist analysis

of the first-person perspective. In: T. Metzinger (ed), Neural Correlates of Consciousness– Empirical and Conceptual Questions. Cambridge, MA: MIT Press.

Metzinger, T. 2003. Being No One. The Self-Model Theory of Subjectivity. Cambridge, MA:MIT Press.

Metzinger, T. 2004, forthcoming. The incoherent strawman. Invited commentary for Alva Noëand Evan Thompson: “Are there neural correlates of consciousness?” Journal of Conscious-ness Studies 11.


Metzinger, T. and Walde, B. 2000. Commentary on Jakab’s “Ineffability of Qualia.” Conscious-ness and Cognition 9: 352–362. doi:10.1006/ccog.2000.0463

Metzinger, T. and Gallese, V. (2003). The emergence of a shared action ontology: buildingblocks for a theory. In: G. Knoblich, B. Elsner, G. von Aschersleben, and T. Metzinger(eds), Grounding Selves in Action. Special issue of Consciousness and Cognition. (in press).

Moore, G. E. 1903. The refutation of idealism. Mind 12: 433–453.Newen, A. 1997. The logic of indexical thoughts and the metaphysics of the “self”. In: W.

Künne, A. Newen, and M. Anduschus (eds), Direct Reference, Indexicality and Prop-ositional Attitudes. Stanford: CSLI.

O’Brien, G. and Opie, J. 2001. Notes toward a structuralist theory of mental representation.In: H. Clapin, P. Staines, and P. Slezak (eds), Representation in Mind: New Approaches toMental Representation. Greenwood Publishers.

Perry, J. 1979. The problem of the essential indexical. Noûs 13: 3–22.Pöppel, E. 1988. Mindworks: Time and Conscious Experience. New York: Hartcourt Brace

Jovanovich.Posner, M. I. 1995. Attention in cognitive neuroscience: An overview. In: M. Gazzaniga (ed),

The Cognitive Neurosciences. Cambridge, MA: MIT Press.Raffman, D. 1995. On the persistence of phenomenology. In: T. Metzinger (ed), Conscious

Experience. Thorverton, UK: Imprint Academic & Paderborn: mentis.Séglas, J. 1897. Le Délire des Négations: Séméiologie et Diagnostic. Paris: Masson, Gauthier-

Villars.Shoemaker, S. 1968. Self-reference and self-awareness. Journal of Philosophy 65: 555–578.Shoemaker, S. 1990. Qualities and qualia: What’s in the mind? Philosophy and Phenomen-

ological Research 50 (Supplement): 109–131. Reprinted in S. Shoemaker 1996. The First-Person Perspective and Other Essays. Cambridge, UK: Cambridge University Press.

Siegel, R. K. and West, L. J. (eds) 1975. Hallucinations. Behavior, Experience, and Theory.New York: Wiley.

Tye, M. 1991. The Imagery Debate. Cambridge, MA: MIT Press.Tye, M. 1995. Ten Problems of Consciousness – A Representational Theory of the Phenom-

enal Mind. Cambridge, MA: MIT Press.Tye, M. 1998. Inverted Earth, Swampman, and representationism. In: J. Tomberlin (ed), Philo-

sophical Perspectives, Vol. 12: Language, Mind, and Ontology. Malden, MA/Oxford, UK:Blackwell.

Tye, M. 2000. Consciousness, Color, and Content. Cambridge, MA: MIT Press.Van Gulick, R. 1988. A functionalist plea for self-consciousness. The Philosophical Review

47 (2): 149–181.Wittgenstein, L. 1953. Philosophical Investigations, trans. G. E. M. Anscombe. London:

Macmillan.Wittgenstein, L. 1953; 1982. Philosophische Untersuchungen. Frankfurt am Main: Suhrkamp.Wittgenstein, L. 1958. Philosophical Investigations, trans. G. E. M. Anscombe. The English

Text of the Third Edition. Upper Saddle River, NJ: Prentice Hall.Yates, J. 1985. The content of awareness is a model of the world. Psychological Review 92:

249–284.Young, A. W. and Leafhead, K. M. 1996. Betwixt of life and death: Case studies of the Cotard

delusion. In: P. W. Halligan and J. C. Marshall (eds), Method in Madness: Case Studies inCognitive Neuropsychiatry. Hove, UK: Psychology Press.

Young, A. W. 1999. Delusions. The Monist 82: 571–590.


Documents

Phenomenal transparency and cognitive self-reference1 · PHENOMENAL TRANSPARENCY AND COGNITIVE© 2003 SELF-REFERENCE 353 Phenomenology and the Cognitive Sciences 2: 353–393, 2003