Publié le 03/12/2021

Extrait du document

Ci-dessous un extrait traitant le sujet : Consciousness. Ce document contient 10402 mots soit pages. Pour le télécharger en entier, envoyez-nous un de vos documents grâce à notre système gratuit d’échange de ressources numériques. Cette aide totalement rédigée en format PDF sera utile aux lycéens ou étudiants ayant un devoir à réaliser ou une leçon à approfondir en: Echange.

Philosophers have used the term ‘consciousness' for four main topics: knowledge in general, intentionality,introspection (and the knowledge it specifically generates) and phenomenal experience. This entry discusses thelast two uses. Something within one's mind is ‘introspectively conscious' just in case one introspects it (or ispoised to do so). Introspection is often thought to deliver one's primary knowledge of one's mental life. Anexperience or other mental entity is ‘phenomenally conscious' just in case there is ‘something it is like' for one tohave it. The clearest examples are: perceptual experiences, such as tastings and seeings; bodily-sensationalexperiences, such as those of pains, tickles and itches; imaginative experiences, such as those of one's own actionsor perceptions; and streams of thought, as in the experience of thinking ‘in words' or ‘in images'. Introspectionand phenomenality seem independent, or dissociable, although this is controversial.Phenomenally conscious experiences have been argued to be nonphysical, or at least inexplicable in the manner ofother physical entities. Several such arguments allege that phenomenal experience is ‘subjective'; thatunderstanding some experiences requires undergoing them (or their components). The claim is that any objectivephysical science would leave an ‘explanatory gap', failing to describe what it is like to have a particularexperience and failing to explain why there are phenomenal experiences at all. From this, some philosophers infer‘dualism' rather than ‘physicalism' about consciousness, concluding that some facts about consciousness are notwholly constituted by physical facts. This dualist conclusion threatens claims that phenomenal consciousness hascausal power, and that it is knowable in others and in oneself.In reaction, surprisingly much can be said in favour of ‘eliminativism' about phenomenal consciousness; thedenial of any realm of phenomenal objects and properties of experience. Most (but not all) philosophers deny thatthere are phenomenal objects - mental images with colour and shape, pain-objects that throb or burn, innerspeech with pitch and rhythm, and so on. Instead, experiences may simply seem to involve such objects. Thecentral disagreement concerns whether these experiences have phenomenal properties - ‘qualia'; particularaspects of what experiences are like for their bearers. Some philosophers deny that there are phenomenalproperties - especially if these are thought to be intrinsic, completely and immediately introspectible, ineffable,subjective or otherwise potentially difficult to explain on physicalist theories. More commonly, philosophersacknowledge qualia of experiences, either articulating less bold conceptions of qualia, or defending dualism aboutboldly conceived qualia.Introspective consciousness has seemed less puzzling than phenomenal consciousness. Most thinkers agree thatintrospection is far from complete about the mind and far from infallible. Perhaps the most familiar account ofintrospection is that, in addition to ‘outwardly perceiving' non-mental entities in one's environment and body, one‘inwardly perceives' one's mental entities, as when one seems to see visual images with one's ‘mind's eye'. Thisview faces several serious objections. Rival views of introspective consciousness fall into three categories,according to whether they treat introspective access (1) as epistemically looser or less direct than innerperception, (2) as tighter or more direct, or (3) as fundamentally non-epistemic or nonrepresentational. Theoriesin category (1) explain introspection as always retrospective, or as typically based on self-directed theoreticalinferences. Rivals from category (2) maintain that an introspectively conscious mental state reflexively representsitself, or treat introspection as involving no mechanism of access at all. Category (3) theories treat a mental stateas introspectively conscious if it is distinctively available for linguistic or rational processing, even if it is not itselfperceived or otherwise thought about.1 Pre-Cartesian uses of ‘consciousness'As elsewhere in philosophy, Descartes' writings mark a major shift in philosophical preoccupation withconsciousness. Pre-Cartesian philosophers of mind rarely emphasize the terms ‘conscious' or‘consciousness' (or clear equivalents). Post-Cartesian philosophers of mind rarely avoid such emphasis. Thissection compares Descartes' usage with earlier usage.Descartes typically speaks of being ‘conscious' to refer to an allegedly intimate source of knowledge about one'sown mental occurrences. In the ‘Conversation with Burman' he says that ‘to be conscious is both to think and toreflect on one's thought' (1648: 335), where the term ‘thought' extends widely, as in the ‘Second Replies', to‘everything that is within us in such a way that we are immediately conscious of it' including ‘all the operations ofthe will, the intellect, the imagination and the senses' (1641: 113). Descartes seems to treat everything mental asintrospectively conscious, in these passages and elsewhere ( the ‘Fourth Replies', 1641: 171). Elsewhere,however, Descartes seems to deny that introspection is complete, as in the Discourse on Method: ‘many people donot know what they believe, since believing something and knowing that one believes it are different acts ofthinking, and the one often occurs without the other'.The resulting focus on the scope and limits of introspective knowledge ( §§6-7) is apparently responsible forthe modern uses of ‘conscious' and ‘non-conscious' to mark a potential distinction between two kinds of mentalstates. Introspected states are (introspectively) conscious, while others, if any, are (introspectively) unconscious.Pre-Cartesian authors do not use the word ‘conscious' to mark such a distinction, although some may becommitted to the distinction, either implicitly or in other terms (Whyte 1962). To take a rather spectacularexample, Socrates claims in Plato's Meno that since one's soul ‘has been born many times' and ‘has learnedeverything that there is', ‘seeking and learning are in fact nothing but recollection' (81c-d). This seems to requirethat one has latent knowledge of which one can at best become aware with great difficulty ( Plato §11).Many pre-Cartesian writers share commitment to a special fountain of reflective knowledge, often called ‘innersense'. In Summa theologiae Aquinas posits a ‘common sense' which enables one, for example, to ‘tell white fromsweet', and adds that this common sense ‘is also able to sense sensation itself, as when somebody sees that he isseeing' (I.78.4). This responds to Aristotle's apparent claim, in On the Soul, that it is ‘by sight that one perceivesthat one sees' rather than by another sense (III 2, 425b12). On these views, the subject matter of such innerawareness is more restricted than for Descartes, including sensation but perhaps not other mental processes. It mayeven be that, for Aristotle, the relevant ‘inner' perception - by which one sees that one sees - is directed at one'sexternal sense organs themselves and not at anything Descartes would consider strictly ‘mental'.A nearer equivalent to ‘introspective consciousness' is the Sanskrit term ‘manas', used widely in Hindu texts for a‘mind-organ' that functions like the external sense organs. For instance, the Vai?e????ika S?tra (c. 3rd century BC)claims that ‘Intellect, pleasure, pain, desire, aversion, effort are perceptible by the internal organ'. This inner perception is carefully distinguished from ‘inferential knowledge', and is framedso as not to involve introspection of a ‘self'. Indeed, while Descartes holds that he can introspect himself as a‘thinking thing', Hinduism characteristically claims that introspection reveals no ‘true self' distinct from one'smental states, and Buddhism characteristically denies that there are ‘thinking things' at all.The word ‘conscious' derives from the Latin words ‘cum' (‘together with') and ‘scire' (‘knowing'). In the originalsense, two people who know something together are said to be conscious of it ‘to one another', with the irresistibleconnotation that they are privy to a scandalous secret. By extension, one can be conscious ‘to oneself' of secretshames - whence the original use of ‘consciousness' for conscience, the inner accuser silently sharing knowledgeof one's transgressions. This archaic moral sense of ‘consciousness' is not a concern in this entry.The Latin conjunction of ‘cum' and ‘scire' also has a use in which the prefix is merely emphatic, so that being‘conscious of' something simply means knowing it, or knowing it well. In this sense the word ‘conscious' can alsobe used as an adjective: a ‘knowing' being such as a normal person is a conscious being, while an ‘unknowing'being such as a plant or sleeping person is an unconscious being. ‘Conscious', like ‘knowing', can be used in thisway for things with minds but not for things within minds, such as mental states. People and animals know things -are conscious of things - but mental states do not themselves know things. Thus, for example, Aquinas uses‘conscious' to describe bearers of mental states - such as human beings, animals and God - but not to describemental states - not even ‘seen' seeings. Since the main philosophical problems about consciousness concern themore modern distinction between conscious and unconscious states, this entry focuses neither on the distinctionbetween conscious and unconscious subjects, nor on the broad ‘knowing' sense of ‘conscious' ( Knowledge,concept of). In effect, Descartes refashions the ‘knowing well' sense of ‘conscious', regimenting it for a particularsource of knowing, introspection. Issues about the specific epistemological status of introspection are a centralconcern of this entry.2 Post-Cartesian uses of ‘consciousness'Descartes' use of ‘consciousness' for reflective knowledge spread rapidly through the next generation of Europeanphilosophers. In An Essay concerning Human Understanding Locke writes, as he does of ‘reflection', that‘Consciousness is the perception of what passes in a Man's own mind' (1689: II.i.19); Leibniz's term for this is‘apperception'. In his Critique of Pure Reason Kant goes on to distinguish between ‘empirical apperception' of a‘flux of inner appearances' - mentioning that ‘Such consciousness is usually named inner sense' - and‘transcendental apperception' which is alleged to be a ‘pure original unchangeable consciousness' that reveals a‘fixed and abiding self' (A107; see Kant, I. §6). This entry discusses one's introspective access to the ‘flux' ofparticular events within one's mind, rather than substantive introspective knowledge about one's self - aboutwhether one is made of physical or spiritual components, and whether one persists through time.In the broader ‘knowing' sense ( §1), a creature is conscious of something just in case it knows something,independently of whether this knowing is itself introspectible. This ancient sense of ‘conscious' lingers on, andbroadens to cover any kind of belief or cognition (whether or not it is ‘knowledge'), and any kind of attitude aboutsomething (whether or not it is ‘cognitive'). In this sense, a creature has consciousness if it has any kind of‘intentional' mental state. By extension, the state itself can be said to be a state of consciousness, even if it is notintrospectible. (This is distinct from the widespread claim that all conscious states are intentional - that ‘allconsciousness…is consciousness of something' (Sartre 1943: 11). On this broad sense of ‘conscious', bydefinition, all ‘ofness' is conscious ofness.) As twentieth-century philosophers of mind and language most oftenpursue concerns about intentionality using terms other than ‘consciousness', intentionality will not be explored inthis entry.In a still broader sense, ‘mind' and ‘consciousness' are synonyms, as are ‘being mindful of' something and ‘beingconscious of' it, so that any kind of mental state (whether or not it is an ‘attitude') is a state of consciousness.When Hegel (§5), Marx or Lukács speak of ‘unhappy', ‘false' or ‘class' consciousness, or when political activistsattempt to ‘raise' consciousness, their concerns are usually equally well rendered using a general term such as‘knowledge', ‘thinking', ‘attitudes' or ‘mentality' in place of ‘consciousness'. It is not clear that their concern iswith introspection, since they refer mainly to thoughts about (or seemingly about) non-mental things,independently of whether these thoughts are themselves introspectively conscious. (When Hegel refers to thoughtsexplicitly about the mind, he uses ‘self-consciousness' rather than ‘consciousness'.) Likewise, many scientificwritings officially on ‘consciousness' are about mentation and mentation-like activity in general, avoiding anyquestion of whether the activity is introspectively conscious. Since these broad uses of ‘consciousness' seem tointroduce no distinctive philosophical perplexities, this entry puts them aside.With the dawn of scientific psychology in the late nineteenth century, the central philosophical controversies aboutconsciousness centre around whether consciousness can ever be explained by an objective science of the mind.The biologist Thomas Huxley provides an early attempt to express the sense of mystery: ‘How it is that anythingso remarkable as a state of consciousness comes about as a result of irritating nervous tissue, is just asunaccountable as the appearance of Djin when Aladdin rubbed his lamp' (1866: 210). Since that time there havebeen many scientific advances in understanding the mechanisms of perception, thought and communication, andmany philosophical advances in understanding the nature of intentionality and meaning. According to ThomasNagel and many other philosophers, such advances must leave an unexplained residue, concerning what it is like tohave phenomenally conscious experiences (as illustrated in the introduction; see Nagel, T. §4). Nagel writes that‘Consciousness is what makes the mind-body problem really intractable', identifying ‘subjectivity' as its mosttroublesome feature:Fundamentally an organism has conscious mental states if and only if there is something it is like to be thatorganism - something it is like for the organism. We may call this the subjective character ofexperience…every subjective phenomenon is essentially connected with a single point of view, and it seemsinevitable that an objective, physical theory will abandon that point of view.(1974: 166-7)Until quite recently, the agenda for the philosophy of mind has been set more by epistemology than by the sciencesof mind. Consequently, introspective consciousness has been the most central and important philosophical notionof consciousness. The remainder of the entry is divided between the explanation of phenomenal consciousness andthe epistemology of introspective consciousness.3 Subjectivity and the phenomenalNagel argues that physical theories cannot explain one's phenomenal consciousness, because they abandon one'spoint of view ( §2). Science does indeed abandon one's point of view, in so far as one need not be able tounderstand or defend the theory (if one lacks relevant concepts or evidence), and in so far as subjects with otherpoints of view should be able to understand and defend the theory. But does abandoning a point of view preventdescribing the point of view, or the features viewed from that point? Even if Nagel is correct that one's point ofview and one's phenomenal experience are ‘essentially' connected, they may not be exclusively connected. Anobjective physical theory of phenomenal consciousness would presumably allow that phenomenal features areaccessible from multiple points of view - for instance, both by some form of introspection and by some form ofneurophysiological or psychological observation and theory. Furthermore, if experiences are necessarilyintrospectible or introspected ( §6) this might explain Nagel's claim that phenomenal features are‘essentially connected with a single point of view'.Frank Jackson (1982) amplifies Nagel's challenge in his ‘knowledge argument' against ‘physicalism', the thesisthat the world is wholly physical. He imagines a super-scientist, Mary, who has never seen anything colouredbecause she lives her life in a black-and-white room. From a black-and-white television in this room, she learns allthe objectively specifiable physical (and causal or ‘functional') facts in the world. When she finally leaves theroom and first sees colour, Jackson argues, she learns a new fact about the nature of phenomenal experience; shemight exclaim, ‘Oh! It is like this to see red!' The new fact Mary learns cannot be identical to a physical orfunctional fact, or else it would be among the facts Mary already knows before leaving the room. So no whollyphysicalist account of phenomenal facts can be true.Most responses to Jackson's argument involve denying that Mary learns a new fact upon experiencing red. Onsome views, she learns how to do new things - to imagine experiencing redness or to recognize redness visually -without coming to know that any new fact obtains. On others, she learns that an old physical fact about experienceobtains, but comes to know it in a new way - via introspective access or via new concepts. For example, oneproposal (Lycan 1990) compares the relevant introspective ways of representing one's experiences with simpleinner-perceptual demonstratives, ways of knowing colour experience that would be unavailable to Mary fromwithin the black-and-white room. Consider an analogy: when one perceives a banana and thinks of itdemonstratively - as ‘this' - those who do not perceive the banana cannot think of it in the same way - simply as‘this' (while staring at something else). Likewise, if one's introspections of one's experiences involve somedemonstratives of them, this would explain why these representations cannot strictly be shared by someone whodoes not ‘perceive' the same experiences but merely thinks about them.Even among philosophers who accept that phenomenal consciousness is some physical or functional process, thereare doubts about the possibility of explaining it. Colin McGinn (1991) suggests that human beings are foreverblocked from knowing the ‘link' between the brain and consciousness, roughly because introspectiveconsciousness gives no knowledge of brains, while neuroscientific access to brains gives no access toconsciousness. Critics respond that one might learn the explanatory link by theoretical inference from jointintrospective and scientific data, such as correlations between phenomenal features and brain states ( Flanagan1992: ch. 6).In slightly different ways, Jackson, Joseph Levine (1993) and David Chalmers (1996) argue that even if one canknow what this ‘link' in fact is, explanations based upon it cannot be as satisfying as other scientific explanations.In essence, the argument is that for any objective, scientific account of phenomenal consciousness, one canconceive of a creature that meets the conditions in the account but lacks phenomenal consciousness. In the extremecase, it is said, one can conceive of a world that is an exact physical duplicate of the actual world - complete withduplicate stars, planets, rocks, plants, animals and philosophers - but which lacks any phenomenal consciousness.All the human-like beings in that world would be non-phenomenal ‘zombies'. So the prescientific concept ofphenomenal consciousness is not such that scientific premises could necessitate, a priori, conclusions about thephenomenal. In Levine's terms, there is an ‘explanatory gap' between physical reality and phenomenalconsciousness. By contrast, for example, it is claimed that the prescientific concept of ‘water' is such that scientificpremises about H2O, can necessitate, a priori, conclusions about water. In particular, it is held to be part of ourconcept of water that if there is anything roughly unique in the lakes and rivers around us that has a preponderanceof features such as boiling, eroding rocks, quenching thirst and so on, then it is water. If science establishes thatH2O meets this condition, then there is no further conceptual possibility that H2O is not water.At least three lines of response may be advanced against the explanatory gap argument. One concedes that noscientific premises a priori necessitate conclusions about the phenomenal, but insists that the same is true forscientific explanations of water, and so on. Perhaps someone who believes in water but denies that it boils, is inlakes and so on is making a false and bizarre claim but are not strictly contradicting themselves. The fate of thisresponse presumably depends on the fate of general reservations about a priori or conceptual necessity.The second strategy also concedes the lack of a priori necessity, but attributes it to the idiosyncratic ways in whichphenomenal facts are represented - for example, by introspection-based demonstratives ( Tye 1995: 178-81;Demonstratives and indexicals). By comparison, suppose that agent A holds banana B in front of his face, andcomes to accept the demonstrative ‘This is a banana'. This conclusion cannot strictly be deduced from anydemonstrative-free descriptions of the banana - that B is a banana in A's hand, that A sees B, and so on. Thesubject could believe all of these premises, and still conceive that this is not a banana - perhaps by conceiving thathe is not in fact agent A. This response to the explanatory gap seems at best to apply only to particular phenomenalconclusions, of the form: it is like this to have a given experience. It is an incomplete response, since there arephenomenal conclusions without demonstratives, namely, those of the form: there is something it is like to have agiven experience.The third strategy is to deny that there is an unbridgeable conceptual gap; in effect, to deny that zombies(non-phenomenal physical duplicates of phenomenal creatures) are even conceptually possible. In Nagel-likefashion, Robert Kirk analyses the concept of phenomenality as follows: a creature's phenomenal states are thosewith different ‘characters' that are ‘for the creature as a whole' (1994: ch. 5). He explains this in terms of neuralstates with different ‘patterns of activation' that are ‘directly available' to ‘have different effects on processes ofassessment, decision-making, and action initiation'. It follows that if a creature has states with characters for thecreature, any physical duplicate of the creature does also. The problem is that Kirk's ‘having character' does notexpress the same concept as Nagel's ‘being like something'. Perhaps Kirk demonstrates how states can be ‘for acreature', but he does not demonstrate that they are ‘like something' for the creature, that is, phenomenal, inNagel's full-blown sense. For example, beliefs may also be realized in different patterns of activation with therelevant direct availability, and so in that sense beliefs may be ‘for' a creature, but beliefs are not clearlythemselves ‘like something' for a creature ( §6). Further analysis of the ‘like something' idiom would be neededto complete this line of response to the explanatory gap.4 Dualism and the phenomenalThe attraction of subjectivity-based arguments against physicalism makes it important to understand the potentialramifications of various non-physicalist alternatives. The central metaphysical issues concern ‘mental causation'and ‘emergence' ( Mental causation). The two central epistemological ramifications concern knowledge ofother minds and knowledge of one's own mind ( Other Minds).Physical science promises an explanation of physical events wholly in terms of other physical events. For example,unsupported objects fall, not because they want to fall, but because of gravitational forces. Similarly, a large groupof neurons cause another neuron to fire, not because of its embarrassment due to peer pressure or its fear of a mob,but because of electrochemical forces. In this way, there should be a purely physical explanation of the activities ofbrains in producing (reflex or non-reflex) behaviour. Yet phenomenal consciousness also seems to have physicalconsequences on behaviour. When one feels an itch, one scratches; this seems to be in large part because of what itis like to itch. But how can an itchy feel make a difference, if there is a purely physical explanation of one's handmotions? The physicalist may make room for phenomenal causation by maintaining that the feel of an itch iswholly constituted by (or, perhaps, identical with) features of the brain and body ( Mind, identity theory of).The feel causes the hand motion, because it is made of things that do so, just as the brain causes the hand motionby being composed of things that do so (that is, neurons). But this strategy is unavailable to the non-physicalistwho thinks there are mental objects, properties or states that are not wholly composed of physical entities (Dualism). One possibility for the dualist is to maintain that some physical events are inexplicable purely inphysical terms - this is Descartes' own position about non-reflex behaviour. Another prominent strategy is todefend ‘epiphenomenalism' about the phenomenal, or the idea that, contrary to appearances, phenomenal featuresare irrelevant to the generation of physical events ( Jackson 1982; Chalmers 1996: ch. 4; Epiphenominalism).If phenomenal states and features are not wholly constituted by physical states and features, how are theyconstituted? One idea, prevalent among neuroscientists, is that the phenomenal ‘emerges' from physicalinteractions. By analogy, water has properties that are not explicable by the properties of hydrogen and oxygenseparately. Water dissolves salt, but neither hydrogen nor oxygen does so separately (nor does hydrogen dissolvepart of salt and oxygen the rest). Analogously, the whole of a phenomenal experience is taken to be more than thesum of its physical parts (that is, the physical entities that help realize it). The initial plausibility of such analogiesseems to depend on a too-narrow conception of the ‘parts' of a complex entity. A water molecule does havehydrogen and oxygen atoms as parts, but it also has various relations among these atoms as other ‘parts'. Thephysical relation of bonding, for instance, is as necessary for water as the atoms are. If all of the parts of water arecounted, it is not clear that water has properties inexplicable by the properties of these parts. This kind of‘emergence' conclusion - one that forgets to ‘sum' some of the parts of a whole - is compatible with a whollyphysicalist account of water, and of consciousness. In the absence of better analogies, non-physicalist philosophershave more commonly denied that there is genuine emergence, emergence of new features that are inexplicable interms of combinations of other features. This leads some non-physicalists to the startling ‘panpsychist' claim thatmental ‘ingredients' of phenomenal consciousness must be present in the tiniest bits of matter capable ofcomprising brains, and to the claim that phenomenal consciousness is a fundamental part of nature - perhaps alongwith subatomic mass and charge.Epiphenomenalism and panpsychism exacerbate the epistemological problem of other minds, but in oppositedirections. It seems that we have excellent, even if not quite perfect, justification for believing that other peoplehave experiences. Plausibly, talk of experiences is not simply disguised talk of others' observable behaviour, but isinstead to be inferred from behaviour with some degree of theoretical risk. Perhaps the inference is in partgrounded on drawing an analogy between oneself and others, but inference from a single case gives one at bestmeagre justification for conclusions about others. The residual problem of other minds, then, is to explain how oneattains an appropriate level of justification. The physicalist can maintain that one is justified in part because thephenomenal features of experience help explain behaviour - in fact, help cause it. But the epiphenomenalist seemsforced to deny this, potentially rendering it too hard to attain knowledge of other minds. The panpsychist, on theother hand, renders it too easy to attain such knowledge; not only other people but other animals, plants, rocks andprotons have mental features.A final danger with the idea that phenomenal consciousness depends on nonphysical features is that it threatensself-knowledge about one's phenomenal states. The nonphysicalist who believes that zombies are possible believesthat two people could be alike in all non-phenomenal respects, while one has phenomenal experience and the otherdoes not. Each could be fully convinced of their own rich, detailed phenomenal experience, but one would bewrong. It is difficult to see how either could justify their belief that they are not a zombie, if zombies are possible.Chalmers tries to address this problem by stipulating that non-zombies are necessarily distinctively justified:To have an experience is automatically to stand in some sort of intimate epistemic relation to the experience - arelation that we might call ‘acquaintance.' There is not even a conceptual possibility that a subject could have ared experience like this one without having any epistemic contact with it: to have the experience is to be relatedto it in this way…. [This] is to stand in a relationship to it more primitive than belief: it provides evidence forour beliefs, but it does not in itself constitute belief.(1996: 196-7)This conclusion anticipates the idea that inner perceptions of one's experiences - taken to be more ‘primitive' thanreflective beliefs about one's experiences - are necessary for phenomenal consciousness ( §6). But where thenon-zombie has inner perceptual states, by hypothesis the zombie has them also. Perhaps what zombies have aremisperceptions, but this need not lessen their justification in accepting these misperceptions at face value, and inbelieving they have experiences.Also, the heavy reliance on introspections of experience opens up two alternatives for physicalism. The first is todeny that there is experience at all, but only the favoured kind of introspections as of experience. The second -perhaps the more attractive - is to maintain that the favoured kind of introspections are sufficient for experience.This is to invert Chalmers' claim that ‘to have the experience is to be related to it in this [inner perceptual] way'; ifso, this is precisely because to be related to a state in this inner perceptual way is to have an experience. Neitherstrategy carries obvious commitment to anything nonphysical. (For related objections to ‘absent qualia', seeShoemaker (1984: chaps 9, 14); and for a response, see Block (1980). Rey (1986) extends the point, arguing thatnon-psychological requirements on phenomenal consciousness - perhaps biological or neurophysiological ones -would also jeopardize introspective knowledge of experience; see also Chalmers (1996: 7).)5 Eliminativism and the phenomenalIn addition to subjectivity-based arguments, phenomenally conscious experiences provide other interestingarguments against physicalism. In certain experiences one seems to be aware of phenomenal denizens of an innermental world: coloured and shaped mental ‘images', bodily ‘sensations' such as ‘pains' that may be throbbing orin one's limb, and inner ‘speech' with ‘private' volume and pitch. Such alleged mental objects as afterimages,pains and inner speeches, that are naturally reported as having properties of non-mental objects (for example,roundness, throbbingness or loudness), may be called ‘phenomenal objects'. The argument against physicalismbased on reports of phenomenal objects is simple. In such experiences nothing in one's brain or body or (causallyrelevant) environment is literally purple and round, literally throbbing and in a limb, or literally soft andmedium-pitched. So if phenomenal objects do exist with these properties, they are not among the things in one'sbrain or body or environment. One must be either a dualist ( §4) or an ‘eliminativist' about mental entities withthese properties; that is, the physicalist must deny that objects with such properties exist ( Eliminativism).The challenge for the eliminativist about phenomenal objects is to explain why people are often tempted to claimsof phenomenal objects, with ordinary perceptible properties. Broadly, the temptation may be attributed to anambiguity or looseness in ordinary reports of experiences, or to an illusion built into the experiences themselves,which may then be reported strictly and faithfully. Following a suggestion due to Ned Block, Michael Tye (1995)pursues the former strategy by pointing out that people often describe representations as having properties thatthey merely represent: the phrase ‘a warm thermostat setting' may be used for a thermostat setting of warmth, and‘a nude painting' may be used for a painting of nudity. It should be no surprise then that people describeexperiences of colour and shape as themselves being coloured and shaped, and so no surprise that, speaking freely,they treat them as images. The other strategy is latent in J.J.C. Smart's suggestion that ‘There is, in a sense, nosuch thing as an after-image…though there is such a thing as the experience of having an image' (1959: 151). Onone straightforward construal of ‘the experience of having an image', an experience itself represents that there isan image with certain features, although there is no such thing (compare with Sartre's ‘illusion of immanence',1940: 5). The illusory-experience view has an advantage over the reporting-based view to the extent thatafterimages look purple and round, pains feel dull or in motion, and inner speech seems to sound faint orhigh-pitched. By contrast, a warm thermostat setting need not itself feel warm, and a nude painting need not itselflook nude (any more than other paintings look clothed). However, without an account of why experiencesmisrepresent phenomenal objects, the illusory-experience view does not adequately discharge theeliminativist's explanatory burden.In addition to phenomenal objects, eliminativists have targeted alleged ‘phenomenal properties' of experiences, or‘qualia'. There is not merely ‘something' it is like to have a phenomenally conscious experience, but someparticular ‘thing' or things it is like. People sometimes try to describe these particular properties, for example, bysaying that a given pain is ‘sharp' or ‘throbbing' to some degree, or that a given visual image is ‘blurry' or‘moving'. Even if the eliminativist about phenomenal objects is correct that there are only experiences as of sharpthrobbing pains and as of blurry moving images, descriptions such as ‘sharp' and ‘blurry' seem in some indirect ornonliteral way to convey particular aspects of what it is like to have these experiences. In a relatively cautious useof the word ‘qualia', particular what-it-is-like properties, whatever their nature turns out to be, are qualia.There are bolder uses of ‘qualia' on which the word can apply only to properties of experience that posechallenges to scientific explanation. Some require qualia to be infallibly and completely accessible tointrospection, which is puzzling on any plausible scientific explanation of introspection ( §§7-8). Some requirequalia to be inaccessible without introspection - for instance by purely behavioural or neurophysiological tests thatone may perform on other people, without relying on an introspective understanding of one's own experience; thisseems to preclude explanation of qualia as physical properties discoverable in multiple objective ways ( §3).Perhaps the most controversial philosophical idea about qualia, however, is that they are ‘intrinsic' properties ofexperience. Metaphysicians dispute the correct account of intrinsicality, but the following may serve to convey theintuitive idea, as it applies to experience: for an experience to have a property intrinsically, the experience musthave the property solely in virtue of the spatiotemporal parts of what realizes the experience. (This is meant toexclude everything that even in part exists when or where the experience does not, for example, stimuli that causethe experience, and behaviour and other mental states that the experience causes.) This seems to precludeexplaining qualia in ‘functionalist' terms by appeal to the causal role of experiences, or in ‘intentionalist' terms byappeal to the representational content of experiences ( Functionalism). Defenders of the intrinsicality of qualiaoften argue for the possibility of ‘inverted qualia', cases in which two experiences differ in qualia even thoughthey have identical causal or representational relations to their mental and non-mental surroundings ( Block1990).Qualia in such bold senses have been rejected most forcefully by Dennett (§3). He describes several examples inwhich changes in whether we like or dislike certain tastes seem to change the tastes themselves, concluding thatwhen someone ‘thinks of "that taste" he thinks equivocally or vaguely' and ‘need not try - or be able - to settlewhether he is including any or all of his reactions' (1988: 61-3). Nevertheless he accepts that, through the changesin likes and dislikes, ‘the taste is (sort of) the same', and defenders of intrinsicality may hope to explain such tastesimilarities by appeal to reaction-independent components of experiences ( Lormand 1994 for further defenceof bold qualia).An argument against the existence of intrinsic qualia can be built upon what G.E. Moore calls the‘diaphanousness' of perceptual experience:The moment we try to fix our attention upon consciousness and to see what, distinctively, it is, it seems tovanish: it seems as if we had before us a mere emptiness. When we try to introspect the sensation of blue, allwe can see is the blue: the other element is as if it were diaphanous.(1903: 450)Gilbert Harman argues as follows:When you see a tree, you do not experience any features as intrinsic features of your experience. Look at a treeand try to turn your attention to intrinsic features of your visual experience. I predict that you will find that theonly features there to turn your attention to will be features of the presented tree, including relational featuresof the tree ‘from here'.(1990: 39)Defenders of intrinsicality may claim not to satisfy Harman's prediction. This response gains plausibility in casesof degraded perception (for example, blurred or double vision), and it is not clear how Harman's argument issupposed to generalize from perceptual experiences to bodily-sensational, imaginative or thought experiences (bodily sensations; Imagery). (For discussion of imagery by a philosopher in sympathy with Harman, see Tye(1995).) The prediction may hold for non-degraded perceptual experiences and perhaps ‘upgraded' imaginingssuch as dreams, and in these cases perhaps it shows that no experienced features seem intrinsic to experience. Onemay hold that some experienced features are intrinsic to experience, but only by incurring a burden of explainingwhy these features seem to belong to trees and other non-mental objects. In fact, this burden remains even if qualiaare taken to be relational features of perceptual experience, since the relevant experienced features are experiencedas if they belong objectively to trees, and do not seem in experience to depend on the participation of the perceiver.Just as the eliminativist about phenomenal objects of experiences would do well to explain the illusory experienceof their presence, so the non-eliminativist about phenomenal properties of experience - whether intrinsic orrelational - would do well to explain the illusory experience of their absence ( Qualia).6 Introspection and the phenomenalHere the discussion begins to shift from phenomenality to introspection. To clarify the apparent differencebetween phenomenal and introspective consciousness, consider whether a state can be conscious in one but not theother sense. Can there be nothing it is like to have a state, even when one is introspectively aware of it? Can therebe something it is like to have a state, even when one is wholly unaware of it? The facts are murky andcontroversial, but it is important to be clear about the possibilities.The introduction lists four kinds of states that are most clearly phenomenal: perceptual experiences,bodily-sensational experiences, imaginative experiences and streams of thought. There are mental states notexplicitly on this list, notably ‘propositional attitudes' such as the belief that snow is white ( Propositionalattitudes). Usually one's belief that snow is white is latent and unintrospected, though one can raise it tointrospective consciousness easily. Normally there seems to be something it is like to have such an introspectivelyconscious belief ( Belief). However, there is another possibility to explore. What having the belief ‘is like' maybe completely accounted for by what it is like to have experiences accompanying the belief, such as auditoryimaginings of asserting the words ‘Snow is white' (or ‘I believe snow is white', or ‘Mon Dieu! La neige!Blanche!'), or visual imaginings of some fictitious white expanse of snow, together with what William James(1890: 287-8) describes as feelings or imaginings of moving eyeballs, eyelids, brow, breath, jaw-muscles and soon as one thinks. Pending evidence of further aspects of what it is like to have the belief, this illustrates how therecan be something it is like when one has an introspectively conscious state, although the state itself has nophenomenal character (Lormand 1996). Introspective consciousness - at least of the sorts available to beliefs - isunlikely simply to be phenomenal consciousness, and is unlikely to be sufficient for it.Might introspection nevertheless be necessary for phenomenality? There is a tension between a ‘yes' answer andthe view that many species of animals can have experiences - that there is something it is like for cats and dogs tofeel pain or to see bright lights, for instance. It is implausible that these beings have Cartesian reflective knowledgethat they feel pain and see. This would require having concepts of feeling pain and of seeing, and perhaps aself-concept, and all this would seem to involve capacities beyond the reach of most nonhuman animals - forexample, the ability to conceive of others as feeling pain and seeing, and the ability to remember or envisageoneself feeling pain and seeing ( Animal language and thought). Notoriously, Descartes himself accepts thatnonhuman animals are ‘automata' without mental states of any sort. Defenders of a reflective-knowledgerequirement may either mimic this strategy, denying that animals have conscious experiences (Carruthers 1992), orelse attempt to minimize the conceptual sophistication needed for the reflective knowledge (Rosenthal 1990).This tension is more commonly taken to be a serious strike against a reflective-knowledge requirement onphenomenality, especially given that a similar tensionarises in the case of human infants. Thomas Reid objects against Locke that ‘reflection ought to be distinguishedfrom consciousness', since:From infancy, till we come to the years of understanding, we are employed solely about external objects. And,although the mind is conscious of its operations, it does not attend to them; its attention is turned solely to theexternal objects, about which those operations are employed.([1785] 1969: 57)Even for beings with the requisite conceptual capacities, it seems implausible that reflective knowledge mustaccompany each of their experiences. At any given moment one can attend only to a small proportion of thesensory stimuli one encounters. It is also difficult to attend simultaneously to the outside world and to one'sexperience of it, as Auguste Comte argues (1842: 21): ‘The thinking individual cannot cut himself in two - one ofthe parts reasoning, while the other is looking on' ( Introspection, psychology of). Nevertheless, plausibly,there is something many inattentive perceptions of unattended stimuli are like; experience would be quiteimpoverished were it not for the contributions of background noises and odours, pressures on one's feet or seat,moisture under one's tongue, peripheral vision and so on. It is possible to maintain that one continually formsreflective beliefs about these experiences, but this fits poorly with the difficulty of remembering these experiences(after they change, for example).An introspective requirement on phenomenality can blunt much of the force of these objections by distinguishinginner perception from the formation of reflective beliefs, and by distinguishing inattentive introspection fromattentive introspection. Just as one might sense a daffodil without having a concept of daffodils, or a tendency toremember the daffodil, so perhaps one can inwardly sense an experience without having a concept of experiences,or a tendency to remember the experience. Animals and babies might sense even if they cannot form beliefs;likewise, perhaps they can inwardly sense even if they cannot form reflective beliefs ( Belief). Also, accordingto Locke, just as there can be passive sensation, so reflection need not be done intentionally or with attention(1689: II.i.7). Along these lines, Brentano distinguishes between inner perception, which may be automatic andinattentive, and inner observation, which is actively guided by purposeful attention (1874: 29). It is true that acreature's most pressing cognitive needs require mental resources to be directed at the external world, but if innerperception is normally inattentive, it need not draw resources away from attentive outer perception ( Bodilysensations). It is an open question whether there can be phenomenal experiences without any kind of introspectiveawareness of them, even of a primitive sort.7 Completeness of introspectionPerhaps more thoroughly than Descartes ( §1), Locke identifies the mental with the introspectively conscious,claiming that it is unintelligible ‘that any thing thinks without being conscious of it, or perceiving, that it does so'because ‘thinking consists in being conscious that one thinks' (1689: II.i.19). This completeness claim has beenrejected by most subsequent philosophers, with the prominent exception of some in the broadly phenomenologicaltradition, following Franz Brentano and Edmund Husserl ( §10).The main evidence for introspectively unconscious mental processes is that they would fill certain theoretical gapsin the scientific explanation of behaviour and introspectible mental activity. This inferential strategy is mostprominent in psychoanalytic attempts to explain otherwise bizarre dreams, associations among concepts, apparentslips of the tongue, emotional disorders, neurotic physiological reactions and so on ( Freud, S.; Psychoanalysis,post-Freudian; Unconscious mental states). Although the scientific status of such clinically-based explanations iscontroversial ( Psychoanalysis, methodological issues in), alternative explanations are elusive; witnessSartre's difficulties in trying to explain self-deception and ‘bad faith' without appeal to unconscious mentation(1943).Furthermore, many replicable psychological experiments lead to parallel conclusions about introspectivelyunconscious mentation in more mundane settings. In one family of experiments (Lackner and Garrett 1972),subjects are presented with an ambiguous sentence in one ear, and disambiguating words in the other ear, but soquietly as not to be noticed consciously by the subjects - typically the stimulus is reported as a meaningless noise.Nevertheless, the subjects' interpretations of the ambiguous sentence are predictable from the meanings of thedisambiguating words. This is evidence that subjects not only identify but understand the words, withoutintrospective consciousness of doing so.Similarly, there is evidence of vision without introspective consciousness of vision in cases of subliminal visualperception (Dixon 1987) and ‘blindsight' (Weiskrantz 1988). In these cases subjects act on the basis of informationabout the visual features of objects, despite denying - sincerely - that they have relevant visual experiences.Blindsight subjects have damage to certain neural pathways connecting portions of the retina to the visual cortex,yet in some sense they have perceptual states sensitive to these stimuli. For example, some ability to discriminatean ‘X' from an ‘O' is intact. This is evidenced by the preponderance of correct answers they can give to questionsabout the stimuli. When asked to reach for objects in blindsight regions, also, some subjects reflexively pre-orienttheir hand and fingers in ways suited to the specific shapes of the objects. What blindsight patients lack mostclearly is any ability to introspect their perceptual states and abilities: they deny that they are perceiving; theyrespond to the stimuli only when coaxed to do so; and even then they take themselves merely to be guessing.Nevertheless, their guesses tend to be correct and their behaviour appropriate to the stimuli.It is not only in unusual cases that there seem to be introspectively unconscious perceptual states. On virtually alldetailed theories of normal vision, for example, cells in each retina register the amount of incoming light at variouspoints, and cause further states representing sudden discontinuities of incoming brightness, which cause furtherrepresentational states and, eventually, introspectively conscious visual experiences ( Vision). These earlylayers of visual processing seem well beyond the reach of introspection. Although these processes (unlike, say,blood flow in the brain) have many mentation-like features - they are assessable as correct or incorrect in relationto external stimuli; they may increase gradually in stability as evidence for them mounts; they may play a directrole in modulating intentional visuomotor action, and so on - they are at best somewhere in the vague boundarybetween the mental and the non-mental.To the extent that introspectively unconscious perceptual states are phenomenal, they present a new threat to theclaim that introspection is necessary for phenomenality ( §6). Disagreement arises about whether cases ofsubliminal perception or blindsight involve phenomenal consciousness - perhaps blindseeing is like something,despite the subject's sincere denials; at any rate the cases are not clear enough to weigh decisively against thenecessity claim.There is more dispute about whether non-introspectible information-bearing states in early visual processing arephenomenally conscious. On the view that they are, it would be difficult to explain why phenomenal visualexperiences are not continually like double images, given that one has separate left-eye-caused andright-eye-caused early visual states. On behalf of the view, one possibility is that there is something it is like forone's early visual systems to have certain states, although there is nothing it is like for one to have them. Althoughthis is a possibility, it seems no more likely than the possibility that there is something it is like for one's neuronswhen they fire.8 Reliability of introspectionEven if introspective consciousness is limited, it may be epistemologically interesting as an especially reliablemeans of access within its domain ( Introspection, epistemology of). Many philosophers have thought thatintrospective access to mental facts is more reliable than access to other empirical facts. Augustine writes in On theTrinity that ‘nothing can be more present to the mind than the mind itself' (X.3.5), and asks rhetorically, ‘what isso intimately known as the mind, which perceives that it itself exists and is that by which all other things areperceived?' (VIII.6.9). Descartes devotes his Second Meditation to an argument that the mind is ‘better known'than the body, and Locke also claims that our knowledge of ‘Things without us' is ‘not altogether so certain, as ourintuitive Knowledge' (1689: IV.xi.3). As with completeness, most subsequent philosophers reject the infallibilityof introspection, although many would agree that it is relatively reliable.The claim of introspective infallibility is extremely bold. In other empirical domains, at best certain mechanismskeep one's beliefs in rough accord with the facts (for example, mechanisms of perception, reason and memory, andthe persistence of facts when one is not continually checking them). But mechanisms fail; a mechanism of thiscomplexity that could never possibly fail would be a miracle ‘at least as mysterious as papal infallibility',according to Dennett (1988: 55). The same reason to expect fallibility holds for introspection: if there is theslightest mechanism correlating one's thoughts with one's thoughts about them, it should be breakable, and if thereis no mechanism, a perfect correlation between the two would seem to be sheer luck.Furthermore, scientific investigations of introspection have revealed widespread ‘confabulation' in self-access. Inidentifying one's beliefs and motivations, one systematically but sincerely reports attitudes one thinks rational orstatistically normal in the circumstances, even if one does not have them ( Nisbett and Wilson 1977;Introspection, psychology of). For instance, in the ‘bystander effect', increasing the number of joint witnesses to aperson's need decreases the likelihood that any of them will assist. But bystanders rarely report this as a factor intheir decision whether to help, often claiming instead to have reached a decision based solely on their ownlikelihood of success. According to Nisbett and Wilson, much allegedly ‘introspective' access to attitudes consistsof self-directed, fallible guesses, based at best on common-sense abilities to rationalize behaviour. Also, sincethese abilities are at work in one's access to others' mental states, this self-directed guesswork provides somereason to suppose, with Ryle (1949), that introspective consciousness is not interestingly more reliable than accessto other minds.In the face of this evidence, the Cartesian may attempt to identify a restricted domain in which a kind ofintrospection is comparatively reliable (Ericsson and Simon 1993) or even infallible (Lormand 1994). The modelof rationalizing or statistical guesswork does not extend easily to introspection of phenomenally consciousexperiences. For example, untutored subjects offer consistent and apparently reliable reports of ‘feelings' ofstinging (rather than throbbing) pain when a limb has restricted blood flow. They seem not to infer these feelingsin the way one might form beliefs about another's pain feelings, since no common-sense principles of rationalitydictate that one should feel stinging rather than throbbing, and since the subject need know no relevant statisticalinformation about how people feel in these circumstances.Against any attempt to find a restricted domain safe for infallibility, Dennett argues that one can easily be wrongabout the ‘changes and constancies' in one's experience over even brief intervals of time (1988: 59). This weighsagainst infallible memory access to what it was like to have past experiences, though not against infallible orespecially reliable access to current experiences.9 Introspection as inner perceptionHow should introspective consciousness be explained? While Lockean inner perception ( §2) has somecontemporary defenders, most notably David Armstrong (1980), this view has fallen on hard times in philosophyof mind.How is inner perception supposed to be distinctively analogous to outer perception? Of course, inner perceptionsare not generated by literal inner eyes, ears and their attendant experience-forming processes. Armstrong explainsinner perception as being, like outer perception, ‘selective' (incomplete), ‘fallible' and ‘causal'. This illuminateslittle, since probably all cognitive processes have these features, even one's most theoretical (versus perceptual)scientific beliefs, for example, those about quantum mechanics or cosmology. Another initially tempting idea is theHindu one ( §1) that inner perception is a causal but non-inferential source of evidence about mental states (or,more cautiously, that it is as low as outer perception on flexible, all-things-considered inference). Might a mentalstate be introspectively conscious only if its bearer inwardly perceives the state, in this sense? One strongobjection is that introspective access often involves self-directed theoretical inferences, often confabulatory ones. It is therefore best to restrict inner-perception views to current phenomenal experience (in accordancewith the suggestions at the end of §8).An influential objection is that inner perception requires phenomenal ‘sense data' interposed between physicalobjects and one's perceptions of them ( Sense-data; Perception). Accepting inner perception may seem toinvolve accepting that one at best perceives outer objects indirectly through perceptions of phenomenal mentalentities. But such a mediation theory would have difficulty explaining why inwardly perceiving sense-data did notin turn require perceiving further entities (‘sense-data data') and so on, infinitely. Such arguments againstsense-data have been mounted against inner perception (as in Shoemaker 1994), but with unclear effect. Innerperceptions need not be interposed between objects and one's perceptions of them - the causal chain in perceivinga table need not proceed from the table to an inner perception and then to a perception of the table. Rather, on amore natural view, the causal chain goes directly from the table to a perception of the table, and then (in cases inwhich the table perception is introspectively conscious) to an inner perception of the perception of the table.Dennett argues that inner perception of a ‘Cartesian theatre' of consciousness would be too sharp to explain thevague conscious/unconscious distinction, and that inner perception is wasteful - ‘once a discrimination has beenmade once [in outer perception], it does not have to be made again' (1991: 127). Yet it can be vague whethersomething is inwardly (or outwardly) perceived, and inner perception of an outer perception need not re-representfeatures the outer perception is of, but instead may newly represent features of the outer perception ( Lormand1994, for further defence of inner theatres).Perhaps the most influential objection to inner-perceptual introspection is based on Moore's ‘diaphanousness'claim ( §5). The objection is that, since each outer-perceptual modality (ing, hearing and so on) makes itsown distinctive contribution to what experience is like, an additional modality of inner perception should beexpected to make its own contribution, to change what it is like. But what it is like to introspect a perceptualexperience seems simply borrowed from what it is like to have the experience itself (McGinn 1982: 50-1). Whenone tries to attend to features of normal experiences, one normally ‘sees through' the experiences to outer objects.So a fundamental disanalogy between outer perception and alleged inner ‘perception' is that the former, but notthe latter, has its own phenomenology or perceptual quality. This is evidence against inner perception of currentphenomenal experiences.On the other hand, if inner perception is necessary for phenomenality ( §6), then instead of borrowingphenomenal qualities from an outer perception, as the diaphanousness objection alleges, inner perception helpsgenerate these qualities together with the (otherwise phenomenally unconscious) outer perception. This mayexplain why inner perception does not add further qualia to an outer-perceptual experience; inner perception mayalready make its phenomenal contribution in the generation of an outer experience with qualia ( Qualia).10 Alternatives to inner perceptionTwo rivals to inner-perception accounts treat introspective access as epistemically less intimate than innerperception. First, James maintains that introspection is always retrospective (1890: 187-), largely in reaction toComte's denial that the mind can simultaneously split between ordinary thinking and awareness of that thinking( §6). James is unhappy with Brentano's response to Comte - that inattentive perception can be split betweenouter and inner domains - because James seeks to defend the reliance on careful, attentive introspective reports inexperimental psychology ( James, W. §2). Second, some maintain that introspection is always laden withtheoretical (versus perceptual) inferences. The experimental evidence for confabulation and expectation-driveninference ( §8; compare with Lyons 1986) suggests that introspection is often theory-laden and retrospective,but does not suggest that all cases of introspective consciousness are, including the seemingly non-inferentialconsciousness of phenomenal experiences that seem to persist while being accessed.One theory of consciousness that combines naturally with theory-ladenness is Rosenthal's ‘higher-order thought'theory (1990). (A thought or belief is ‘higher-order' in virtue of being about mental entities rather thannon-mental entities.) According to Rosenthal's view, a mental state is conscious just in case one forms, in asuitably direct way, a thought that one has the state. The state may generate its higher-order thought throughinference so long as these inferences are not themselves conscious. This condition is intended to rule out cases inwhich one comes to think about a state through very indirect inference - say, solely through believing thetestimony of a psychologist. On this view, even if introspective access to phenomenally conscious experiences issomehow inferential, it would seem non-inferential simply because one lacks higher-order thoughts about theinferences.At the other extreme from defenders of theory-ladenness and retrospection, many phenomenologists reject innerperception as not being intimate enough to explain introspective access. On one suggestion, some introspectivelyconscious mental states ‘reflexively' represent themselves (in addition to representing other things). Thisconclusion is often embraced to avoid an infinite regress threatening the assumption that introspection is complete( §6): an experience represents itself rather than being represented by a separate introspective state which(assuming completeness) must in turn be represented by a separate introspection of the introspection, and so on.Brentano argues that ‘The presentation which accompanies a mental act and refers to it is part of the object onwhich it is directed' (1874: 128). Husserl also suggests that ‘In the case of a perception directed to somethingimmanent [that is, roughly, mental],…perception and perceived form essentially an unmediated unity, that of asingle concrete cogitatio' (1913: 112). And Sartre insists that ‘the first consciousness of consciousness' - what hecalls ‘pre-reflective consciousness' - ‘is one with the consciousness of which it is [a] consciousness' (1943:13-14). Given the prevalence of inferential, confabulatory access to one's mental states, reflexivity theories, likeinner-perceptual ones, are best restricted to current phenomenally conscious experiences rather than to othermental states. Again, there is a possibility that one's only access even to these experiences is somehow much moreconfabulatory or inferentially sensitive to expectations than ordinary perception is, but pending evidence for thispossibility the restriction to phenomenal experiences is reflexivity's best hope.Since one often suffers ordinary perceptual illusions, the more analogous introspection is to perception, the morelikely it would be that one would suffer naïve introspective illusions about what one's conscious experiences arelike. But it rarely if ever happens that one mistakes, say, a dull pain for a sharp pain, in the way that one mistakes a