110

A plug for generic phenomenology doi: 10.1017/S0140525X07002841 Rick Grush Department of Philosophy, University of California San Diego, La Jolla, CA 92093. rick@mind.ucsd.edu http://mind.ucsd.edu Abstract: I briefly sketch a notion of generic phenomenology, and what I call the wave-collapse illusion– a less radical cousin of the refrigerator light illusion– to the effect that transitions from generic to detailed phenomenology are not noticed as phenomenal changes. Change blindness and inattentional blindness can be analyzed as cases where certain things are phenomenally present, but generically so. Anumberofresearchershaveusedmismatchesbetweenwhatsubjects think they are aware of and what subjects can report on, to argue that we are subject to a grand illusion– that the idea that we are presented with a rich, detailed visual scene is false. Block 504 shows that with some of these phenomena (e.g., the phenomena in the Sperling [1960] and theLandmanetal.[2003] experiments), thebetteranalysisisthattheitemssubjects thinkarephenomenally present are in fact present, but inaccessible in a certain sort of way. But there are other kinds of cases– change blindness and inattentional blindness– to which Block’s analysis doesn’t seem to apply, and so the threat of this illusion remains. I want to assist Block in dulling the teeth of the illusion argument. The first topic I will discuss involves a sketch of the notion of generic phenomenology, the idea that some of what is phenomenally present is present only generically. Roughly the idea is that, for example, when reading a page of text, a word on the page in the visual periphery is phenomenally present, but not as the specific word it is, just as the generic text word. If I am right, then much mental representation, including phenomenology, is sui generis, and not well described as either picture-like or sentence-like; and in particular, generic phenomenology is neither adescription,norablurryimage.Nor is itaninterpreted image (a combo-metaphor of both picture and sentence). The second part suggests how the problematic phenomena, change blindness and inattentional blindness, can be given plausible, and less alarmist, interpretations using the notion of generic phenomenology. Whilethere isnospaceforafull positive characterization of this sui generis format, I sketch three of its features here: First, it is possible that incorrect genera are employed in some cases. Second, a generic phenomenal presentation can present its content as clear without presenting the details. In this respect it differs from an image. This is, I currently believe, an affordance issue: a clear phenomenal presentation presents its content as affording answers to queries of detail. Images only offer this affordanceif they actually have the detail. Third and relatedly, this representational format supports and is subject to a less radical version of the refrigerator light illusion, what I will call (with a great deal of trepidation, since I don’t want to encourage quantum-mechanical approaches to the study of consciousness) the wave collapse illusion: If a given generic phenomenal presentation is clear because it presents itself as affording detail queries of a certain kind, and the phenomenal presentation changes from generic todetailed asaresultofsuch aquery, thenthistransition is not noticed as a phenomenal transition. This is actually a simplification, as I think there are two kinds of clarity: being presented (1) as affording answers to queries effected by modulation of sensory apparatus (e.g., moving the eyes), and (2) as affording answers to queries effected by attention modulation alone without modulation of sensory apparatus; both contrast with being presented as query-resistant, for example as occluded or blurry. For purposes of this commentary I speak here of a univocal clarity, though if pressed I would hypothesize that the wave-collapse illusion is present, or strongest, only in the latter sort of case. (As I understand it, the refrigerator light illusion is more radical, because it claims that when something can be expected to be present upon query, then the transition from nothing to the presence of that thing is not noticed as a phenomenal change. It is not obvious to me that we are really subject to this illusion.) So, most of the words on the page are generically phenomenally present; some at the small region of foveation are present in a phenomenally detailed way, and some near this area, but just outside it, are presented as clear but only as the generic text word. When the eyes are moved to foveate the word, the added sensory detail effects a transition from the phenomenally generic text word to a phenomenally specific and detailed presentation of that text word. Now, because the genera employed can be wrong, it is possible to arrange for a situation where all the “words” in this penumbral area are switched around when an eye movement to that area is detected. The subject will notice nothing because (1) the genus text word was used for the clear generic; (2) that genus remained applicable when more detail was available; and (3) the wave collapse illusion was in effect. Whenbeingaskedtocount ball passes between people clothed in white, the perceptual system employs the ad hoc genera BEHAVIORAL AND BRAIN SCIENCES (2007) 30:5/6Commentary/Block: Consciousness, accessibility, and the mesh person-clothed-in-white and person-clothed-in-black. A person in a gorilla suit walking through the middle of the scene is phenomenally presented as clear and under the genus person-clothed-inblack. Many instances of change blindness would seem to be cases where some corresponding regions or objects in two scenes or images are phenomenally present as clear but generic, and under the same genus– even though either that one genus is applicable to both but the details differ, or in one case the genus is actually wrong. If this admittedly sketchily presented hypothesis is on the right track, then we are under an illusion regarding the phenomenal content we entertain– but not a terribly radical illusion. We seem to have a full, richly detailed, phenomenal representation of the visual scene, though in fact what we have is, albeit full and clear (in the specified sense), only actually detailed in some places and in some respects, and in other places and other respects it is clear but generic. This might seem like no large departure from what some radicals have defended, but in fact it provides traction at exactly the point where the slide from “no fully detailed rich internal representation” to “no internal representation” occurs. What is cognitively accessed? doi: 10.1017/S0140525X07002853 Gilbert Harman Department of Philosophy, Princeton University, Princeton, NJ 08544-1006. harman@princeton.edu http://www.princeton.edu/~harman/ Abstract: Is Block’s issue about accessing an experience or its object? Having certain “flow” experiences appears to be incompatible with accessing the experience itself. And any experience of an object accesses that object. Such access either counts as cognitive or does not. Either way, Block’s issue seems resolvable without appeal to the scientific considerations he describes. Block is concerned with the question of whether there are cases of phenomenology in the absence of cognitive access. I assume that, more precisely, the question is whether there are cases in which a subject S has a phenomenological experience E to which S does not have direct cognitive access? (S might have indirect cognitive access to E through scientific reasoning. I take it that this is not the sort of cognitive access in question.) It is somewhat unclear– in at least two ways– what Block means by “cognitive access.” First, it is unclear what cognitive access is supposed to be access to. Second, it is unclear what makes access cognitive. Let mebegin my discussion with the first question, about what cognitive access is supposed to be access to. Suppose first that E is not an experience of or awareness of a phenomenal or intentional object X. Then it would seem that the relevant cognitive access can only be to E itself, that is to S’s having E. On the other hand, suppose that the relevant phenomenological experience E involves being aware of something X, the phenomenal or intentional object of E, what S is aware of in having experience E. In that case, would the relevant cognitive access be access to X or to E? It may be that S’s experience of X is compatible with and perhaps even sufficient for S to cognitively access X, although S’s trying cognitively to access E is incompatible with S’s having the experience E. In such a case it would seem that S does not have (direct) cognitive access to E. For example, S’s being completely engaged in what S is doing– as in optimal “flow” experience (Csikszentmihalyi 1990)– is compatible with (and even sufficient for) S’s cognitively accessing what S is doing while at the same time at least sometimes being incompatible with S’s being aware of being so engaged. In such cases it would seem that S does not have direct cognitive access to having such flow experiences. So, if the relevant cognitive access is access to E, there seem to be clear cases of phenomenology without cognitive access, and no need for the sort of investigation Block describes. Alternatively, the relevant cognitive access might be to the (intentional or phenomenal) object X of S’s experience E. This seems to be the sort of cognitive access Block has in mind. Consider his discussion of the subjects in Sperling’s (1960) experiment who reported being aware of all the items in a briefly displayed grid even though they could identify only some of the items. The items in question are the objects of the subject’s perceptual experiences. But can one have a phenomenal experience of X without having cognitive access to X? Indeed, can one have a phenomenal experience of X without that phenomenal experience of X being a cognitive experience of X that constitutes cognitive access to X? Or, to put the question the other way, can there be a phenomenal experience of X that is not itself a cognitive experience of X? What could possibly distinguish a cognitive experience of X from a noncognitive experience of X? Clearly, it depends on what is meant by “cognitive” experience. Suppose that a necessary condition of E’s being a “cognitive” experience of X is that E should have a certain sort of “intentional content”– an experience of X’s being F, for some relevant F. Given this supposition, if S’s having E is a cognitive experience, E consists at least in part in X’s seeming (appearing, looking, etc.) to S to be F for some relevant F. Furthermore, it might be suggested that, S’s having a phenomenal experience E of X need not (and maybe never does) consist even in part in X’s seeming to S to be F. If so, then given the aforementioned supposition, it would seem that there could be a phenomenal experience E that is not cognitively accessible. So it may seem highly relevant whether some S could have phenomenal experiences of X that did not consist at least in part in X’s seeming to S to be F for some relevant F. In Harman (1990) I argue (in effect) that the answer is “No, this is not possible.” In fact, I argue that phenomenal content is the same thing as intentional content, a conclusion that is widely (but not universally) accepted in recent philosophical discussion. In this (controversial) view, S must have cognitive access to the object of a phenomenal experience E because any phenomenal experience of X is itself a cognitive experience of X. Not only are the objects of phenomenal experience cognitive accessible, they are ipso facto cognitively accessed. If Block assumes that phenomenal content is not always intentional content, so that the phenomenal content of experience is not always cognitively accessed, this assumption by itself appears to guarantee a negative answer to Block’s question without any appeal to the scientific considerations he mentions. The “mesh” as evidence– model comparison and alternative interpretations of feedback doi: 10.1017/S0140525X07002865 Oliver J. Hulmea and Louise Whiteleyb aAnatomy Department, Wellcome Trust Centre for Neuroimaging, University College London, London WC1E6BT, England, United Kingdom; bGatsby Computational Neuroscience Unit, University College London, London WC1N 3AR, England, United Kingdom. o.hulme@ucl.ac.uk l.whiteley@ucl.ac.uk http://www.gatsby.ucl.ac.uk/~louisew/ Abstract: We agree that the relationship between phenomenology and accessibility can be fruitfully investigated via meshing, but we want to emphasise the importance of proper comparison between meshes, as well as considerations that make comparison especially difficult in this 505 BEHAVIORAL AND BRAIN SCIENCES (2007) 30:5/6Commentary/Block: Consciousness, accessibility, and the mesh domain. Wealso argue that Block’s interpretation of the neural data in his exemplar mesh is incorrect, and propose an alternative. WhatBlockcalls a “mesh”canalso beconsidereda “model”ofthe interrelationship between neural and psychological data, and of the relationshipofboth sourcesofdatatotheconceptsofphenomenology and accessibility. This kind of model fitting is massively underdetermined, and there is a real danger of overfittingmassaging the data to create meshes that can support anyassumption. In a Bayesian framework, competing quantitative models are evaluated by comparing their marginal likelihoods– how likely the observed data is under each model. The marginal likelihood incorporates an Occam’s razor–like penalty, penalising complex models that have more parameters, and are thus more likely to overfit the data (Mackay 2004). The kind of “mesh” at issue here is clearly not yet a quantitative model for which a marginal likelihood can be computed, but the principles of model comparison should still be followed as closely as possible. Block presents a single mesh whose explanatory power gives us “reason to believe” (target article, sect. 1, para. 2) the assumption embedded in it. We would like to emphasise the necessity of comparing fully developed competing meshes before we can have more reason to believe Block’s assumption than its counterpart– asingle entrant in a race will always be the winner. Wewill now use Block’s example to demonstrate some of the issues inherent in evaluating whether a particular mesh really does provide the “best explanation” in this domain. In his example, Block claims that without making the assumption that accessibility is constitutive of phenomenology it is not possible to build a mesh that provides a mechanism for overflow. Thestarting point is an argument from behavioural data that phenomenologyoverflows accessibility, which is then taken as a given for the mesh to explain. This conclusion is itself a kind of inference to the best explanation, which ultimately contains an assumption about the answer to the methodological puzzle. The importance of the puzzle is diminished by embedding an assumption about the answer within a larger explanatory structure. However, how muchthestructurerelies upon such assumptions will affect the strength of the evidence it provides, and it is therefore essential to be extremely precise about which assumptions are made, and where. Another issue that makes model comparison particularly difficult in this domain is the imprecise terminology. In his example, Block uses unitary concepts of accessibility and reportability to refer to different phenomena, in different contexts. This can lead to confusion. For instance, whatever is currently part of access-consciousness is reportable in the sense that it can be reported at any time, without any reorienting of attention or the sense organs. Objects in a Sperling task are reportable only if attended, but no terminological distinction is made between these two types of reportability even though the distinction is critical to Block’s disentangling of phenomenology and accessibility. We propose that a more precise taxonomy of the different types of accessibility and reportability would be easy to develop and would resolve much of this confusion. A significant source of difficulty in building and comparing meshes is how we should describe the neural data, and how it maps to psychological states. Block acknowledges that the neural data he invokes may be “wrong or at least highly incomplete” (sect. 13, para. 3). Therefore, great care should be taken at this stage because an error here can cascade through the mesh, severely limiting the strength of the evidence it provides. In demonstration of this point we suggest an alternative mapping between neural states and phenomenology which, in our opinion, would strengthen the mesh by better accounting for the data, or at least cast doubt over Block’s version. Block proposes that cortico-cortical (CC) recurrent loops from higher to lower areas are the core neural basis for phenomenologyfor example, that feedback from V5 to V1 is constitutive of phenomenal motion. We propose that motion phenomenology 506 should instead be viewed as being composed of different component phenomenal characters that correspond to the activity of independent but interacting cortical areas, and that recurrent feedback is then necessary for binding these phenomenal characters together. As an illustrative oversimplification, the phenomenal experience of a simple moving grating could be characterised as being composed of a phenomenal experience of a retinotopic array of contrast elements, and a phenomenal experience of motion within a particular retinotopic region. It has been argued elsewhere that the best candidate for the core neural basis of the phenomenal experience of contrast elements is V1 (Pollen 1999) and of motion is V5 (Zeki & Bartels 1999). Under normal conditions the two attributes are bound to each other and one directly experiences the surface qualities of the thing that is moving– a modal experience. If the two are not bound, one will have an amodal experience of motion. This hypothesis can be tested by looking at what happens when feedback from V5 to V1 is prevented. Block claims that V5 activity over a certain threshold in the absence of feedback to lower areas is not sufficient for the experience of motion (sect. 13). We would claim that it is in fact sufficient for an amodal phenomenal experience, supported by the observation (cited by Block in his Note 10) that patient G.K. can experience fast motion in his “blind field,” described as looking like “black moving on black” (Zeki & ffytche 1998), despite having no possibility of recurrent feedback from V5 into V1/V2 in his damaged hemisphere (Semir Zeki, personal communication). The same logic can be applied to the transcranial magnetic stimulation (TMS) evidence that disrupting feedback from V5 to V1 prevents the perception of a moving phospene (Pascual-Leone &Walsh 2001). Subjects reported that the phosphene was “present but stationary,” which is equally consistent with the role of CC feedback in binding phenomenal contents. A true test would be whether subjects can distinguish between sham and real V5 stimulation in the context of disrupted feedback to V1, and whether this is correlated with an experiential report of amodal motion. Adopting this alternative mapping would potentially strengthen Block’s mesh by better f itting the neuroscientific evidence. Block’s proposed approach compares meshes as a proxy for a direct comparison of competing hypotheses about the relationship between accessibility and phenomenology. To make this comparison meaningful, we must be very precise about the necessity of the hypothesis to the explanatory power of the mesh. From this perspective the explanatory gap may prove to be more like a crack in the pavement than a gaping hole, but we must still take great care in stepping over it. Many ways to awareness: A developmental perspective on cognitive access doi: 10.1017/S0140525X07002877 Carroll E. Izard, Paul C. Quinn, and Steven B. Most Department of Psychology, University of Delaware, Newark, DE 19716. izard@udel.edu pquinn@udel.edu most@udel.edu Abstract: Block’s target article makes a significant contribution toward sorting the neural bases of phenomenal consciousness from the neural systems that underlie cognitive access to it. However, data from developmental science suggest that cognitive access may be only one of several ways to access phenomenology. These data may also have implications for the visual-cognitive phenomena that Block uses to support his case. Prelingual infants perform mental operations that signal levels of awareness beyond phenomenal consciousness and that may be observed or indexed by processes other than those involved in cognitive accessibility. These include: (1) using action selection processes to group stimuli into perceptual categories, thereby BEHAVIORAL AND BRAIN SCIENCES (2007) 30:5/6Commentary/Block: Consciousness, accessibility, and the mesh laying groundwork for subsequent concept formation; (2) recognizing and responding appropriately to faces and facial expressions of basic emotions and making stable emotion-cognition connections; and (3) exhibiting movements indicative of intentionality, goal-directed behavior, and problem-solving (Slater & Lewis 2007; cf. Merker 2007). These behaviors may reflect the developmentofdifferent levels or complexities of awareness and offer possibilities of extending current conceptualizations of ways to access phenomenal experience. Mechanisms of accessibility and levels of awareness. Developmental scientists routinely observe evidence suggesting that prelinguistic infants not only experience objects and events phenomenologically, they respond to them in meaningful ways. They discriminate between animate and inanimate faces (Ellsworth et al. 1993), familiar and strange persons (Bushnell et al. 1989), and among a wide variety of objects sufficiently well to place them in categories (Quinn 2006). As well, they respond differentially to others’ emotion expressions (WalkerAndrews 1998), and execute movements to influence persons, objects, and events, and to solve problems (Sommerville & Woodward 2005). Foundations of concept formation. Through visual tracking and eye movements, young infants have shown that they can parse visual experiences into perceptual groups that subsequently attain conceptual significance. For example, 3-to4–month-olds presented with visual images of realistic photographs of cats generalize their looking-time responsiveness to subsequently presented novel cats, but use comparison and selection processes to display visual preferences for exemplars from novel categories including birds, dogs, and horses (Eimas &Quinn 1994; Oakes & Ribar 2005; Quinn et al. 1993). Studies demonstrating categorization abilities early in life indicate that infants do not experience objects in the world as undifferentiated, but as separate groups that fall into distinct representations (Quinn & Eimas 1996). These representations may then serve as placeholders for the acquisition of the more abstract and non-obvious information that occurs beyond infancy, through language and more formal learning of semantic categories (Quinn & Eimas 1997; 2000). Thus, over time, the perceptual placeholder representation for cats will come to include the information that cats eat tuna, hunt mice, give birth to kittens, have cat DNA, and are labeled as “cats.” Emotion expression. In mother-infant face-to-face play, young infants display emotion expressions in synchrony with the mother but also periodically take the lead, indicating behavior other than imitation (Stern 1974; Tronick 1989). Their behavior in the “stillface” procedure is typically interpreted as an effort to re-engage the mother (Muir & Lee 2003). 2.5-to-9–month-old infants respond differentially and predictably to the mother’s discrete emotion expressions (Izard et al. 1995; Montague & WalkerAndrews 2001). Through their emotion expressions, children without a cerebral cortex also show evidence of access to phenomenal experience (Merker 2007). Body movement/action. Kinematic data have shown that 10-month-old infants reach more rapidly for a ball that they have been encouraged to throw into a basket than for one they have been encouraged to fit into a plastic tube (Claxton et al. 2003). From ages 9 to 19 months, infants show clear developmental changes in target selection and movements that ultimately lead to problem solving (McCarty et al. 1999). The three foregoing types of data suggest that prelingual infants have emotion- and action-systems that mediate access to contents of phenomenal experience independently, or largely independently, from those involved in cognitive access. Thus young infants (like hydranencephalic children; cf. Merker 2007) appear to possess forms of accessibility that may lie outside the pale of Block’s “cognitive” criteria. Independence and interdependence of access modes. Evidence suggestive of accessibility via emotion- and actionsystems (in the absence of cognitive accessibility) can be found in adults as well as in children. However, due to maturation and resultantly enriched connections among neural systems, examples of functionally independent access pathways may become rarer with age. In adults, observations of such independence might often require artificially constrained manipulations or incidental brain lesions. Milner and Goodale’s (1995) work with patient D.F., for example, suggests that dorsal and ventral visual pathways output separately to action and cognitive systems respectively (also see James et al. 2003). When asked to indicate explicitly the orientation of a slot, D.F. (who had a compromised ability to process information via her ventral visual pathway) was unable to do so. However, when asked to insert a card into the slot as if posting a letter, her action was immediate and correct. Examples also abound of situations where emotion processes seem to glean aspects of experience that lie beyond reach of cognitive accessibility. These include emotion’s role in decision making (Bechara et al. 2000) and in the power of emotional stimuli to guide attention even when people cannot report them (Jiang et al. 2006). Recent evidence suggests that contents accessible to the neural systems of emotion can be made available to the systems of cognitive access (e.g., particularly emotional task-irrelevant stimuli appear to gain access to explicit report mechanisms at the expense of non-emotional target stimuli; Arnell et al. 2007; Most et al. 2005a). However, the quality and function of the experience change after the emotion-cognition connection (Izard, in press). For example, when people label emotions, there follows an observable activation decrease in neural areas associated with emotional reactivity (Lieberman et al. 2007). Emotion-cognition-action connections and interactions have played a critical role in the evolution and functioning of consciousness and continue to influence the development of higher levels of awareness in ontogeny. Implications for visual cognition. Evidence of multiple routes for accessing phenomenology might help reframe findings from the visual cognition literature, several of which Block described in making his argument. Rather than casting phenomena such as the attentional blink (Chun & Potter 1995; Raymond et al. 1992), inattentional blindness (Mack & Rock 1998; Most et al. 2001; 2005b; Simons & Chabris 1999), and object substitution masking (Di Lollo et al. 2000; Reiss & Hoffman 2006; Woodman & Luck 2003) in terms of a conscious versus unconscious divide, it may be more fruitful to regard them as failures of cognitive access but not necessarily as failures of other types of access. The limitations that constrain cognitive accessibility might not generalize to other forms of accessibility. For example, evidence suggests that whereas people often fail to detect neutral targets during an attentional blink, emotion stimuli are much less susceptible to this effect (except in cases of bilateral amygdala damage; Anderson & Phelps 2001). A different look at accessibility and reportability of levels of awareness. Developmental research potentially allows us to examine a time before connections between cognition-, emotion-, and action-systems are fully formed, thereby providing insights that might not be as readily gleaned from, but which may affect interpretation of, the adult literature (e.g., evidence of separable access systems). The developmental data also have implications for Block’s endeavor to pinpoint neural bases of phenomenal experience unadulterated by access mechanisms. His suggestions for ruling out the machinery of cognitive accessibility are important, but ruling out mechanisms involved in accessibility via emotion and action processes may prove even more challenging. Block’s ability to construct an empirically based proposition for how phenomenology might overflow cognitive accessibility gives testimony to the burgeoning of cognitive science investigations relevant to this enterprise. Empirical investigations of similar constraints on emotion- and action-accessibility have yet to catch up. We hope that Block’s path-making endeavor will kick-start this process. 507 BEHAVIORAL AND BRAIN SCIENCES (2007) 30:5/6Commentary/Block: Consciousness, accessibility, and the mesh What is “cognitive accessibility” accessibility to? doi: 10.1017/S0140525X07002889 Pierre Jacob Institut Jean Nicod, UMR 8129, EHESS/ENS/CNRS, DEC, Ecole Normale Supe´rieure, Pavillon Jardin, 75005 Paris, France. Pierre.Jacob@ehess.fr http://www.institutnicod.org Abstract: I first argue that some of Block’s formulations may misleadingly suggest that the function of mechanisms of so-called cognitive accessibility is to makeoneaware,notofvisiblefeaturesofthe visible world, butofone’s own psychological life. I then ask whether Block’s view of phenomenology in the present target article is consistent with his endorsement of non-representationalism elsewhere. According to Block’s diagnosis, the present scientific evidence is compatible with the hypothesis that the neural machinery underlying cognitive accessibility is not a constitutive part of the neural machinery underlying visual phenomenology. The evidence shows that locally recurrent activity within the occipital areas is not sufficient for cognitive accessibility: the latter requires broadcasting to a global workspace involving frontal and parietal areas by means of long-range projections from the occipital areas. But the evidence does not rule out the hypothesis that recurrent activity within occipital areas is sufficient for visual phenomenology. I find Block’s diagnosis compelling and his hypothesis quite plausible. Here, I want to press him on a pair of conceptual issues respectively raised by his present conception of cognitive accessibility and by his present conception of visual phenomenology. Block’s hypothesis presupposes the acceptance of some such psychological distinction as the distinction between belief (or judgment) and phenomenal experience. One can visually experience (or see) an object that exemplifies the determinate property F without using (or even possessing) the concept of F. But one cannot believe that some object is F unless one possesses the concept of property F. As Block recognizes, the belief–experience distinction faces a direct challenge from the behaviorist reliance on reportability. Since the content of one’s report depends on what one believes, not on the content of one’s visual experience, one basic reason for rejecting the distinction between experiencing and believing is the behaviorist suspicion that, unlike the content of one’s belief, the content of one’s experience runs the risk of being unreportable and thereby escaping the scope of scientific investigation. If one accepts the belief–experience distinction, then the question may also arise: What does it take to be aware of one’s own experiences? At one extreme is Dretske’s (1993; 1994) view that one might have a conscious experience and not be conscious of having it. At another extreme is Dennett’s (1991) firstperson operationalist view, according to which one cannot be aware of a stimulus unless one believes that one is aware of it. Blockrejects first-personal operationalism.But insteadofendorsing the view that one might have a conscious experience and not be conscious of it, he considers a deflationary view and the “same order” view. I wonder why. On the deflationary view, one is supposed to experience one’s experience just as one dances one’s dance or one smiles one’s smile. On the same order view, a conscious experience is reflexive. The deflationary view sounds to me like a failed attempt at dissolving rather than solving the problem of self-awareness. Certainly, one is not having a visual experience of one’s own visual experience of a red rose when one is having a visual experience of a red rose. Nor is it clear that the same order view can accommodate Block’s explicit purpose of providing a unified account of experience: Is a mouse in a reflexive state of awareness when it sees a piece of cheese? Furthermore, Block sometimes gives, I think, the misleading impression that the function he assigns to the mechanisms of 508 cognitive accessibility is to make one aware, not of features of distal stimuli, but of one’s own phenomenology or of features of one’s own phenomenally conscious experiences. For example, early on he introduces cognitive accessibility in terms of Fodor’s criterion of modularity according to which we do not have cognitive access to some of our own perceptual states and representations. This is slightly misleading because Block’s view is that by being broadcasted to the global workspace for further processing, the content of a dominant visual representation makes one cognitively aware of visible features of one’s non-mental environment, not of one’s own psychological life (or computational architecture). I now turn to Block’s present conception of visual phenomenology. So far as I can see, none of Block’s arguments in this paper presupposes his (2003; 2007) anti-representationalist view of the character of phenomenal visual experiences. Given my own representionalist bias, this independence makes his arguments for the view that the neural machinery of visual phenomenology does not include the neural machinery of cognitive accessibility more easily acceptable. In particular, I fully concur with the main line of Block’s accounts of the Sperling experiment, change blindness experiments, and the Landman et al. experiment. Consider Block’s present account of the Sperling experiment. To say of subjects that they visually experience the whole array of 12 alphanumeric characters is to say that they are able to bring each character under the general concept “letter.” The reason they fail to report more than 4–5 of such characters is that they fail to bring more than 4–5 of them under the concepts of their distinctive shapes. On this view, the content of a reportable representation seems to stand to the phenomenal character of a non-reportable experience of one and the same stimulus just as the concept of a determinate shape property (e.g., rectangle) stands to the concept of a determinable shape property (e.g., polygon). If so, then the question arises: Is the rejection of representationalism, which Block has endorsed elsewhere, consistent with his present account of visual phenomenology? The question, I think, is made more pressing by the fact that Block is willing to draw a contrast between two memory systems with different storage capacities: The capacity of the so-called visual phenomenal memory system is said to overflow the capacity of working memory. But on the face of it, the argument for the view that the neural machinery for visual phenomenology does not contain the neural machinery for cognitive accessibility seems to presuppose that only contents that are cognitively accessible, not the contents of visual phenomenology, are available for further processing by such cognitive mechanisms as attention, memory, and reasoning. Would Block be willing to postulate a special phenomenal attention system and a special phenomenal reasoning system for visual phenomenology? If not, then why not? Whysingle out visual phenomenal memory among other cognitive mechanisms? Incomplete stimulus representations and the loss of cognitive access in cerebral achromatopsia doi: 10.1017/S0140525X07002890 Robert William Kentridge Department of Psychology, University of Durham, Durham DH1 3LE, England, United Kingdom. robert.kentridge@durham.ac.uk Abstract: When processing of stimuli occurs without attention, phenomenal experience, as well as cognitive access, may be lost. Sensory representations are, however, constructed by neural machinery extending far beyond sensory receptors. In conditions such as cerebral achromatopsia incomplete sensory representations may still elicit BEHAVIORAL AND BRAIN SCIENCES (2007) 30:5/6Commentary/Block: Consciousness, accessibility, and the mesh phenomenal experience but these representations might be too aberrant to be integrated into the wider cognitive workspace. In the target article Block aims to identify circumstances in which stimuli might elicit phenomenal experience but not elicit a representation which is cognitively accessible. In the examples Block considers, cognitive accessibility is impeded through deficits in attention (the extinction example) or through limitations in processing time (the partial report examples). Both mechanisms depend upon some constraint in processing the stimulus. There must be a concern that this might affect sensory processing, andhencethat phenomenal experience of the stimulus is affected along with cognitive access. We know, for example, that attention affects visual sensitivity (Solomon 2004). Changes in the response gain of neurons in sensory areas of cortex are as likely to be affecting phenomenal experience as cognitive access (Carrasco et al. 2004; Treue & Martı `nez Trujillo 1999). Are there circumstances in which the object of sensory processing can be examined at leisure and can be fully attended, and yet cognitive access is lost when phenomenal experience survives? What does it mean for a representation to be cognitively accessible? As Block notes, “mechanisms of reporting, reasoning, evaluating, deciding, and remembering” (sect. 11, para. 2) should be able to make use of such a representation– a key aspect of cognitive access in models such as Baars’ Global Workspace Theory (Baars 1997) is that information about a stimulus becomes widely available. The representation must therefore take a form that permits interaction with memories or plans about other stimuli (in Piagetian terms, for a cognitive system to assimilate a representation, the former must be accommodated to the latter). If stimuli no longer engage representations that can be integrated in any sense with the rest of a person’s representational framework, then surely they cannot be said to be cognitively accessible. How can representations become isolated from cognition? One might erroneously assume that sensations are “raw”– that they do not need to be prepared in a fashion to make them accessible to cognition. Sensations do not, however, correspond simply to the activity of sensory receptors. In color vision, for example, color appearance is far more closely related to a distal property of surfaces (things in the world about which we have cognitions), their spectral reflectance, than to the activations of cone photoreceptors in the retina (Hofer et al. 2005). Color sensation is the endpoint of a complex process. If this process is prevented from running to its conclusion we may be left with an incomplete signal that gives rise to sensation but cannot be integrated with cognition. Stimuli may therefore potentially elicit phenomenal experiences and these experiences may be discriminable yet they remain cognitively isolated. There is no framework within which to reason about or evaluate their differences, nor can they be remembered (I am not sure I see mere discrimination or decision as an act of cognition, but neither does Block when he discusses “guessing” in blindsight). Dosuch circumstances exist? I will argue there are neurological patients who have all the time in the world to process stimuli, who can attend to them, but who have sensations divorced from cognition. Cerebral achromatopsia is a neurological condition in which color vision is lost as a consequence of damage to ventromedial occipital cortex, usually in the vicinity of the fusiform and lingual gyri (Meadows 1974). Unlike more usual forms of color blindness there are no deficits or abnormalities in the retinal cone photoreceptors which form the starting point of normal color perception. Patients with cerebral achromatopsia do not usually make spontaneous comments about color. Questions about color sensation seem quite alien to them. They are unable to name the colors of stimuli presented to them or to perform nonverbal tests of color perception such as sorting or odd-one-out tasks. Although they may remember some semantic associations of color words (e.g., that bananas are yellow), they appear to have no sensation of color or means of remembering any aspects of the nature of color sensation (see, e.g., Heywood & Kentridge [2003] for a recent review). It is, however, well known, that cerebral achromatopsics do see (quite consciously) the border formed between regions of equiluminant color (Heywood et al. 1998). If a red and a green that a patient cannot tell apart are used to construct a red square against a green background, then the patient will effortlessly see the square although they will be unable to explain how the square and its background differ. My colleagues and I (Kentridge et al. 2004) have shown that not only can these patients see such color borders but they can discriminate between borders formed from different colors (and that this discrimination cannot be accounted for on the basis of chromatic contrast salience) even though they cannot see the colors of the adjoining surfaces that form the borders. Again, the discrimination is conscious. The borders somehow look different from one another. The patients cannot, however, explain in any sense how or why the borders differ (they have no accessible representations of color)– they just know that the borders look different. The patient is surely having differing phenomenal experience of these borders, yet these experiences are not cognitively accessible. It is true that they know of the existence of these borders, but subjects in a partial report experiment know of the existence of items they cannot describe. I suggest that this situation, in which sensory representations simply cannot be integrated into the global workspace, provides a better example of phenomenology in the absence of cognitive access than cases in which the integration of representations into the workspace is possible but temporarily unachieved. Phenomenology without conscious access is a form of consciousness without top-down attention doi: 10.1017/S0140525X07002907 Christof Kocha and Naotsugu Tsuchiyab aDivision of Biology, California Institute of Technology, Pasadena, CA 91125; bDivision of Humanities and Social Sciences, Psychology and Neuroscience, California Institute of Technology, Pasadena, CA, 91125. koch.christof@gmail.com http://klab.caltech.edu/~koch/naotsu@gmail.com http://www.emotion.caltech.edu/~naotsu/Site/index.html Abstract: We agree with Block’s basic hypothesis postulating the existence of phenomenal consciousness without cognitive access. We explain such states in terms of consciousness without top-down, endogenous attention and speculate that their correlates may be a coalition of neurons that are consigned to the back of cortex, without access to working memory and planning in frontal cortex. We agree with Block’s hypothesis that phenomenally conscious states may sometimes not be cognitively accessible. Partial report and dual-tasks paradigms show that we have only limited access to some aspects of phenomenal experience. Or, to adopt Block’s earlier language, phenomenal consciousness can occur without access consciousness, a revision of our earlier position (Crick & Koch 1998a; Koch 2004). We argue here that sensory psychology has a long-standing framework to consider such cases, involving attentional selection processes. We recently (Koch & Tsuchiya 2007; Tsuchiya & Koch 2008) summarized the empirical evidence that consciousness and topdown, volitionally controlled endogenous attention are distinct neurobiological processes with distinct functions (see also, Iwasaki 1993; Lamme 2003). In particular, sensitive psychophysical techniques can dissociate these two. Indeed, a range of phenomena exists in which subjects are conscious of certain stimuli attributes without top-down attention. This list includes 509 BEHAVIORAL AND BRAIN SCIENCES (2007) 30:5/6Commentary/Block: Consciousness, accessibility, and the mesh the pop-out target in a visual search task, gist perception of a scene, categorization of peripheral targets when a very intensive cognitive-resource task has to be performed at fixation, and iconic memory. Consider Sperling’s original iconic memory experiment (Sperling 1960) or Landman et al.’s (2003) variant. Subjects report that they clearly, vividly, and consciously see a field of letters or a bunch of bars arranged on a circle. This is also what we experience when we look at such displays. However, it is well known that subjects have only very limited access to the detailed properties of the individual elements, unless top-down attention is directed to a subset of stimuli using appropriately timed cues. Our basic point is that phenomenology without conscious access is an example of consciousness without top-down attention processing, though the converse is not true; that is, not every example of conscious perception in the absence of top-down attention is cognitively non-accessible. For example, the gender of a briefly presented face can be accurately reported even if subjects are engaged in a highly demanding task at the fixation (Reddy et al. 2004). So what is the story at the level of the brain? Decades of electrophysiological recordings in monkeys have proven that the spiking response of neurons in the ventral visual stream (e.g., in areas V4 and IT) representing attended stimuli is boosted at the expense of the response to non-attended items. According to Crick and Koch (1995), this enables these neurons to establish a reciprocal relationship with neurons in the dorsolateral prefrontal cortex and related regions that are involved in working memory and planning (and language in humans), leading to reverberatory neuronal activity that outlasts the initial stimulus duration. Critical to the formation of such a single and integrated coalition of neurons are the long-range axons of pyramidal neurons that project from the back to the front of cortex and their targets in the front that project back to the upper stages of the ventral pathway (possibly involving stages of the thalamus, such as the pulvinar [Crick & Koch 1998b], and the claustrum [Crick & Koch 2005]). The subject now consciously sees these stimuli and can report on their character (e.g., identify the letter [Sperling 1960] or the orientation of the square [Landman et al. 2003]). Furthermore, the subject also has a strong conscious senseof theentirescene(“I seeanarrayofletters”) thatis likewise mediated by a loop that involves the inferior temporal cortex and the frontal lobes half-way across the brain. But what happens to those stimuli that do not benefit from attentional boosting? Depending on the exact circumstances (visual clutter in the scene, contrast, stimulus duration) these stimuli may likewise establish coalitions of neurons, aided by local (i.e., within the cortical area) and semi-local feedback (i.e., feedback projections that remain consigned to visual cortex) loops. However, as these coalitions of neurons lack coordinated support from feedback axons from neurons situated in the prefrontal cortex, thalamus, and claustrum, their firing activity is less vigorous and may decay much more quickly. Yet, aided by the neuronal representation of the entire scene, these weaker and more local coalitions may still be sufficient for some phenomenal percepts. Block cites functional magnetic resonance imaging (fMRI) studies of patients with visuospatial hemi-neglect (Driver & Vuilleumier 2001; Rees et al. 2000; 2002b) that offer evidence of enhanced hemodynamic activity in the fusiform face area contralateral to a face that the patient is not aware of. For Block, this raises the question of whether this is likewise an example of phenomenal consciousness without cognitive access. We answer this question clearly in the negative. First, one should trust the f irst-person perspective: That is, in the absence of compelling, empirical evidence to the contrary (such as Anton’s blindness, also known as hysterical blindness; Sackeim et al. 1979), if the subjectdeniesanyphenomenalexperience,thisshouldbeaccepted as a brute fact. If we take the existence of mere recurrent, strong neuronal activation as evidence for consciousness, why not argue 510 that the spinal cord or the enteric nervous system is conscious but is not telling me (Fearing 1970)? Second, the relationship between neuronal firing activity and the associated hemodynamic BOLD response is a very complex one. In particular, there are well-documented cases where a vigorous fMRI signal is observed in the absence of any spiking from the principal neurons in that area (Harrison et al. 2002; Logothetis 2003; Logothetis & Wandell 2004; Mathiesen et al. 1998). Synaptic activity is a much larger driver of hemodynamic activity than are action potentials. Therefore, a much more cautious reading of these studies is that they demonstrate synaptic input into the fusiform face area in these patients; however, whether or not this input is vigorous enough to establish a sustained coalition of neurons is totally up in the air and requires further investigations. In conclusion, the quiddity of the neuronal correlates of conscious access are long-range loops between the back and the front of cortex and its associated satellites (thalamus, basal ganglia and claustrum), enabled by top-down attention. Without this amplification step, most coalitions in the back are fated to die; however, given the right conditions, a few may survive and may be consciously experienced by the subject. Yet, as the informational content of these coalitions are not accessible to working memory and planning circuits in the front, the subject cannot consciously access the detailed stimulus attributes. Our explanation provides a plausible account of how phenomenal consciousness can occur without cognitive access. Partial awareness and the illusion of phenomenal consciousness doi: 10.1017/S0140525X07002919 Sid Kouider, Vincent de Gardelle, and Emmanuel Dupoux Laboratoire de Sciences Cognitives et Psycholinguistique, Ecole Normale Supe´rieure, 75005 Paris, France, and CNRS/EHESS/DEC-ENS,75005Paris, France. sid.kouider@ens.fr gardelle@ens.fr dupoux@lscp.echess.fr www.lscp.net Abstract: The dissociation Block provides between phenomenal and access consciousness (P-consciousness and A-consciousness) captures much of our intuition about conscious experience. However, it raises a major methodological puzzle, and is not uniquely supported by the empirical evidence. We provide an alternative interpretation based on the notion of levels of representation and partial awareness. In his target article, Ned Block is dealing with a difficult problem: how to empirically demonstrate that phenomenal consciousness (hereafter P-consciousness) is dissociable from access consciousness (hereafter A-consciousness). An a priori argument in favor of this dissociation is the common intuition that the representational content of phenomenal experience is much richer than the limited content we can access at a given time. In Block’s words, “phenomenology overflows cognitive accessibility” (sect. 8, para. 6). This intuition is so strong that it appears very easy, at f irst glance, to show how much richer P-consciousness is, compared with A-consciousness. However, providing an empirical demonstration of this dissociation leads to a major methodological difficulty: any measure of consciousness seems inevitably to require the involvement of A-consciousness. From there on, it seems impossible to show evidence for P-consciousness without A-consciousness. This methodological puzzle arises also in dissociating consciousness and top-down attention. Demonstrating consciousness without attention seems impossible for similar reasons: To assess consciousness of the stimulus, one needs to direct the subject’s attention on the stimulus! Although there is converging evidence that attention can affect both conscious and unconscious BEHAVIORAL AND BRAIN SCIENCES (2007) 30:5/6Commentary/Block: Consciousness, accessibility, and the mesh perception, the reverse dissociation involving the possibility of consciousness without attention remains highly debated (see Dehaene et al. 2006; Koch & Tsuchiya 2007). Block acknowledges this methodological issue and proposes to take the set of evidence at hand as a whole and see whether it points towards the researched dissociation. Using, among others, examples from perception of unattended objects (e.g., attentional blink, change blindness), and from partial report Sperling-like experiments, Block assumes that weshould adopt the A- versusP-consciousness dissociation and look for its respective neural bases. Unfortunately, the evidence Block is using cannot unequivocally prove his theory. Furthermore, we think that the empirical data reviewed in his target article do not provide more support for his accounts over alternative and crucially simpler explanations. In the empirical phenomena that Block is using, one can distinguish two types of situations: those involving partial access and those involving undetectable stimuli. The first type of situation involves stimuli that are visible but unattended– and importantly, not even detected– implying the absence of any conscious access. This is usually the case during attentional blink and inattentional blindness experiments. Block uses the fact that the stimulus is supra-threshold (it can be reported when attention is drawn towards it) to argue for Pwithout A-consciousness. But there is no evidence for this claim, since subjects do not even detect the unattended stimulus. Moreover, Block’s claim that participants forget their phenomenal episode appears impossible to test, because any probe would modify their attention and hence make the stimulus consciously accessible (Kouider et al. 2007). Inthesecondtypeofsituation,bycontrast,subjectsareawareof “some” information: stimuli are presented very briefly and/or in a degraded fashion, such that they are not fully visible but not subliminal either. Block assumes that this situation gives rise to P- without A-consciousness. We propose instead that what happens here is rather a form of partial awareness in the absence of full awareness. Partial awareness reflects the situation where subjects have transient access to lower but not higher levels of representation. For instance, visual word recognition implies the processing of several hierarchically organized levels (e.g., fragments, letters, whole word). With degraded presentation conditions, lower levels can be accessed (e.g., fragments/letters) while higher levels cannot (whole word). Still, subjects can use such partial information in conjunction with context/expectations to make hypotheses about the representational content at higher levels of processing (Kouider & Dupoux 2004). Under this perspective, Block’s richness of phenomenal experience can be reinterpreted as the transient activation of a large quantity of degraded low-level information. In such partial awareness conditions, the available information is quantitatively rich but qualitatively poor. This hypothesis allows us to construe the Sperling phenomenon (see Sperling 1960) as resulting from partial awareness: subjects have a transient and degraded access to fragments of all the letters in the grid. As subjects are not expecting anything other than letters, fragments are used to reconstruct as many letters as possible. Due to mnemonic decay and attentional overf low, subjects are able to reconstruct at most about four letters. Crucially, the unreported items are never identified as letters per se and remain coded as unidentified letter fragments. A similar situation of partial without full access is the McConkie experiment (McConkie & Rayner 1975) in which subjects see “letter-like” fragments in the periphery and infer that these are real letters. In our previous work (Kouider & Dupoux 2004) we have extended this phenomenon to a dissociation between the letter and the word level. We have induced subjects to access some but not all letters of a real or false color word (GREEN or GENER). We found that both the real and false color words are identified and treated as real words, as assessed by both subjective reports and the magnitude of the Stroop effect. Animportant question for future research will be to characterize whether such reconstruction processes imply metacognitive/ inferential interpretations or rather more direct perceptual illusions. Block acknowledges that McConkie’s experiments involve a reconstruction process (what Block labels “cognitive illusions”). However, Block assumes that subjects in Sperlinglike experiments are not reconstructing the visual scene but genuinely experiencing the whole set of items. To justify this special treatment, Block argues that the Sperling phenomenon is somehow mandatory and does not require subjects to explicitly report the stimuli, implying that it is a perceptual rather than a metacognitive reinterpretation. Yet, these statements remain highly speculative, as none of them has been empirically demonstrated. For us, it is highly probable that Sperling-like paradigms also lead to the “experience” of letters even when the uncued items consist of false letters. Of course, disentangling this issue requires further empirical research. All these remarks point towards the same direction: Including a typology in terms of levels of representation during conscious access, along with the associated notion of partial awareness, provides a unified description of the empirical evidence at hands. In particular, this account offers more explicit specifications of the functional mechanisms leading to conscious perception.