Preserved aspects of consciousness in disorders of consciousness: A review and conceptual analysis Jakob Hohwy* & Emma Fox Philosophy Department, Monash University Abstract. The last decade has seen impressive and intriguing advances in the exploration of vestiges of consciousness in patients with disorders of consciousness (DOC). Consciousness is an extremely complex area of research so it is difficult to provide unequivocal interpretations of these new findings from DOC-studies. This review therefore provides a conceptual analysis of a series of key studies in this area of research. The main upshot is that different studies of preserved consciousness in DOC are best seen as targeting different aspects of consciousness (such as arousal, awareness, volition, global availability, access, attention, self, and first-person perspective) rather than a unitary property of being conscious. These different aspects of consciousness thus seem to dissociate and fractionate in DOC. This provides a deeper understanding of the nature and structure of consciousness in general, and it provides suggestions for future research in DOC and the science of consciousness. Further methodological, conceptual and ethical issues are discussed. Contents 1. Introduction .................................................................................................................................................................... 2  2. Conceptions and aspects of consciousness ..................................................................................................................... 2  2.1. Arousal and awareness dimensions ............................................................................................................................................. 3  2.2. Further aspects of consciousness ................................................................................................................................................ 4 

3. Disorders of consciousness and the Other Minds Problem ............................................................................................. 5  4. Proposed tests of consciousness in DOC ........................................................................................................................ 6  4.1. Awareness ................................................................................................................................................................................... 6  4.2. Volition ........................................................................................................................................................................................ 7  4.3. Endogenous attention as a sign of volition .................................................................................................................................. 7  4.4. Volitional signs utilised as covert communication ....................................................................................................................... 8  4.5. Volition and endogenous attention as a sign of levels of cognitive function ............................................................................... 8  4.6. Endogenous attention as a sign of global workspace .................................................................................................................. 9  4.7. Non‐volitional aspects of consciousness? .................................................................................................................................. 10 

5. Potential tests for preserved aspects of consciousness.................................................................................................. 10  5.1. Exogenous attention grabbing ................................................................................................................................................... 11  5.2. The self, self‐related mental states, and intrinsic awareness .................................................................................................... 11  5.3. Perceptual awareness and binocular rivalry .............................................................................................................................. 13  5.4. Conscious access and masking ................................................................................................................................................... 14  5.5. First person perspective and multisensory integration ............................................................................................................. 15 

6. Discussion ..................................................................................................................................................................... 17  6.1. Probing different, potentially dissociable, aspects of consciousness via different tests ........................................................... 17  6.2. Methodological issues ............................................................................................................................................................... 17  6.3. Absence of evidence and evidence of absence ......................................................................................................................... 18  6.4. Variations on the argument from analogy to the existence of other minds .............................................................................. 18  6.5. What type of consciousness comes with volition and covert communication? ........................................................................ 19  6.6. Ethical issues .............................................................................................................................................................................. 20 

7. Conclusion .................................................................................................................................................................... 21 

*

Corresponding author. [email protected] This paper is accepted for publication in Journal of Consciousness Studies.

Hohwy & Fox, forthcoming, Journal of Consciousness Studies

1. Introduction Disorders of consciousness (DOC) include the minimally conscious state (MCS), the vegetative state (VS) and coma. They are characterised by low scores on behaviourally based tests such as the Glasgow Coma Scale (GCS), Full Outline of UnResponsiveness scale (FOUR) (Posner, Saper et al. 2007) and JFK Coma Recovery Scale – Revised (CRS-R) (Giacino, Kalmar et al. 2004). Patients without the appropriate behavioural signs have therefore generally been considered unconscious. Recent fMRI evidence, showing activity in VS patients comparable to that of healthy individuals in imagery tasks, has dramatically opened the question of whether there is preserved consciousness in spite of the absence of outward behavioural signs (Owen 2006; Monti, Vanhaudenhuyse et al. 2010). This unavoidably undermines trust in the purely behavioural tests. The very concept of consciousness is exceedingly hard to characterise explicitly and it is even harder to give it an unequivocal and operationalisable definition. It is very common for the concept to differ according to the discipline utilising it. Examples include clinical neurology, cognitive neuroscience, the science of consciousness, as well as philosophy. One problem is that consciousness is often treated as a fairly unitary property whereas the notion is in fact used to refer to a cluster of different properties. Therefore, it is not immediately obvious how to interpret a finding of ‘being conscious in the vegetative state’. Conscious in what sense? This question is crucial for our scientific understanding of the very nature of consciousness, its disorders, as well as for ethical concerns about end of life decisions. Here, we review and conceptually analyse existing studies of tests of consciousness in DOC, which do not rely on purely outward behavioural signs. The aim is to categorise the aspect of consciousness assessed by each such test. We also discuss further aspects of consciousness and suggest potential tests for them. The upshot from looking at this series of individual studies from a conceptual point of view is that different actual and potential tests quite clearly target quite different aspects of consciousness. This suggests that common definitions of consciousness, for example the common definition in terms of arousal and awareness, will benefit from conceptual refinement. In DOC, consciousness seems to fractionate into different seemingly dissociable aspects. This puts pressure on the common idea that consciousness in principle is a unitary phenomenon. This approach thus helps clarify the question of how to interpret evidence of consciousness in DOC, and it raises as an important research question the issue of how these different aspects of consciousness relate to each other in general. Section 2 sets out the different aspects of consciousness. Section 3 discusses the common inferential tool that must be used in this area of research, the argument from analogy to the existence of other minds. Section 4 discusses actual tests of consciousness in DOC one by one against the list of aspects of consciousness, and Section 5 discusses potential such tests. Section 6 provides a discussion of the implications of this review for our understanding of DOC, for consciousness in general, and briefly touches on ethical issues.

2. Conceptions and aspects of consciousness The mind-body problem in philosophy is about finding a place for consciousness in the natural world (Chalmers 2002). This is a perennial philosophical problem in part because the very concept of consciousness is very difficult to pin down with much precision. As a result of this difficulty, it has become customary to indicate what consciousness is by way of 2

Hohwy & Fox, forthcoming, Journal of Consciousness Studies

examples: consciousness is what it is like in the morning when you wake up after a dreamless sleep; it is what it is like when you feel the hurtfulness of pain, or what it is like when you experience seeing and smelling a red rose, and so on. Underlying this general notion of consciousness as “what-it-is-like” is a range of more specific notions of consciousness, which are however often equally hard to define in a non-circular and operationalisable manner. The following is a non-exhaustive catalogue.

2.1. Arousal and awareness dimensions It is common in clinical neurology (Posner, Saper et al. 2007) and in the neuroscience of disorders of consciousness (Laureys 2005; Giacino, Schnakers et al. 2009) as well as in consciousness science and philosophy (Koch 2004; Bayne 2007; Hohwy 2007) to make a fundamental distinction between arousal and awareness. Arousal is variously explicated in terms of alertness or awakeness, or level of consciousness (Posner, Saper et al. 2007). Arousal is severely diminished in coma but higher in VS and MCS. Behaviourally it is indicated by opening of the eyes and intuitively it is easy to grasp that arousal is what happens upon waking up. It depends on subcortical systems in the brainstem, midbrain and thalamus. Awareness is variously explicated in terms of the contents of consciousness or the representational capacity of consciousness (Posner, Saper et al. 2007). It is absent in coma, assumed absent in VS (though this is currently a topic of debate), and assumed to be varying in MCS. It presumably depends on cortical and thalamocortical connectivity (Baars 1996; Koch 2004). It is typically assessed behaviourally by command following and observation of non-reflex behaviour. Awareness is often specified in terms of internal or external perception either of self or of the environment, respectively. Notice that neither of these two terms would be of much help to someone who did not already grasp the problematic notion of consciousness. In this context ‘being aroused’ just seems to mean ‘being conscious’ as does ‘being awake’ or ‘being alert’. The same applies to ‘being aware’ and ‘having conscious representational contents’. Therefore it is difficult to flesh these notions out in anything but behavioural terms. Each of these notions are themselves complex, related to the other notions and subject to neurophysiological research. For example, ‘awakeness’ can be characterised in terms of ‘arousal/responsiveness, circadian rhythms, sleep cycle, and homeostasis’ (Bekinschtein, Cologan et al. 2009: 172). In this characterisation, of ‘awakeness’, ‘arousal’ has connotations with ‘consciousness’ and the other terms are better understood as standard concomitants of consciousness rather than explications of the concept itself. Arousal is generally supposed to subserve or interact with awareness such that individuals’ conscious states lie in some two-dimensional space defined by arousal and awareness with brain death and coma low in both arousal and awareness and normal conscious state high in both dimensions (Laureys 2005; Mormann and Koch 2007; Demertzi, Vanhaudenhuyse et al. 2008). Anomalies include REM sleep which is low in arousal and higher in awareness, and sleep walking, complex partial seizures (Englot, Blumenfeld et al. 2009), and absence seizures which are supposed to be low on awareness though higher on arousal. Absence seizures exemplify some of the problems with this two-dimensional construal of consciousness since sufferers can navigate their environment and so must be able to represent it; awareness cannot therefore be simply the ability to perceive.

3

Hohwy & Fox, forthcoming, Journal of Consciousness Studies

Finally, though it is clear that the notions of arousal and awareness are intended to allow for gradation, which can be captured to some extent by behavioural ratings, it is far from clear how the notion of levels of arousal or awareness would translate to the subjective experience of being at a decreased or increased level of consciousness.

2.2. Further aspects of consciousness It is common to appeal to other aspects of consciousness in attempts to further characterise the notions of arousal and awareness. Awareness, as subserved by arousal, is often elucidated in terms of having the ability to attend to environmental contingencies. It is controversial however whether, and if so how, attention and consciousness may relate (Koch and Tsuchiya 2007). Sometimes the level of awareness is quantified in terms of the degree of information integration (Tononi and Koch 2008; Boly, Massimini et al. 2009), connectivity, or degree of global access to cognitive consumer systems such as memory, executive functioning and verbal report (Baars 1996; Dehaene and Naccache 2001). It is however controversial whether and how consciousness relates to such notions of global access (Block 2001; Block 2005). It is not uncommon to see consciousness understood in terms of the self. This appeal to the self however comes in various flavours. A species of awareness is conscious representation of self-related contents, such as recognition of one’s own name or mirrored face. A deeper and more obscure notion is that of self-consciousness, being aware of oneself as having a self. And sometimes ‘being conscious’ seems to be used to mean ‘having a self’ or more broadly still, being a person. The notion of the self is however also notoriously difficult to pin down. On one influential philosophical account, having a self is a matter of maintaining a type of self-model, a representationally transparent model of the current state of the biological organism (Metzinger 2009). Such a self-model would encompass a first-person perspective, which is often seen as a defining aspect of the subjective aspect of consciousness – it is hard to envisage someone being consciousness without a first-person perspective. In the background of the preoccupation with the self lie ideas of volitional creatures or persons with desires to act on preferences for the future. These are all philosophically troublesome notions. For example, it is controversial how much self-reflection it takes to have volition in a rich person-relevant sense rather than displaying mere stimulus-driven or reflexive behaviour (where is the dividing line as one moves down from human persons towards, say, non-person ants?). This notion is however crucial in many ethical discussions concerning DOC (Hohwy and Reutens 2009; Levy, Savulescu et al. 2009). These are all aspects of consciousness that have exercised philosophers. There is no agreement about if and how they may relate or come apart. There is also little agreement about how they relate to a core idea about “what-it-is-like”, namely qualitative or phenomenal aspects of consciousness. This aspect, which sometimes gets the label of art ‘qualia’ is at the core of the mind-body problem. Some philosophers see this as the true aim of any consciousness research, and it has been suggested that it can at least in principle come apart from any other notion of consciousness and any kind of introspective access we might have to such purely phenomenal states (Block 2008), making it exceedingly hard to operationalise for empirical study. This should be recognised as a possibility but it has little bearing on the difference between DOC and normal states of consciousness since the problematic notion of unaccessed, purely phenomenal consciousness is equally vivid in both types of cases. Therefore, this review assumes, together with most of consciousness science, that phenomenal aspects of consciousness goes together with at least some of the aspects of consciousness, listed above, to which people have cognitive and introspective access and which they can commonly report to each other. 4

Hohwy & Fox, forthcoming, Journal of Consciousness Studies

3. Disorders of consciousness and the Other Minds Problem For each of the many different aspects of consciousness it is possible to ask meaningfully whether it is absent or preserved to some degree in DOC. Very little is known about how the aspects of consciousness relate and depend on each other so preservation of each aspect could, at least in principle, be independent of the presence or absence of any other aspect. This suggests that a diversified research strategy on which preservation of distinct aspects of consciousness is explored separately. Ideally, one would execute such a diversified research strategy by appeal to the gold standard for ascertaining the presence of consciousness, namely behavioural measures such as introspective report. But, of course, these are compromised in DOC. DOC thus presents as a corollary to the venerable Other Minds Problem that has exercised philosophers for centuries: I can know my own mind in a direct, privileged manner but can only attempt to know the minds of other creatures indirectly via observation of their behaviour. In DOC there is no reliable behaviour to observe so how can inference about mental states in the patients even begin? Though this philosophical ancestry is not always explicitly recognised, there is acute awareness of this problem in the scientific field (Coleman, Bekinschtein et al. 2009; Monti, Coleman et al. 2009; Overgaard 2009; Stins and Laureys 2009; Zeman and Steven Laureys 2009). The standard philosophical solution to the Other Minds problem appeals to an argument from analogy: if I have a certain conscious state when I have a certain type of behaviour, then I may conclude (defeasibly) that other people have similar conscious states when they display similar behaviour (Mill 1865). More sophisticated versions are available, which employ more broad-ranging inference to the best explanation, and which allow appeal to biological commonalities within a population (Hill 1991; Hyslop 2005; Shea and Bayne 2010). The current research on consciousness in DOC may, as we seek to demonstrate below, benefit from using the different versions of the argument from analogy as backdrop and organising epistemic principle. In DOC, behavioural signs of conscious states are absent or impaired. This presents a hurdle to employing the argument from analogy in DOC cases. The breakthrough, which promises to overcome this hurdle, are studies where neurophysiological evidence, such as the Blood Oxygen Level Dependant (BOLD) response elicited in fMRI imaging, is found to stand proxy for normal kinds overt behaviour, such as verbal report. These proxies are then utilised in a version of the argument from analogy. It is therefore potentially promising to pursue the diversified research strategy by obtaining covert behavioural signs in the form of different types of neurophysiological and psychophysical evidence. Several research groups now explore such classes of evidence in DOC (Owen 2006; Coleman, Bekinschtein et al. 2009; Monti, Coleman et al. 2009). There is increasing awareness that different types of data may pertain to different aspects of consciousness, rather than to an amorphous notion of being conscious as such, though the consequences of this for the study of consciousness is rarely discussed explicitly (Owen 2008; Bekinschtein, Dehaene et al. 2009; Gaillard, Dehaene et al. 2009; Monti, Coleman et al. 2009; Soddu, Boly et al. 2009). Here, we bring a more philosophical, conceptual analysis of the aspects of consciousness to bear on the study of DOC. This aids our understanding of how this rapidly growing and intriguing research field relates to the more fine-grained aspects of consciousness, it demonstrates how quite different and potentially dissociable aspects of consciousness could be preserved in DOC, and it throws light on the different forms the argument from analogy 5

Hohwy & Fox, forthcoming, Journal of Consciousness Studies

can take. We first align some existing studies of DOC with different aspects of consciousness (see Table 1) and then suggest some other types of studies as having some potential as tests of further aspects of consciousness (see Table 2).

4. Proposed tests of consciousness in DOC Two studies in particular have catapulted the issue of consciousness in DOC into prominence. In one PET-study, a comparison of brain activity in healthy and VS patients revealed that thalamocortical connectivity seems impaired in VS (Laureys, Faymonville et al. 2000). In a later fMRI-study, VS patients and healthy volunteers were shown to have very similar activation patterns during visual imagery tasks (Owen 2006; see also Monti, Vanhaudenhuyse et al. 2010). Though there is much debate about consciousness science at large (Block 2001; Dennett 2001) and the neural correlates of consciousness (Hohwy 2007) there is no serious doubt that these studies tap into important aspects of consciousness and genuinely raise the question of the extent to which consciousness may be preserved in DOC. In the following section we review these and other studies and find evidence that they do indeed relate to different aspects of consciousness.

4.1. Awareness In Owen’s 2006 study (Owen 2006; see also Monti, Vanhaudenhuyse et al. 2010), a VS patient and healthy controls were asked to visually imagine either playing tennis or navigating through their house. Using fMRI, comparable activity was detected in both groups (e.g., in the supplementary motor area (SMA) for tennis imagery). Which aspects of consciousness does this tap into? The VS patient and controls were assumed to already be somewhat aroused so initially the expectation would be that the brain activity reveals increased levels of awareness or conscious representation. It is important to note, however, that the study only addresses awareness indirectly: activity in the SMA is not known to contribute directly to the constitution of conscious content. That is, though SMA activity can be observed during conscious motor imagery it is not established whether it belongs to the minimally sufficient conditions (Chalmers 2000) for specifically conscious motor imagery. Neither is it known what other brain areas, which could be damaged in DOC, belong to these minimally sufficient conditions. It is therefore possible that there can be SMA activity without conscious motor imagery. The more principled point in the background is that there is a difference between the neural correlates of consciousness (i.e., the minimally sufficient conditions for awareness) and what might be called the causal concomitants of consciousness, which is a potentially much broader set of neural activations that co-occur with consciousness but can be causally upstream or downstream from the activity that makes the state conscious. At best, one can use this kind of study in an attempt to make a case that it is somewhat likely that a VS patient, who has such an overlap in activity with patently conscious healthy controls, is also conscious. However, for the reasons just given, this is a rather loose argument from analogy: the overlap in activation is not known to be a similarity that is sufficiently relevant for the target property of being conscious of the motor imagery. More research is thus needed to make the case for awareness more confidently on the basis of this type of study. For present purposes, the main point is that this study targets, albeit indirectly, the preservation of awareness (in the form of having representational content in the form of conscious imagery) in DOC.

6

Hohwy & Fox, forthcoming, Journal of Consciousness Studies

4.2. Volition As seen, one immediate objection to Owen et al.’s (2006) tennis imagery study is that it remains possible that the evoked activity is merely unconscious and automatic, and as such fails to engage any aspect of consciousness (Naccache 2006). This objection is acknowledged implicitly in the 2006 paper and later met explicitly with an appeal to another aspect of consciousness, namely volition. Owen and others (Boly 2007; Owen 2008) note that the SMA activity was sustained for 30 seconds and time-locked to requests to imagine tennis and to then stop imagining tennis. Owen makes the case that the activity therefore reflects volitional and intentional compliance with the instructions. In so far as true volition is somehow necessarily associated with being conscious the implication is that this study more directly targets volitional aspects of consciousness than awareness as such. It is an open research question as to what extent such conscious volition depends on conscious awareness (i.e., representation) and as such it is theoretically possible that there was conscious volition without conscious content. However, it seems unlikely that the patient could have failed to represent the content of the verbal instructions (e.g., “imagine playing tennis”), and feed it to volitional processing, given she differentially recapitulated the patterns of neural activity seen in controls for tennis and navigation imagery. What the evidence tracks here is thus the occurrence of conscious mental agency, that is, an individual volitionally enacting mental operations. This stance on volition is independent of worries about the extent to which she is also aware of the tennis imagery itself. (The Discussion section below further considers the relation between volition and consciousness).

4.3. Endogenous attention as a sign of volition Attention is divided in to two types. In exogenous attention, attention is grabbed by salient environmental or internal states of affairs. In endogenous attention, attention is voluntarily directed at states of affairs. Evidence for endogenous attention would therefore be evidence of volition. In one study healthy controls and a MCS patient were presented with series of words; in one condition they were asked to attend to a specific word and in the other condition they listened passively (Monti, Coleman et al. 2009). fMRI indicated very similar activity in both groups for the active vs. passive listening tasks, and the activity was time-locked to the response blocks. The best explanation of this seems to be that the patient voluntarily followed the instruction to attend to the word. A similar logic was employed in a study of evoked-related potentials in MCS and VS patients (Schnakers, Perrin et al. 2008). Patients were presented with series of names, including sometimes their own name. In the active condition they were instructed to count their own name or another name. MCS patients reliably displayed a normal P300 response (a positive deflection in voltage in the EEG after about 300 ms, typically associated with decisionmaking) in the active counting condition. This is again evidence of voluntary task compliance in MCS, though here it is not time locked to the task requirements. No P300 was found in VS. In a different study, healthy controls, and MCS and VS patients were presented with locally and globally deviant auditory stimuli (Bekinschtein, Dehaene et al. 2009). In one condition participants were asked to count the occurrences of global pattern deviations. Controls reliably displayed a P300 complex ERP but only when they consciously attended (endogenously) to the pattern. Most MCS patients had a similar response (when asked to attend to the pattern) but the VS patients did not. As part of a very impressive multimodal approach to assessment of DOC, patients are first tested for visual perception of faces and houses via fMRI activity recorded from the fusiform 7

Hohwy & Fox, forthcoming, Journal of Consciousness Studies

face area and parahippocampal place area. If they manifest this kind of response they are then shown superimposed images of houses and faces and are asked to attend to one or the other and again scanned for differential activity in the relevant areas of the brain (Coleman, Bekinschtein et al. 2009). If patients manifest fusiform activity when asked to attend to the face rather than the house, then a reasonable interpretation is that they consciously attend to faces thus displaying endogenous attention and therefore volition. This type of paradigm targets volitional consciousness as well as the ability for endogenous attention. Whereas the evidence for volition is quite strong, the evidence for attention is good but somewhat more uncertain since it is unclear how well the patients could attend. For example, it is possible that the effort at volition drives the increased activity in the fusiform face area but that attention to the face stimulus itself failed.

4.4. Volitional signs utilised as covert communication If indeed a patient manifests signs of volition, then these very signs can be exploited as communicative signs. Explicit conventions can tie different mental volitional acts to different meanings (e.g., imagine tennis for “yes”, imagine navigating your house for “no”), such that patients can communicate with carers using covert behaviour such as BOLD responses or ERP (Sorger, Dahmen et al. 2009). This strategy would be immensely useful for clinical, ethical and quality of life purposes. In and of itself it does not reveal any more aspects of consciousness specifically, in addition to the demonstration of volition. But it can of course be used to probe directly which aspects of consciousness are preserved (e.g., by asking whether touch is consciously felt, or whether patients can recognise themselves etc). There is growing hope (or perhaps fear) that at least some VS patients will be able to communicate covertly in this way. In a recent study of 54 VS and MCS patients, 5 were able to replicate the 2006-tennis imagery findings. One of these patients (VS) was further tested and found able to use this wilful imagery as a means of communication (Monti, Vanhaudenhuyse et al. 2010). This is very startling evidence because it provides some reason to believe that there is conscious volition in about 10% of patients and the ability for covert communication in at least 2%. This evidence is consistent with the possibility that conscious volition can dissociate from the ability to covertly communicate. This means that patients could volitionally follow instructions for mental activity but be unable to use this ability for communicative purposes. Imagine, as a possible case, a DOC patient with the equivalent of Broca’s aphasia. She would be able to understand and volitionally follow instructions to imagine tennis but would be unable to utilise imagining tennis as a conventional sign of “yes”. (Majerus, Bruno et al. 2009 discusses aphasia as a problem for overt behavioural assessment in DOC, what we suggest here is that aphasia could also be a problem for covert communicative protocols in DOC). It is also possible that the communicative efforts are there but that the fMRI paradigm is not always able to detect such efforts.

4.5. Volition and endogenous attention as a sign of levels of cognitive function A number of the paradigms discussed above explore the preservation of volition and endogenous attention in DOC: volitional instruction following while imagining tennis; attention to faces rather than houses for unchanging superimposed visual stimuli; attention to (and counting of) global rather than local violations of auditory patterns; attention to (and counting of) occurrences of auditory target words and names; attention to occurrences of own names. 8

Hohwy & Fox, forthcoming, Journal of Consciousness Studies

Each of these paradigms can yield evidence of preserved volition, in the form of endogenous attention. But they also each recruit different levels of cognitive function, which are in principle dissociable. For example, a VS patient may fail to endogenously attend to violations of global auditory patterns but be able to volitionally follow instructions for visual imagery. Such a difference could reflect failure of recruitment of higher, executive systems for the task involving violations of auditory patterns. Similarly, it is not unlikely that a MCS patient could manifest volition when asked to count occurrences of her own name but fail to volitionally imagine playing tennis (see (Schnakers, Perrin et al. 2008; Bekinschtein, Dehaene et al. 2009) for relevant studies; see (Coleman, Bekinschtein et al. 2009) for a multimodal approach that tries to address issues of level of content processing). Preservation of volition and its concomitant aspects of consciousness is therefore a complex issue in itself. Dissociations can go in a number of different directions: the ability for volition may be preserved or impaired, or it may be preserved but lacking appropriate awareness of representational content to operate on given a certain type of request, or it may be lacking even though there is awareness of a range of representational content. (Volition and cognition is dealt with further in the Discussion section below).

4.6. Endogenous attention as a sign of global workspace It is not unreasonable to think that conscious awareness requires that representational content occur in the global workspace, such that content is broadcast to a range of otherwise unconscious brain processes embodying cognitive consumer systems (Baars 1996; Dehaene and Naccache 2001). Therefore tasks presented to DOC patients that engage something like a global workspace are of special interest. Monti et al. (Monti, Coleman et al. 2009) asked healthy participants and one MCS patient to count occurrences of certain words in series of auditory words. fMRI revealed very similar results in the two groups. This suggests preservation of at least a threefold process: ability to perceive heard words, ability to adopt and maintain the instruction (or “mind-set”), and ability to make the contents available to be counted (and perhaps reported). It is possible that the patient tried to count but failed, perhaps because conscious contents of sufficient quality to operate on were not available. However, given the activation pattern in this condition was very similar to that seen in healthy participants, this does not seem likely (of course, the counting may not have been correct). This study involves at least part of the global workspace and as such would be evidence for conscious awareness of representational content, such as heard words in MCS. A similar line of reasoning is available for processing of global violations of auditory patterns (Bekinschtein, Dehaene et al. 2009). Notice that here we have an indirect argument from analogy, which rests not only on similarity in BOLD signal across communicative and non-communicative groups, but also on a theory of consciousness. Though it is reasonable to theorise that global availability is sufficient for (without being constitutive of) consciousness, it is not known whether the neural correlates of the global workspace are identical to the neural correlates of consciousness (for discussion, see Block 2005; Block 2008). A related theory-driven approach, which is more speculative but very intriguing, rests on the notion of information integration (Tononi 2005). This is a quantity derived from information theory that is sensitive both to the informativeness (i.e., ability to discriminate among alternatives) of a given state and its degree of integration throughout the system. If level of consciousness coheres with level of information integration, then the conscious brain should be characterised by highly informative but well-integrated states whereas the unconscious brain should have non-informative islands of disintegrated activity. In a combined EEG/TMS 9

Hohwy & Fox, forthcoming, Journal of Consciousness Studies

study, awake and non-REM sleep participants showed just this kind of difference in the propagation of brain activity after single pulse TMS. This type of paradigm has therefore been suggested as a potential bedside test of DOC (Massimini, Boly et al. 2009). Though the theory is potentially groundbreaking it is difficult at this stage to say which, if any, aspect of consciousness this test would probe. After all, information integration is something that an unthinking machine could plausibly possess. An inference to consciousness in DOC based on this test would still be an argument from analogy, given that it begins with ascertaining neural activity from awake participants, but most of the evidential weight in interpreting this activity comes from the theory rather than introspective reports or direct correlation of activity with conscious states.

4.7. Non‐volitional aspects of consciousness? The impetus for many of the studies discussed thus far is the insight that consciousness can dissociate from the ability for overt verbal or otherwise behavioural communication. This is why it is important to investigate preservation of volition in DOC, the cognitive abilities that may be needed for a certain domain of volition, as well as the ability to exploit volition to establish conventions of covert communication. But it is likely that conscious volition– the ability for mental agency such as wilfully imagining playing tennis – can dissociate from other non-volitional aspects of consciousness. It is therefore an important and live research question whether there are preserved aspects of consciousness in the absence of any signs of volition. This issue was evident already in the paradigm-changing tennis imagery study by Owen and colleagues (Owen 2006). The first question that arose was if the patient had preserved awareness, irrespective of volition. This type of question is extremely difficult to resolve with much certainty because it stretches the sources of evidence one appeals to in the argument from analogy beyond the normal behavioural and communicative sources. The argument must now proceed more indirectly: healthy subjects have and are able to report conscious experience C which is correlated with neurophysiological evidence N, therefore noncommunicative DOC patients with N also have experience C. The strength of this inductive inference depends on the strength of the correlation between C and N in healthy individuals and on the generalisability from healthy to damaged brains. The problem with using the tennis imagery finding as evidence of non-volitional awareness of imagery is that the correlation between conscious imagery and just SMA activity is very weak (for discussion, see Hohwy 2009). For an inference to a non-volitional aspect of consciousness to be strong the evidence would need to rest on a body of studies that have come close to revealing the neural correlates – the minimally sufficient neural conditions (Chalmers 2000) – for a certain kind of conscious state. As it stands, the SMA activity can at best be said to belong to the broader notion of upstream or downstream causal concomitants of consciousness.

5. Potential tests for preserved aspects of consciousness At the outset , we listed aspects of consciousness that could be preserved in DOC. What we have just seen is that, for some of the aspects on this list, tests have actually been proposed. We discussed the structure and limitations of these tests and found that they do indeed seem to tap into different aspects of consciousness. We now move to a more speculative part where we consider potential, future paradigms for probing some of the further aspects of consciousness. 10

Hohwy & Fox, forthcoming, Journal of Consciousness Studies

5.1. Exogenous attention grabbing Exogenous attention occurs when attention is grabbed by environmental or internal states of affairs. It is debated whether attention can occur in the absence of awareness of the attendedto states of affairs (Koch and Tsuchiya 2007). Without recourse to communication it is hard to be very confident that attention was or was not accompanied by consciousness. Even though the brain can respond preferentially to apparently salient stimuli, there is reason to think some such responses are not conscious. A local deviation of an auditory pattern can elicit mismatch negativity (MMN; a component of ERP for unexpected stimuli) in VS and MCS but it does so in coma and for inattentive healthy individuals too (e.g., Fischer, Morlet et al. 1999). Conversely, global violations of auditory patterns only elicit a response in healthy individuals under conditions of endogenous attention (Bekinschtein, Dehaene et al. 2009). In short, what this suggests is that exogenous attention grabbing could be a useful tool for probing non-volitional consciousness but only if there is prior reason to believe that the attention comes with conscious awareness of the object of attention. Here it would be useful to have a robust neural correlate of conscious (as opposed to unconscious) exogenous attention in healthy subjects, something we currently do not have.

5.2. The self, self‐related mental states, and intrinsic awareness The notion of consciousness is often related to the idea of the self, often under the aegis of self-consciousness, as if the self was a kind of object the mind could be directed at. There is much discussion in philosophy and beyond about the nature, and existence of, the self (Metzinger 2009). As Hume pointed out, it is difficult to pin down the self in introspection: ‘For my part, when I enter most intimately into what I call myself, I always stumble on some particular perception or other, of heat or cold, light or shade, love or hatred, pain or pleasure. I never can catch myself at any time without a perception, and never can observe any thing but the perception.’ (Hume 1739-40: Pt 4; Sect 6). Accordingly, there is no obvious test of preservation of the self in DOC. It is much less controversial that awareness sometimes is directed at self-related mental states rather than at mental states relating to the external world. This is acknowledged in core definitions of the notion of awareness (‘Consciousness is the full awareness of self and one’s relationship to the environment’ (Posner, Saper et al. 2007: 5). Self-related awareness can be as simple as recognising oneself in the mirror or the sound of one’s own name (Perrin, Schnakers et al. 2006; Laureys, Perrion et al. 2007). Or it can be more involved such as when episodic memory, concerning the individual’s own experiences, are utilised in “mental time travel”, envisioning a future for oneself, considering what to do, and so on. It would be too hasty to label neural systems underpinning such self-related cognition “the self”. But the ability to consider oneself as a unique, experiencing agent in the world seems an important aspect of consciousness, which at least in principle can come apart from the ability to merely have experiences of the world. It would also be an ethically important ability, given that some ethical views assign personhood only to creatures with conscious future-directed desires (see Discussion). It seems very likely that there are a number of different coherent networks in the brain (Greicius, Krasnow et al. 2003), some of which are extrinsic, directed at the outside world, and some of which are intrinsic, directed at the “internal” world. The default mode network (Raichle 2001), tends to decrease activity during attention demanding tasks, it returns to baseline during rest, and tends to increase activity during self-related tasks such as envisioning one’s future (Gusnard 2005; Buckner, Andrews-Hanna et al. 2008). There is 11

Hohwy & Fox, forthcoming, Journal of Consciousness Studies

evidence that activity in the default mode network during rest increases from coma over VS and MCS to the fully conscious state (Boly, Phillips et al. 2008). Damage causing hypoactivity in the default mode network would, in theory, lead to a state of attentive immersion in external states of affairs – one would ‘lose oneself’ in the world (Gusnard 2005; Soddu, Boly et al. 2009). Conversely, damage to attention networks, but with the default mode network intact, would lead to detachment from the external world and perhaps some kind of mainly self-related absorption (Soddu, Boly et al. 2009). Conceivably, in different cases of DOC such differential damage to the various networks could occur. This could in theory result in quite different aspects of preserved consciousness. What it is like to be conscious just of external states of affairs but never of self-related contents would be quite different from what it is like to be conscious only of self-related contents but never of states of affairs in the external world. Intermediate or intermittent mixes – some distance from the normal mix – of the two extremes are also conceivable. Assessment of the integrity of the default mode and attention networks can happen during rest using ultra-slow fluctuations in the BOLD signal. This has therefore been suggested as a test with prognostic impact of the level and degree of consciousness in DOC (Soddu, Boly et al. 2009; Soddu, Vanhaudenhuyse et al. 2011). This would be very valuable in particular because much of the testing can occur without asking patients to perform any task. It is not known how closely activity in these networks correlates with consciousness in healthy subjects, let alone DOC patients. So it cannot be concluded confidently that a patient with intact processing in the default mode network during rest is conscious of self-related content. To target self-related mental activity more directly, exposing patients to self-related stimuli would be desirable. Qin, Di et al., (2010) studied fMRI signal changes in default mode areas during passive listening to self-related vs. non-self-related stimuli (own name vs. names of others). They observed signal changes in anterior midline regions during the presentation of the self-related stimuli comparable to those observed in healthy control participants. Importantly, they also addressed the issue of whether this activity is associated with consciousness. They found correlation between the midline signal change and level of consciousness as measured behaviourally by CRS-R. This is evidence that self-related processing is linked to level of consciousness. However, in the context of doubt about the ability of behavioural tests such as CRS-R to reveal consciousness in VS, the evidence appears less robust: for patients rating relatively low on CRS-R who still have significant signal change for the self-related stimuli it would be an open question whether this selfrelated processing occurs consciously or not. To address some of these issues, it would be useful to explore an active rather than a passive self-related task, such as a task engaging mental time travel by envisioning a future for oneself. If such a task was found to increase default mode activity in a VS or MCS patient (but not a coma patient), then that would be fairly strong evidence of self-related awareness. It does seem implausible that there could be this kind of predicted default mode activity in response to an instruction to “think about your future” but wholly without consciousness. It is an open empirical question whether such self-related awareness can occur in the absence of, for example, the ability for endogenous attention to words, names, faces, or auditory pattern violations. Adopting an active rather than a passive task is more revealing but also more methodologically demanding. There is a sense in which asking a DOC patient to actively envisage a future for herself presupposes that she has conscious volition and some capacity 12

Hohwy & Fox, forthcoming, Journal of Consciousness Studies

for conscious awareness. If the task presupposes these aspects of consciousness then it becomes difficult to interpret the results. It is a bit like asking someone “Are you asleep?” Superficially, it is senseless because it presupposes awakeness so cannot be an independent test of awakeness. But we ask it nevertheless because any answer (“no” as well as “yes!”) is informative (whereas an absence of answer may not allow us to conclude much). Taken in this vein this kind of task would be similar to taking SMA activity, when asked to imagine playing tennis, as evidence of conscious volition. However, if care is taken in characterising the dependent and independent variables this approach can be extended. What is presupposed (the independent variable) is conscious volition and some degree of conscious awareness, what is not presupposed (the dependent variable) is the ability to engage self-related representational content in mental time travel. We can imagine someone who can understand instructions and decide to follow them but who is unable to engage in mental time travel. By analogy, if someone answers “yes!” rather than “no” to “Are you asleep?” we gain more finegrained information about their overall state of mind than just that they are awake. This discussion shows that it is possible to work around some of the potential confounds in this area by being specific about the aspects of consciousness one intends to test for. To round off the discussion in this section, it should be noted that a more indirect measure of self-related processing is also possible if attention demanding tasks (such as those reported in (Bekinschtein, Dehaene et al. 2009; Monti, Coleman et al. 2009) cause deactivation of the default mode network. This would indicate that patients, just as healthy participants, are able to “lose themselves” (Gusnard 2005; Hohwy 2007) in a task (cf. McKiernan, D'Angelo et al. 2006).

5.3. Perceptual awareness and binocular rivalry The original tennis imagery study gave better evidence for volitional consciousness than for basic awareness of the tennis imagery itself. The initial problem was that the observed SMA activity could have been merely automatic, though automaticity is unlikely in the light of the sustained, volitional nature of the response. However, SMA activity itself is not known to be a tight neural correlate of conscious motor imagery; the evidence at best suggests it is a mere causal concomitant to conscious motor imagery. It is therefore difficult to say much about the content and quality of the conscious awareness that may accompany the volitional effort in this study. It would be desirable to investigate perceptual awareness more directly, that is, without task requirements and based on a tight correlation with conscious awareness in healthy controls. One possibility here is to expose patients to binocular rivalry and observe how closely neural activity dynamics fit what is observed in healthy individuals. Binocular rivalry would be an appropriate approach as it is one of the most studied paradigms in consciousness science; moreover, studies of DOC are coming tantalisingly close to using the stimuli seen in standard rivalry approaches (Coleman, Bekinschtein et al. 2009). Binocular rivalry occurs when different stimuli (e.g., a house and a face) are presented to each eye and conscious perception alternates between them in spite of the unchanging physical stimulus. fMRI reveals a complex, distributed pattern of activity that correlates with reports of each conscious percept: there is activity in medial temporal cortex, in early visual cortex, in lateral geniculate nucleus and in right prefrontal cortex during transitions (for reviews, see Tong, Meng et al. 2006; Hohwy, Roepstorff et al. 2008). Though this is still some way from the best estimate of the neural correlates of awareness, it does reveal a great deal about the specific neural structures underpinning selection of one rather than another representational content into consciousness. In addition to the distributed fMRI signature much is known about 13

Hohwy & Fox, forthcoming, Journal of Consciousness Studies

the temporal dynamics of rivalry under different stimulus perturbations. Similar neural and temporal responses in DOC (excluding coma) in passive viewing conditions would be suggestive of preservation of conscious awareness independently of volitional, attentional and other aspects of consciousness. It would be technically challenging to assess the whole distributed range of activity as different fMRI regimes have been used in different studies. Much progress is happening in this area and there is hope that sophisticated neural decoding techniques can bring us closer to content-specific neural correlates of conscious perception during rivalry (Haynes 2009; Overgaard 2009). A less involved, but more indirect, approach would be to utilise a recent finding documenting pupil dilations during rivalry (Einhauser, Stout et al. 2008). Healthy participants were found to have pupil dilations just before reported perceptual alternations for a range of bistable auditory and visual stimuli. If a DOC patient displays the same dilations it would not be unreasonable to suggest she also experiences percept alternations. Such tests of awareness would have the added benefit that they are not primarily based on a prior theory of consciousness, which was a worry for the global workspace and information integration approaches noted above. In contrast, these tests rely more strongly on the assumption that we can generalise from biological markers of consciousness in healthy individuals to patients with DOC. A potential problem with this approach is that even if one supposes that the minimally sufficient conditions for conscious awareness are intact, the wider, enabling conditions for them (relating for example to arousal levels) are impaired in DOC. The problem would be that even though contents are selected for presentation in consciousness, there may be no overall conscious state for them to be presented in. This problem can only be addressed directly once the thorny issues of the role of enabling conditions for consciousness is resolved (for discussion, see Hohwy 2009). It is therefore possible that such activity is in the end not associated with conscious content processing – that it is in the end merely automatic. But this type of automaticity objection has less traction here than in the tennis imagery case because here the neural activity is more directly correlated with conscious awareness in healthy individuals. It is not that the observed activity may be irrelevant to consciousness; rather it is that some additional damage may prevent it from making its normal contribution to consciousness.

5.4. Conscious access and masking In healthy individuals, some content processing is not accessible to introspective report and is therefore assumed to be unconscious. This happens for example for masked stimuli. A recent study of masked vs. unmasked stimuli using intracranial electroencephalogram (iEEG) found that the neurophysiological markers of conscious access include a distinct signature of ‘sustained voltage changes, particularly in prefrontal cortex, large increases in spectral power in the gamma band, increases in long-distance phase synchrony in the beta range, and increases in long-range Granger causality’ (Gaillard, Dehaene et al. 2009). If a non-invasive version of this iEEG paradigm were to be developed, then it could probe conscious access to representational content in DOC. It would have the advantage that patients are not required to perform any active task. The neurophysiological markers would be correlated in healthy individuals specifically with conscious access to the presented content. Therefore, it would be harder to object that the markers are signs of unconscious, automatic processing. The remaining problem would be if there is damage to some of the wider enabling 14

Hohwy & Fox, forthcoming, Journal of Consciousness Studies

conditions for the neural systems underpinning conscious access, which would make them, as it were, function blindly. A rough analogue of this test is used in the multimodal approach to DOC where the integrity of sensory neural axes is ascertained before volitional tests are administered (Coleman, Bekinschtein et al. 2009). However, this marker is not as fine-grained as the ones revealed with iEEG and it is not as tightly correlated with conscious access. This type of test would also be relevant for the global workspace approaches discussed above.

5.5. First person perspective and multisensory integration It seems that in so far as there is any consciousness at all it comes with some kind of first person perspective. This notion of a first-person perspective is weaker than the notion of having a self or being self-conscious, it just means that there is some privileged point from which any conscious bodily sensations and perceptions are ordered (Blanke and Metzinger 2009; Metzinger 2009). Having a first-person perspective is crucial for binding sensory attributes and for multisensory integration: without a privileged point of view there is no right way to combine colour and movement, or vision with touch or sound. The preservation of multisensory integration could therefore indicate not only a degree of intact sensory processing but also preservation of a first-person perspective. This would suggest the preservation of an important ingredient of consciousness – an indication that there is an individual there, even in the potential absence of conscious volition, attention, access or self-referential experience. The rubber hand illusion is an example of multisensory, visuotactile integration. Objective measures of this illusion include skin conductance responses and skin temperature changes, both of which could potentially be used in passive tasks in DOC (Kammers, Rose et al. ; Botvinick and Cohen 1998; Moseley, Olthof et al. 2008; Hohwy and Paton 2010). Also relevant is the ventriloquist effect where synchronised auditory and visual stimuli are perceived to be spatially co-located. Under different levels of noise, the two stimuli are weighted in a Bayes-optimal way (Alais and Burr 2004) and differential weightings could potentially be detected with eye movement differences in a DOC patient.

15

Hohwy & Fox, forthcoming, Journal of Consciousness Studies

Dependent variable

Proposed tests of consciousness in DOC Type of evidence Study

Visual imagery

fMRI

(Owen 2006; Monti, Vanhaudenhuyse et al. 2010)

Instruction following

fMRI

(Owen 2006; Schnakers, Perrin et al. 2008; Bekinschtein, Dehaene et al. 2009; Monti, Coleman et al. 2009; Monti, Vanhaudenhuyse et al. 2010) (Bekinschtein, Dehaene et al. 2009) (Monti, Coleman et al. 2009) (Schnakers, Perrin et al. 2008) (Soddu, Boly et al. 2009).

ERP

Endogenous attention to global auditory patterns Endogenous attention to visual stimuli Endogenous attention to auditory stimuli Default mode network activity during rest Default mode network activity for self-related stimuli Communication

ERP fMRI ERP fMRI Ultra-slow fMRI fluctuations

fMRI e.g., fMRI

Aspect of consciousness targeted Awareness of imagery; representational content Volition

Attention, awareness Attention, awareness Attention, awareness Ability for self-related thought

(Qin, Di et al. 2010)

Self-related stimuli

(Monti, Vanhaudenhuyse et al. 2010)

Volition, indirectly all aspects

Table 1 Dependant variable

Potential tests of consciousness in DOC Type of Related type of evidence study

Information integration

TMS/EEG

(Massimini, Boly et al. 2009)

Default mode activity during mental time travel

fMRI

Bistable perception

fMRI (possibly with neural decoding)

(McKiernan, D'Angelo et al. 2006; Soddu, Boly et al. 2009; Qin, Di et al. 2010). (Lumer, Friston et al. 1998; Tong, Nakayama et al. 1998; Haynes, Deichmann et al. 2005; Haynes 2009)

Pupil dilation Conscious access

iEEG/EEG

Multisensory integration

Skin conductance response; skin temperature; eye movement.

(Einhauser, Stout et al. 2008) (Gaillard, Dehaene et al. 2009) (Alais and Burr 2004; Moseley, Olthof et al. 2008).

Table 2 16

Aspect of consciousness targeted Binding, theoretical quantity of information integration Self-related thought; Selfconsciousness

Awareness

Awareness, conscious access First-person perspective

Hohwy & Fox, forthcoming, Journal of Consciousness Studies

6. Discussion 6.1. Probing different, potentially dissociable, aspects of consciousness via different tests A number of aspects of consciousness were outlined and a number of studies relevant for the issues of preserved consciousness in DOC were surveyed. Some of these studies are being developed as tests of preserved consciousness, some have been suggested as tests but are not yet developed, and some have potential as such tests. The main finding of this review is that different proposed and potential tests are best seen as targeting distinct and potentially dissociable aspects of consciousness such as awareness, arousal, volition, global availability, access, attention, self, and first-person perspective. Consequently, were different DOC patients to pass different sets of these tests, a cautious inference would be that there is preservation of the corresponding aspects of consciousness. Different DOC may be characterised by different or varying aspects of consciousness. This is a more differentiated picture of consciousness than commonsense experience dictates and than most research assumes. It suggests that preserved consciousness in DOC is a multidimensional matter. This has implications for clinical and ethical issues, some of which are addressed below. For many of these different aspects of consciousness, there is little concrete evidence that they cannot dissociate, and, conceptually, it seems many of them can dissociate. It is a wide open empirical question to what extent there are interactions between these different aspects of consciousness: impairment of one aspect could have dramatic effect on some but not other aspects. In addition, current tests already suggest that different aspects begin to dissociate in VS and MCS. For example, some VS patients can manifest volition in connection with requests for visual imagery and some MCS patients can comply with requests to volitionally engage endogenous attention, but not always vice versa (a confounding factor for these findings is that the MCS vs. VS diagnosis, though greatly improved, is still laden with some uncertainty and may also change between initial assessment and later testing). For healthy individuals, outside of experimental conditions, many of these aspects of consciousness tend to be associated. The DOC research is compatible with the hypothesis that different neural systems underpin different aspects of consciousness. This in turn suggests that care should be taken when we transpose our normal experience of unified consciousness to the level of consciousness that may be preserved in DOC. It may be that the varying levels of consciousness in DOC are less a matter of differences in the overall intensity of unified consciousness and more a matter of the multidimensional structure of the specific aspects of consciousness that happen to be preserved.

6.2. Methodological issues When testing for consciousness in DOC there is a major methodological divide between conscious volition and consciousness without a volitional element.. For the former, the time course is most important, whereas for the latter it is the type of neurophysiological or psychophysical evidence. For volitional consciousness the nature of the covert evidence acquired via fMRI or EEG is less important than its time course, as it is the fact that the test is time locked to requests to follow instructions that provides the evidence for capacity for volition. It is not obvious however whether one can generalise from having found volition in one cognitive domain (e.g., in the imagery task) to volition being intact in other areas. The reason is that manifestation of volition depends on appropriate content for volitional 17

Hohwy & Fox, forthcoming, Journal of Consciousness Studies

processing being available. For the latter, non-volitional consciousness, the specific types of neurophysiological or psychophysical evidence is most important. The reason is that suspicions of automaticity are best dispelled if the evidence of consciousness coheres very closely with known neural correlates of consciousness in healthy individuals, rather than more loosely connected upstream or downstream causal concomitants. Though the charge of automaticity can be addressed in that way, it is still possible for the enabling conditions for the neural correlates to be malfunctioning, meaning that consciousness is, after all, undermined.

6.3. Absence of evidence and evidence of absence There is agreement in the field that an absence of evidence of preservation of consciousness in DOC is not evidence of an absence of preserved consciousness. This is ethically crucial. The suggestion that preservation of consciousness in DOC is a multidimensional matter broadens this judgement. If there are many aspects of consciousness then, even if we took absence of evidence to be evidence of absence, the evidence would concern only limited aspects of consciousness, leaving it possible that other aspects are preserved. This strengthens the call for increased caution concerning these difficult cases (Hohwy and Reutens 2009). It is possible that increased sensitivity to aspects of consciousness in testing for DOC consciousness, together with a tight coupling with the neural correlates of consciousness found in healthy individuals, will result in a fine grained theory that could in principle support an inference from absence of evidence to evidence of absence (Shea and Bayne 2010). The “absence of evidence” slogan is not without exceptions, for example, it seems reasonable to infer from not hearing rain falling on my tin roof to the conclusion that there is an absence of rain, and this kind of reasoning seems to depend on how tightly evidence and states of affairs are tied together (Sober 2009). However, the mooted inference from absence of evidence of consciousness to evidence of an absence of consciousness in DOC involves an additional problematic factor, namely that the inference would occur from healthy to damaged brains. Even if we were in a situation where an absence of evidence in the undamaged brain would be significant, it would be contentious to rule out any vestiges of consciousness in the damaged brain of a DOC patient.

6.4. Variations on the argument from analogy to the existence of other minds The argument from analogy says an inductive inference can be made from one’s own mental states to those of someone else, if there are relevant behavioural similarities. As mentioned, more sophisticated versions of this argument allow more sources of evidence than the purely behavioural, such as neurophysiological commonality or commonality in biological species (Hill 1991). When considering volitional signs of consciousness in DOC, such as fMRI activity that is time-locked to the experimenter’s requests, the picture is a little more complicated and the argument from analogy less direct. The argument from analogy in this case appeals to similarities in covert behaviour between creatures of the same species that may have different neurophysiological states (due to damage in the DOC patient). The appeal to covert behaviour makes the link to mental states less direct than for overt behaviour. For overt behaviour it is relatively easy for us all to relate it to a type of mental state but for overt behaviour we do not have much prior evidence of a stable relation to types of mental states (only a relatively limited number of fMRI studies are available). This makes the inference somewhat weaker. Further, there is a risk that though the DOC patient is the same biological species as the healthy comparison, there are neurophysiological differences caused by damage to the brain of the patient. This also weakens the inference. There are some attempts at overcoming these weaknesses: the multimodal approach to DOC pursued by 18

Hohwy & Fox, forthcoming, Journal of Consciousness Studies

Coleman and colleagues at Cambridge (Coleman, Bekinschtein et al. 2009) seeks evidence of the integrity of sensory neural axes before investigating the preservation of volitional consciousness. This can be viewed as an attempt to establish similarity in neurophysiological functioning before applying the argument from analogy. A further complication for the argument from analogy is that in DOC we cannot, as we can in healthy individuals, ascertain the validity of the evidence as a test of consciousness by subsequently asking for an introspective report (as in “were you in pain when you winced just then?”). The desirability of such subjective validation is not driven by blind faith in introspective reports. It is just that it is always useful to have further, relatively independent sources of evidence. These limitations mean that the argument from analogy is more complicated, and somewhat weaker, in the case of covert evidence in DOC. In healthy individuals, a certain BOLD signal, for example, co-occurs with a certain conscious state such as imagining playing tennis; the validity of this evidence as a test of consciousness can be gauged by asking for an introspective report. The inference that someone displaying a similar BOLD signal is also in that conscious state piggybacks on the strong validity of the inference in the non-DOC case. This does not mean the inference is fallacious, it just means that the inductive strength is somewhat weaker for the argument from analogy in the case of DOC.

6.5. What type of consciousness comes with volition and covert communication? What does the preservation of volition imply about the conscious state of a VS patient? The first thing to notice is that though volitional task compliance probably comes with some awareness of the content of the instructions, it doesn’t indicate preservation of other aspects of consciousness. Volition does not immediately indicate wholly intact consciousness. In particular, it is not settled whether volition can occur without attention, without global access, without self-awareness, consciousness of having a self, or having a first-person perspective. From a commonsense perspective volition means being able to consider, and decide among, a range of options for action in the context of one’s desires and preferences as well as background beliefs. Though some aspects of volition perhaps can be automatised (Bargh and Chartrand 1999), there is little reason to believe that new goals (like “imagine playing tennis”) can be immediately automatised. There is therefore good reason to believe volition, in this case, correlates with some degree of conscious deliberation. Preservation of this ability in VS patients would be immensely important. However, the extent and depth of conscious deliberation is unknown. For all the data shows, it may be that requests for imagining tennis or navigating one’s house are consciously deliberated in a very impoverished context. This may be if the patient does not have access to information about her other desires and preferences in which case there is no deliberative competition for her decision. This would be a kind of “conscious automatism” and could help explain why some patients can display volition but cannot utilise it for covert communication. Different kinds of volitional task compliance may also give different responses in different patient groups. It may be that requests to count occurrences of a name or word (Schnakers, Perrin et al. 2008; Monti, Coleman et al. 2009) recruit different and perhaps more globally distributed abilities than do the request to imagine playing tennis. It can also be the case that only a limited range of desires and preferences can be taken into account when deliberating about an action. For example, it may be that there is access to short term desires but only limited access to long term desires (as could perhaps occur as 19

Hohwy & Fox, forthcoming, Journal of Consciousness Studies

consequence of impairment to networks subserving mental time travel and mentalising (Buckner, Andrews-Hanna et al. 2008; Soddu, Boly et al. 2009). In that case, there will be impairment of potentially vital decisions concerning long time care and end of life decisions. It is also possible that a patient be able to engage in covert communication but unable to engage long term planning. Such a patient may deliberate vital questions about future care in the context of short term desires only, and perhaps without mentalising about others’ mental states. As a dramatic, hypothetical case, imagine such a patient uses tennis imagery to answer “yes” to a question about withdrawal of life support. Even though there is rational deliberation, and no indication of depression or the like, it is not certain we should honour this wish because if the patient had access to a wider range of desires she would perhaps not decide to have her life ended. As a final conceptual twist it is conceivable that what is detected in these patients is a kind of ersatz-volition. True volition is paradigmatically manifested when one decides to act in a certain way. Volition in the DOC cases is manifested when experimenters request a certain mental action from the patient. It is possible that the patients are only able to manifest these responses when requested and not of their own “free will”, making the volition ersatz. This sounds fanciful but there is an analogy in blindsight where subjects can detect (above chance) stimuli presented to their blind field under forced-choice conditions but are unable to themselves initiate action when a stimulus is presented in their blind field. It is possible, that is, that there is volition in DOC but only when patients are requested to manifest it. Empirical inroads can be made on this issue if it is possible to ask patients, via the tennis-imagery route to covert communication, about the nature of their conscious state outside of experimental conditions. In summary, the presence of volition itself doesn’t show how much conscious rational deliberation there is. Neither does it indicate which other aspects of consciousness are preserved. Similarly, the presence of overt communication doesn’t show how much of consciousness is preserved nor the quality of the responses that are being communicated.

6.6. Ethical issues Clearly, if ethical decisions about DOC patients take their state of consciousness into account, then a change in state of consciousness must challenge such ethical decisions. This concerns both issues about the quality of care and issues about end of life decisions. Given the findings to date, it is however difficult to say how ethical decisions should change. Some argue that if there is preserved consciousness then there is increased risk of pain and discomfort and that this suggests quicker intervention to end these patients’ lives (see Kahane and Savulescu 2009). Some argue that the aspect of preserved consciousness suggested by, for example, Owen’s tennis imagery study, is not evidence of conscious personhood, that is, the presence of a creature with conscious desires for the future. Therefore, this argument goes, these findings are not relevant to end of life decisions (Horne 2009; Levy 2009). Others argue that the findings to date are too inconclusive to go either of those two ways. The science of consciousness is not yet advanced enough to allow an interpretation of these imaging and ERP data that could reasonably sway us towards more swiftness or status quo concerning end of life decisions. Instead they simply make end of life decisions even more difficult than they already are. The best we can hope for is progress in the research on DOC in tandem with progress on finding the neural correlates of consciousness. Meanwhile, the change in ethical practice should go towards even better palliative care (Hohwy and Reutens 2009). The discussion in the present review supports this call for increased caution: ethical decisions are 20

Hohwy & Fox, forthcoming, Journal of Consciousness Studies

made more difficult once we realise that a broad range of possibly dissociable different conscious states can be preserved in DOC.

7. Conclusion Consciousness is not a unitary and unequivocal phenomenon. Our commonsense conception of consciousness allows a range of different aspects: phenomenal experience, self-awareness, first person perspective, volition, attention, awareness, arousal, and so on. When we get evidence of preserved consciousness in disorders of consciousness (DOC) it is therefore a legitimate question which aspect of consciousness it is evidence of. We have set out a list of aspects of consciousness and paired each aspect with suggested or potential empirical studies in disorders of consciousness. This shows that not only can we conceptually distinguish aspects of consciousness; it is also possible to garner empirical evidence for preservation of some of these aspects. This holds even in the absence of the type of evidence we normally appeal to, namely overt behaviour, which is of course the case relevant for DOC. This allows more fine-grained inferences about preservation of consciousness in DOC, since we can now say something about the specific aspects of consciousness. However, it also introduces a range of complexities. 1. Most of these aspects of consciousness go together in the case of healthy individuals and there is very little evidence concerning the extent to which they depend on each other and how they can dissociate in the case of DOC or other insults to the brain. It is therefore difficult to generalise from the preservation of one particular aspect of consciousness to the preservation of other aspects in the same patient. 2. The type of argument normally used to infer the presence of mental states in other creatures has its bedrock in the behaviour of those creatures – this is the argument from analogy. This argument type is appropriate but weakened when it is applied to DOC because it generalises from the healthy to the damaged brain and because it relies, not on normal communicative behaviour but on either overlap in neural correlates or covert communication. This is compounded by the need for tight neural correlates, rather than loose causal concomitants, of consciousness as the basis of comparisons for aspects of consciousness, and evidence for these are scarce and primarily obtained in healthy individuals. 3. The case of conscious volition has received most attention. Though preservation of volition is very important it is as yet unclear how volition relates to concurrent aspects of consciousness and related cognitive domains. Therefore the mere presence of volition provides only limited information regarding preservation of a patient’s overall conscious state. The scientific research for aspects of preserved consciousness in disorders of consciousness is advancing rapidly. This opens new possibilities for treatment as well as for basic understanding of consciousness. In particular, it is now possible to gauge potentially dissociable aspects of consciousness, aspects that are difficult to pinpoint in the healthy population. However, this opportunity for progress comes with the price of increased complexity of this field of research, which has ethical implications for treatment possibilities.

References: Alais, D. and D. Burr (2004). "The ventriloquist effect results from near-optimal bimodal integration." Curr. Biol. 14: 257. Baars, B. J. (1996). In the Theater of Consciousness: the Workspace of the Mind. 21

Hohwy & Fox, forthcoming, Journal of Consciousness Studies

Bargh, J. and T. L. Chartrand (1999). "The unbearable automaticity of being." American Psychologist 54: 462-479. Bayne, T. (2007). "Conscious states and conscious creatures: explanation in the scientific study of consciousness." Philosophical Perspectives 21(1): 1-22. Bekinschtein, T., V. Cologan, et al. (2009). You are only coming through in waves: wakefulness variability and assessment in patients with impaired consciousness. Progress in Brain Research, Elsevier. Volume 177: 171-189. Bekinschtein, T. A., S. Dehaene, et al. (2009). "Neural signature of the conscious processing of auditory regularities." Proceedings of the National Academy of Sciences 106(5): 1672-1677. Blanke, O. and T. Metzinger (2009). "Full-body illusions and minimal phenomenal selfhood." Trends in Cognitive Sciences 13(1): 7-13. Block, N. (2001). "Paradox and cross purposes in recent work on consciousness." Cognition 79: 197-219. Block, N. (2005). "Two neural correlates of consciousness." Trends in Cognitive Sciences 9(2): 46. Block, N. (2008). "Consciousness, accessibility, and the mesh between psychology and neuroscience." Behavioral and Brain Sciences 30(5-6): 481-499. Boly, M. (2007). "When thoughts become actions: an fMRI paradigm to study volitional brain activity in non-communicative brain injured patients." Neuroimage 36: 979-992. Boly, M., M. Massimini, et al. (2009). Theoretical approaches to the diagnosis of altered states of consciousness. Progress in Brain Research, Elsevier. Volume 177: 383-398. Boly, M., C. Phillips, et al. (2008). "Consciousness and cerebral baseline activity fluctuations." Human Brain Mapping 29(7): 868-874. Botvinick, M. and J. Cohen (1998). "Rubber hands `feel' touch that eyes see." Nature 391(6669): 756-756. Buckner, R. L., J. R. Andrews-Hanna, et al. (2008). "The Brain's Default Network: Anatomy, Function, and Relevance to Disease." Ann NY Acad Sci 1124(1): 1-38. Chalmers, D. (2000). What is a neural correlate of consciousness? Neural Correlates of Consciousness: Empirical and Conceptual Issues. T. Metzinger. Cambridge, Mass., MIT Press. Chalmers, D. (2002). Consciousness and its place in nature. Philosophy of Mind. D. J. Chalmers. Oxford, Oxford University Press. Coleman, M. R., T. Bekinschtein, et al. (2009). A multimodal approach to the assessment of patients with disorders of consciousness. Progress in brain research, Elsevier. Volume 177: 231-248. Dehaene, S. and L. Naccache (2001). "Towards a cognitive neuroscience of consciousness: basic evidence and a workspace framework." Cognition 79: 1–37. Demertzi, A., A. Vanhaudenhuyse, et al. (2008). "Is there anybody in there? Detecting awareness in disorders of consciousness." Expert Review of Neurotherapeutics 8(11): 1719-1730. Dennett, D. (2001). "Are we explaining consciousness yet?" Cognition 79(1-2): 221. Einhauser, W., J. Stout, et al. (2008). "From the Cover: Pupil dilation reflects perceptual selection and predicts subsequent stability in perceptual rivalry." Proceedings of the National Academy of Sciences 105(5): 1704-1709. Englot, D. J., H. Blumenfeld, et al. (2009). Consciousness and epilepsy: why are complex-partial seizures complex? Progress in Brain Research, Elsevier. Volume 177: 147-170. Fischer, C., D. Morlet, et al. (1999). "Mismatch negativity and late auditory evoked potentials in comatose patients." Clinical Neurophysiology 110(9): 1601-1610. Gaillard, R., S. Dehaene, et al. (2009). "Converging Intracranial Markers of Conscious Access." PLoS Biol 7(3): e1000061. Gaillard, R. l., S. Dehaene, et al. (2009). "Converging Intracranial Markers of Conscious Access." PLoS Biol 7(3): e1000061. 22

Hohwy & Fox, forthcoming, Journal of Consciousness Studies

Giacino, J. T., K. Kalmar, et al. (2004). "The JFK Coma Recovery Scale-Revised: measurement characteristics and diagnostic utility." Arch. Phys. Med. Rehabil. 85: 2020-2029. Giacino, J. T., C. Schnakers, et al. (2009). Behavioral assessment in patients with disorders of consciousness: gold standard or fool's gold? Progress in Brain Research, Elsevier. Volume 177: 33-48. Greicius, M. D., B. Krasnow, et al. (2003). "Functional connectivity in the resting brain: a network analysis of the default mode hypothesis." Proc. Natl Acad. Sci. USA 100: 253-258. Gusnard, D. A. (2005). "Being a self: Considerations from functional imaging." Consciousness and Cognition 14(4): 679. Haynes, J.-D. (2009). "Decoding visual consciousness from human brain signals." Trends in Cognitive Sciences 13(5): 194-202. Haynes, J. D., R. Deichmann, et al. (2005). "Eye-specific effects of binocular rivalry in the human lateral geniculate nucleus." Nature 438: 496. Hill, C. (1991). Sensations: A Defense of Type Materialism. Cambridge, Cambridge University Press. Hohwy, J. (2007). "The search for neural correlates of consciousness." Philosophy Compass 2(3): 461–474. Hohwy, J. (2007). "The sense of self in the phenomenology of agency and perception." Psyche 13(1). Hohwy, J. (2009). "The neural correlates of consciousness: New experimental approaches needed?" Consciousness and Cognition 18(2): 428-438. Hohwy, J. and B. Paton (2010). "Explaining away the body: experiences of supernaturally caused touch and touch on non-hand objects within the rubber hand illusion " PLoS ONE 5(2): e9416. Hohwy, J. and D. C. Reutens (2009). "A case for increased caution in end of life decisions for disorders of consciousness." Monash Bioethics Review 28(2): 14.11-14.12. Hohwy, J., A. Roepstorff, et al. (2008). "Predictive coding explains binocular rivalry: An epistemological review." Cognition 108(3): 687-701. Horne, M. (2009). "Are people in a persistent vegetative state conscious?" Monash Bioethics Review 28(2): 12.11-12.12. Hume, D. (1739-40). A Treatise of Human Nature. Oxford, Oxford: Clarendon Press. Hyslop, A. (2005) "Other Minds." The Stanford Encyclopedia of Philosophy. Kahane, G. and J. Savulescu (2009). "Brain Damage and the Moral Significance of Consciousness." J Med Philos 34(1): 6-26. Kammers, M. P. M., K. Rose, et al. "Feeling numb: Temperature, but not thermal pain, modulates feeling of body ownership." Neuropsychologia In Press, Accepted Manuscript. Koch, C. (2004). The Quest for Consciousness: A Neurobiological Approach. Englewood, Colorado, Robert and Company Publishers. Koch, C. and N. Tsuchiya (2007). "Attention and consciousness: two distinct brain processes." Trends in Cognitive Sciences 11(1): 16-22. Laureys, S. (2005). "The neural correlate of (un)awareness: lessons from the vegetative state." Trends in Cognitive Sciences 9(12): 556. Laureys, S., M. E. Faymonville, et al. (2000). "Restoration of thalamocortical connectivity after recovery from persistent vegetative state." Lancet 355: 1790-1791. Laureys, S., F. Perrion, et al. (2007). "Self-consciousness in non-communicative patients." Conscious. Cogn. 16: 722-741. Levy, N. (2009). "What difference does consciousness make?" Monash Bioethics Review 28(2): 13.11-13.13. Levy, N., J. Savulescu, et al. (2009). Moral significance of phenomenal consciousness. Progress in Brain Research, Elsevier. Volume 177: 361-370. 23

Hohwy & Fox, forthcoming, Journal of Consciousness Studies

Lumer, E. D., K. Friston, et al. (1998). "Neural correlates of perceptual rivalry in the human brain." Science 280: 1930. Majerus, S., M.-A. Bruno, et al. (2009). The problem of aphasia in the assessment of consciousness in brain-damaged patients. Progress in Brain Research, Elsevier. Volume 177: 49-61. Massimini, M., M. Boly, et al. (2009). "A perturbational approach for evaluating the brain’s capacity for consciousness." Progress in Brain Research 177: 201-214. McKiernan, K. A., B. R. D'Angelo, et al. (2006). "Interrupting the "stream of consciousness": An fMRI investigation." NeuroImage 29(4): 1185-1191. Metzinger, T. (2009). The Ego Tunnel. New York, Basic Books. Mill, J. S. (1865). An Examination of Sir William Hamilton's Philosophy. London, Longmans. Monti, M. M., M. R. Coleman, et al. (2009). "Neuroimaging and the Vegetative State." Annals of the New York Academy of Sciences 1157(Disorders of Consciousness): 81-89. Monti, M. M., M. R. Coleman, et al. (2009). Executive functions in the absence of behavior: functional imaging of the minimally conscious state. Progress in brain research, Elsevier. Volume 177: 249-260. Monti, M. M., A. Vanhaudenhuyse, et al. (2010). "Willful Modulation of Brain Activity in Disorders of Consciousness." N Engl J Med: NEJMoa0905370. Mormann, F. and C. Koch (2007) "Neural correlates of consciousness " Scholarpedia 2, 1740. Moseley, G. L., N. Olthof, et al. (2008). "Psychologically induced cooling of a specific body part caused by the illusory ownership of an artificial counterpart." Proceedings of the National Academy of Sciences 105(35): 13169-13173. Naccache, L. (2006). "Is she conscious?" Science 313: 1395-1396. Overgaard, M. (2009). How can we know if patients in coma, vegetative state or minimally conscious state are conscious? Progress in Brain Research. S. Laureys, N. D. Schiff and A. M. Owen, Elsevier. Volume 177: 11-19. Owen, A. M. (2006). "Detecting awareness in the vegetative state." Science 313: 1402. Owen, A. M. (2008). "Disorders of Consciousness." Annals of the New York Academy of Sciences 1124(1): 225-238. Perrin, F., C. Schnakers, et al. (2006). "Brain response to one's own name in vegetative state, minimally conscious state, and locked-in syndrome." Arch Neurol 63: 562 - 569. Posner, J. B., C. B. Saper, et al. (2007). Plum and Posner’s Diagnosis of Stupor and Coma. Oxford Oxford University Press. Qin, P., H. Di, et al. (2010). "Anterior cingulate activity and the self in disorders of consciousness." Human Brain Mapping: n/a-n/a. Raichle, M. E. (2001). "A default mode of brain function." Proc Natl Acad Sci U S A 98(2): 676682. Schnakers, C., F. Perrin, et al. (2008). "Voluntary brain processing in disorders of consciousness." Neurology 71(20): 1614-1620. Shea, N. and T. Bayne (2010). "The Vegetative State and the Science of Consciousness." British Journal for the Philosophy of Science 61: 459-484. Sober, E. (2009). "Absence of evidence and evidence of absence: evidential transitivity in connection with fossils, fishing, fine-tuning, and firing squads." Philosophical Studies 143(1): 63-90. Soddu, A., M. Boly, et al. (2009). Reaching across the abyss: recent advances in functional magnetic resonance imaging and their potential relevance to disorders of consciousness. Progress in Brain Research, Elsevier. Volume 177: 261-274. Soddu, A., A. Vanhaudenhuyse, et al. (2011). "Identifying the default-mode component in spatial IC analyses of patients with disorders of consciousness." Human Brain Mapping: n/a-n/a.

24

Hohwy & Fox, forthcoming, Journal of Consciousness Studies

Sorger, B., B. Dahmen, et al. (2009). Another kind of [`]BOLD Response': answering multiplechoice questions via online decoded single-trial brain signals. Progress in Brain Research, Elsevier. Volume 177: 275-292. Stins, J. and S. Laureys (2009). "Thought translation, tennis and Turing tests in the vegetative state." Phenomenology and the Cognitive Sciences 8(3): 361-370. Tong, F., M. Meng, et al. (2006). "Neural bases of binocular rivalry." Trends in Cognitive Sciences 10(11): 502. Tong, F., K. Nakayama, et al. (1998). "Binocular rivalry and visual awareness in human extrastriate cortex." Neuron 21: 753-759. Tononi, G. (2005). Consciousness, information integration, and the brain. Progress in Brain Research. S. Laureys, Elsevier. Volume 150: 109. Tononi, G. and C. Koch (2008). "The Neural Correlates of Consciousness: An Update." Annals of the New York Academy of Sciences 1124(1): 239. Zeman, A. and N. D. S. a. A. M. O. Steven Laureys (2009). The problem of unreportable awareness. Progress in Brain Research, Elsevier. Volume 177: 1-9.

25