© Springer-Verlag Berlin Heidelberg 2014Grace Lee, Judy Illes and Frauke Ohl (eds.)Ethical Issues in Behavioral NeuroscienceCurrent Topics in Behavioral Neurosciences1910.1007/7854_2014_338
Externalization of Consciousness. Scientific Possibilities and Clinical Implications
Centre for Research Ethics and Bioethics, Uppsala University, Uppsala, Sweden
Coma Science Group, Cyclotron Research Centre, University and University Hospital of Liège, Liège, Belgium
The paper starts by analyzing recent advancements in neurotechnological assessment of residual consciousness in patients with disorders of consciousness and in neurotechnology-mediated communication with them. Ethical issues arising from these developments are described, with particular focus on informed consent. Against this background, we argue for the necessity of further scientific efforts and ethical reflection in neurotechnological assessment of consciousness and ‘cerebral communication’ with verbally non-communicative patients.
KeywordsConsciousnessDisorders of consciousnessNeurotechnologyInformed consent
The instrumental investigation of consciousness has witnessed an astonishing progress over the last years. Different neurotechnological tools and methods have been developed in order to assess residual consciousness in patients with disorders of consciousness (DOCs). Functional neuroimaging technologies, such as electroencephalography (EEG), magnetoencephalography (MEG), functional magnetic resonance imaging (fMRI), positron emission tomography (PET), single photon emission tomography (SPECT), event-related potentials (ERPs), magnetoencephalography (MEG), magnetic resonance spectroscopy (MRS), and transcranial magnetic stimulation (TMS) (Laureys et al. 2009), give researchers the possibility to see what happens in the brain during the execution of particular tasks. These emerging neurotechnologies are very promising in regard to the study and the treatment of DOCs . Notably, identification of activated brain areas and real-time observation of cerebral activity potentially allow a new form of technology-based communication in the absence of overt external behavior or speech, thus going beyond the behavioral manifestation of awareness (Evers and Sigman 2013).
It is important to clarify and to assess some issues emerging from this kind of communication. First of all is the relationship between brain activity, which is the specific object of the neuroimaging investigation, and awareness: how to judge when the first implies the second. Another important issue concerns the kind of consciousness that patients with DOCs retain (e.g., can they perceive the same emotional meaning of the provided information?). As a further development of these analyses, the question of how to assess the capacity of patients with DOCs to make an appropriate informed decision will also arise.
In short, the new advancements in neurotechnological assessment of residual consciousness in patients with DOCs raise important ethical issues, such as how to assess residual capacity of self-determination; whether and how much a prospective ‘cerebral communication’ may be considered as valid for an Informed consent ; and whether a prospective direct communication with patients with DOCs through neurotechnology implies the necessity to rethink their clinical management, particularly the role of legal guardians.
According to Laureys and Schiff, the most relevant result of the progress in the neuroimaging investigation of consciousness is the passage from a monolithic way of looking at DOCs to a more graded nosology based on a quantitative assessment of consciousness and on functional neuroimaging technologies. Neurotechnology allowed researchers to detect important neurological differences between patients that are behaviorally classified as equal: As a result, both description and diagnosis of DOCs are more detailed, and new nosographic criteria and categories have been elaborated (Laureys and Schiff 2012). Furthermore, advancements in neuroimaging research have allowed the development of novel investigational paradigms that provide an imaging indication of volition and awareness: This indication may appear but is not unanimously assumed as unambiguous (Laureys and Schiff 2012). One of the earliest studies, conducted by Owen, Laureys and colleagues in 2006 (Owen et al. 2006), is particularly relevant in showing the possible dissociation between the clinical examination based on the behavioral appearance and the results of a neuroimaging assessment (in this case, an fMRI examination). A young woman who survived after a car accident was behaviorally diagnosed as being in a vegetative state (VS) according to the international guidelines. The researchers’ team pronounced some sentences (e.g., ‘There was milk and sugar in his coffee’) and measured through fMRI her neural responses comparing them with responses to acoustically matched noise sequences. Interestingly, the woman’s neural reaction to the sentences was equivalent to the control subjects’ reactions, yet this result alone is not sufficient to conclude that the woman is aware because there is the possibility of implicit processing: Some aspects of human cognition, as language perception and understanding, can go on without awareness (Fine and Florian Jaeger 2013). For this reason, the research team developed a complementary fMRI study asking the woman to mentally perform two tasks: imagining playing tennis and imagining visiting her house. The relevant result was that the brain activation of the woman was not distinguishable from that of the control subjects, a group of conscious volunteers.
Similar results were obtained in the follow-up study jointly conducted in Liege and Cambridge: 54 patients with severe acquired brain injuries were scanned through fMRI. In response to the request to perform imagery tasks, 5 of them were able to modulate their brain activity by generating blood-oxygenation-level-dependent (BOLD) responses which were judged by the researchers as voluntary, reliable, and repeatable (Monti et al. 2010). Additional tests in one of the 5 responsive subjects revealed his ability to correctly answer yes–no questions through imagery tasks, showing the feasibility of communication. These results are ethically very significant: If new diagnostic tools are available, then it is ethically warranted to use them and to give all the patients the possibility to be rightly diagnosed through them.
Given the possibility that patients with DOCs retain the capacity to communicate and express their own thoughts and preferences, the ethical question of their Informed consent arises.
In this paper, we discuss some technical aspects of fMRI and brain–computer interfaces (BCI) and their prospective use for communicating with patients with DOCs . Furthermore, we analyze the epistemological issue of the relevance of neural activation in the patient for proving or suggesting his/her ability to communicate. Against that background, we analyze emerging ethical issues of Informed consent .
2 The Possibility and Meaning of ‘Cerebral Communication’
To date, fMRI is the most commonly used and one of the most promising tools to study DOCs , especially for its noninvasive nature, ever-increasing availability, relatively high spatiotemporal resolution, its capacity to demonstrate the entire network of brain areas activated in particular tasks, and its capacity to provide both anatomical and functional information in the scanned subject. Besides functional data, fMRI techniques also provide other clinically relevant physiological information (e.g., regarding biochemical status, cerebral blood compartment, perfusion, water molecular diffusion, and cerebral microstructure and fiber tracking) (Laureys et al. 2009). There are some limitations to the use of fMRI, for instance in the case of patients who have implanted materials (e.g., metallic implants) that are incompatible with the scanner. In general, the main limitation, or maybe one of the most difficult to assess, especially in case of patients with DOCs , relates to motion artifacts and the duration of the procedure. First, the scanning procedure requires an average time between 15 and 120 min. Second, the methodology used in the fMRI detection of the activated cerebral areas requires repeating the procedure several times in the same subject and/or in different subjects. According to the so-called ‘subtraction paradigm,’ the brain activation measured before the task (i.e., the control state) is confronted with the brain activation during the task (i.e., the task state), and the difference is assumed to represent the specific brain areas for the task. In order to achieve reliable data, it is necessary to repeat the experiment several times in the same person or in different persons and calculate the average of the results. In this way, it is possible to detect changes in neural activity related to mental activity avoiding the risk of confusing them with false changes resulting from noise (Laureys et al. 2002).
The scientific premise of functional neuroimaging is the functional segregation of the brain. Generally speaking, neuroscientists agree that a cortical area can be specialized for some perceptual or sensorimotor processing and that this specialization is anatomically segregated in the cortex (Laureys et al. 2009). More precisely, it is assumed that the cortical infrastructure of a single function or of a complex behavior can involve different specialized areas combining resources by functional integration between them. As a result, a deep correlation between functional integration and functional segregation is necessary for the brain activities. This coexistence of integration and segregation is the cerebral foundation for functional neuroimaging to be informative about the cerebral activity: Complex behavior can be broken down into more simple and elementary mental operations related to specific cerebral areas.
From a methodological point of view, in the case of the application of fMRI to patients with DOCs , it is important to assess the passive stimulations (i.e., when the patient is not asked to perform any task) and the active paradigms (Boly et al. 2007). Regarding the first point, according to Boly and colleagues, lacking a ‘full understanding of the neural correlates of consciousness , even a normal activation in response to passive sensory stimulation cannot be considered as proof of the presence of awareness in patients with DOCs . In contrast, predicted activation in response to the instruction to perform a mental imagery task would provide evidence of voluntary task-dependent brain activity, and hence of consciousness , in non-communicative patients’ ‘Boly et al. 2007:979’.
We will analyze the issues arising from the assumed ‘neural evidence’ of consciousness with more details below. What is relevant to note here is that the brain activation in response to passive stimulation is currently not necessarily assumed by the neuroscientific community as proof of consciousness (i.e., awareness). From an ethical point of view, this is relevant, especially regarding Informed consent . The problem is that if brain activation is the only way a patient potentially retains for communicating, but this activation cannot be assumed as proof of conscious activity, then the patient cannot be assumed to be either conscious or able to express a valid Informed consent . For this reason, further technical advancement in the detection of residual consciousness in patients with DOCs is essential in order to resolve the ethical issue of their self-determination (i.e., informed consent).
Regarding the paradigm selection, spatial navigation and motor imagery tasks have been detected as useful mental tasks to identify and assess brain activity and consciousness in patients with DOCs. This new paradigm (i.e., imagery tasks for assessing consciousness through fMRI) could be a useful tool to assess willfulness and consciousness, implement a process of communication with patients with DOCs, and overcome the limitation of the behavioral paradigm based on motor responsiveness.
Neuroimaging in general and fMRI in particular have allowed us to objectively differentiate patterns of cerebral activity in patients with DOCs (Boly et al. 2005). Detection of specific areas yielded by particular tasks is clinically relevant because it potentially gives us the possibility to develop an alternative form of communication with patients lacking the ability to speak and to move (Naci et al. 2013). The aforementioned experiment by Laureys and colleagues, for instance, shows the possibility to communicate with patients by detecting through fMRI the willful activation of specific areas in their brains (Monti et al. 2010). This possibility relies on the identification of the different brain regions and the related mental activities, which have been made possible in recent years.
On the basis of such findings, neuroscientists have defined consciousness as the emergent property of the collective behavior of widespread thalamocortical frontoparietal network connectivity (Laureys 2005a). Moreover, it has been possible to identify the different networks elicited by subjective internal self-related thoughts (self-awareness: midline cortical structures) and by external sensory perceptions (external awareness: lateral frontoparietal structures) (Vanhaudenhuyse et al. 2011). On the basis of this knowledge, an experimental paradigm has been developed in which the brain’s response to self-related stimuli such as the patient’s own name (Qin et al. 2010), and not to external stimuli, has been measured.
However, as stated above, the activation of a brain area as such is not enough to conclude that the patient is aware, since it could be a case of, for example, passive stimulation reaction or implicit learning (Laureys 2005b). The assumed condition to interpret the neuroimaging data as evidence of consciousness is a time-related condition: The activation of the cerebral area in response to a specific task has to last at least 30 s. In this way, it is possible to disentangle the cerebral activation related to a voluntary (re)action from unconscious reactions that are fleeting (Boly et al. 2005; Greenwald et al. 1996; Naccache et al. 2005). Furthermore, as emerging from the aforementioned experiment by Laureys and colleagues, correct yes–no answers to simple questions such as ‘Is your mother’s name Yolande?” confirm voluntary origin of the fMRI signal (Monti et al. 2010). Discrimination between voluntary and involuntary brain activity is ethically relevant in regard to the prospective use of neuroimaging for communicating with patients with DOCs and particularly for asking them to give an informed consent .
Research for implementing an fMRI-based communication with patients with DOCs is currently in progress. For instance, a new, noninvasive, relatively fast to apply, and reliable fMRI-based spelling device has recently been proposed as a communication tool, which is potentially promising also for patients with DOCs (Sorger et al. 2012). Yet to date, all these attempts are still at the stage of proofs of concept rather than being practical means to really ensure long-term communication. There are some technical problems in the use of fMRI-based technology to communicate with patients with DOCs. For instance, because of the severe brain damage, the coupling of hemodynamics and neuronal signal, which is at the basis of the fMRI assessment of consciousness , could be very different in patients with DOCs compared to that in healthy people. Moreover, given the plasticity of the brain, the anatomy and functional neuroanatomy could have undergone a functional remapping in patient with DOCs, so that a specific cerebral area could have been functionally replaced by another one.
For the abovementioned difficulties, EEG-based communication devices, the so-called brain–computer interfaces (BCI), are being developed as a potentially more practical, transportable, and cheaper alternative to fMRI for communicating with patients with DOCs (Bruno et al. 2011a; Sorger et al. 2003; Naci et al. 2012; Sellers 2013; Lulé et al. 2013). Other relevant results emerged from a clinical case of complete locked-in syndrome (LIS) showing consciousness via ERP (Schnakers et al. 2009a) and from the measurement of pupil size by a bedside camera to communicate with patients with locked-in syndrome (Stoll et al. 2013).
Another possibility emerging from contemporary neurotechnology is the use of TMS-EEG as a tool to probe consciousness in patients with DOCs (Casali et al. 2013; Jacobo et al. 2013). Furthermore, TMS-EEG potentially gives researchers a tool for developing a communication paradigm with patients with DOCs.
2.2 Brain–Computer Interface
BCI is a direct connection between living neuronal tissue and artificial devices that establishes a non-muscular communication pathway between a computer and a brain (Wolpaw et al. 2002). Through BCI, it is possible to detect changes in neuroelectrical activity or brain activity in response to sensory stimulation. The user is then trained to use these changes to select items, words, or letters in communication software or to make choices for neuroprosthesis control (Kübler 2009).
BCI is grounded in a continuous, real-time interaction between living brain and artificial effectors. In this way, a functional hybridization between brain and technology is realized. The operation scheme of a BCI is quite simple: The input is the user’s intent coded in the neural activity of her/his brain detected through BOLD response. The output is the device controlled by the user’s brain activity.
BCI uses a representation of the subject’s mentation as the essential component. The psychological task or the intention of the subject is detected and recorded through invasive or noninvasive methods, mostly EEG using surface or implanted electrodes, but also MEG, fMRI, or functional near infrared spectroscopy (fNIRS). There is a significant difference between these methods regarding the ease of use. For instance, while MEG and fMRI are more demanding, require quite sophisticated instruments, and are quite expensive, EEG, NIRS, and invasive systems are portable and thus suitable for use in daily life (Kübler 2009).
In the particular case of patients with DOCs , EEG offers significant comparative advantages on the aforementioned points. Furthermore, it can be useful to develop EEG-BCI systems that can be used at the bedside to detect volitional brain activity and to enable basic communication.
Thus, to date, EEG-based techniques are the most suitable BCIs for clinical application to patients with DOCs even if other technologies, such as fMRI, allow a more detailed spatial resolution and a more precise allocation of neuronal activity than EEG. Whatever technology is used, the detected and recorded cerebral signals are digitized and differently processed by filtering, amplitude measurement, and spectral analysis (Wolpaw et al. 2002). Specific algorithms then translate the processed signals into commands expressing the users’ will. In particular, the subject may communicate choosing the words on a screen moving a cursor through his own mind. In this way, BCI provides subjects with a virtual keyboard where the user can press the keys through the brain activity’s modulation.
Importantly, BCI provides the user with real-time feedback on their performance, giving her/him the possibility to improve the use of the BCI over time. BCI thus enables a cerebral communication without motor response. This cerebral communication could give to some behaviorally non-responsive patients, such as patients with DOCs , a new opportunity to communicate.
There are several prerequisites to use BCI for communicating with patients with DOCs . The patient should be able to properly understand verbal commands. The patient should also be able to react to external stimulation and express her/his answers through a minimal form of communication (e.g., a binary yes/no communication) while remaining sufficient cognitive capacities enabling the formulation of a reliable informed decision (Lulé et al. 2013). It is possible that patients retain the ability to partially understand commands, to understand but not to follow commands, or to understand and to follow commands but not well enough to make BCI feasible. In order to use a BCI with patients with DOCs , the understanding of the provided information should be matched with their ability to attend to stimuli, to selectively process the salient ones, and to retain information in working memory (Chatelle et al. 2012).
The results emerging from the aforementioned studies by Laureys, Owen, Schiff and others are relevant and promising also in the direction to use BCI with patients with DOCs . A possible communication protocol through BCI emerges from the experiments by Cruse and colleagues (Cruse et al. 2011). They investigated the capacity of patients with DOCs to perform mental motor tasks that are possible to differentiate in their EEG at the single-trial level. Sixteen patients in VS/UWS were asked to imagine squeezing their right hand or moving all their toes, and in 19 % of the patients a support vector machine predicted the task being executed with an accuracy of between 61–78 %. The same test was performed with MCS patients, where 22 % of them were able to follow commands using motor imagery (Cruse et al. 2012). Starting from these results, it could be possible to implement a binary communication by assuming imagination of right hand as ‘yes’ and the imagination of toe movement as ‘no’ (Chatelle et al. 2012).
Another relevant study has been conducted by Lulé and colleagues who tested an EEG-BCI paradigm on 16 healthy subjects and 18 patients in a VS/UWS, in a MCS, and in LIS (Lulé et al. 2013). The results of the study showed that 13 healthy subjects and 1 LIS patient were able to communicate through BCI, and 1 patient in MCS who was unresponsive at the bedside showed command following with the BCI, while all other patients did not show any response to command and could not communicate through BCI. Even if no patients with DOCs were able to functionally communicate through BCI, this study is relevant and promising in showing command following in one patient in MCS.