Conscious perception of brain states: mental strategies for brain–computer communication

Abstract

Direct brain–computer communication utilises self-regulation of brain potentials to select letters, words or symbols from a computer menu. In this study a completely paralysed (locked-in) patient learnt to produce slow cortical potential (SCP) shifts to operate a binary spelling device. After hundreds of training sessions he gave a detailed description of his mental strategies for self-regulation. His cognitive strategies matched with the electrocortical changes perfectly. Thus he produced a contingent negative variation (CNV) with images of preparation such as an arrow being drawn on a bow. To produce a positive potential shift he imagined the arrow shooting up from the bow. To suppress potential shifts he tried to stop thinking. The study demonstrates that patients become sensitive for their brain states with increasing self-regulation practice. The use of conscious cognitive strategies may, however, be incompatible with the complete automatization of the self-regulation skill.

Introduction

Several neurological diseases, such as brain stem infarct or amyotrophic lateral sclerosis (ALS), can lead to severe or total motor paralysis, referred to as locked-in syndrome: the intact brain is locked into a paralysed body. Locked-in patients experience a loss of all motor functions, whereas cognitive abilities often remain largely unimpaired [24], [33]. As Bauby [2] described in his famous book “The Diving Bell and the Butterfly”, one of the most terrifying aspects of the locked-in syndrome is the inability to communicate: the patients are unable to tell their feelings and wishes and even the expression of the most basic needs is impossible. To re-establish or maintain communication in these patients, brain–computer interfaces (BCI), i.e. direct connections between brain and computer, have been developed [19]. Compared to communication systems based on muscular function, BCI are still slow (between 0.15 characters and 12 words per minute, see [19]). However, muscle dependent communication systems are not feasible for some locked-in patients, so that a BCI is the only possible communication channel. Communication is an important aspect of the will to live in people with such severe disabilities. Bach [1], e.g. showed that establishing and maintaining effective communication greatly increased the quality of life in ALS patients.

The currently most advanced BCI make use of the fact that humans can acquire control over EEG parameters by means of neurofeedback, i.e. real-time visual or auditory feedback of a specific EEG parameter. A typical arrangement of a BCI is depicted in Fig. 1. Existing BCI are based on the self-regulation of the action potential firing rate [15], the 8–12 Hz mu rhythm [25], [36], or slow cortical potentials (SCP) [4], [18]. Strengths and limitations of these systems are described in [19].

The thought translation device (TTD) [4] is a BCI which is based on the self-regulation of SCP. SCP reflect changes in cortical polarisation of the EEG lasting from 300 ms up to several seconds. Functionally, SCP reflect a threshold regulation mechanism for local excitatory mobilisation. Negative SCP shifts, like the Bereitschaftspotential [16] or contingent negative variation (CNV) [34], [35], indicate local excitatory mobilisation, whereas positive potential shifts indicate disfacilitation [3]. Neurophysiologically, negative surface SCP result from a sink caused by synchronous slow excitatory postsynaptic potentials in the apical dendrites of layers I and II in the cortex with a source being located in cortical layers IV and V. The origin of positive SCP is less understood, but may result from sinks in deeper layers or inhibitory sources in cortical layers I and II. SCP are sensitive to several artefacts of non-neuronal origin such as eye-movements, sweating, electrode polarizations, and p(CO₂) changes in the underlying tissue. Special care has to be applied to avoid or correct these artefacts [2]. In the TTD, patients are required to produce voluntary SCP shifts of either positive or negative amplitude, thereby moving a cursor on a notebook screen in two directions. This binary signal can be used to operate applications such as switches in the environment or an electronic communication system. In a sequence of training steps [20], patients can learn to self-regulate their SCP with high accuracy, enabling them to select phrases, words or single letters for verbal communication. Patients have communicated numerous messages by means of this technique [4], [5], [19].

Although it is well known that individuals can acquire self-control over physiological functions, it is still a controversial question how they learn it. Generally, it is believed that the self-regulation of a physiological parameter is acquired according to operant learning principles. After the presentation of a discriminative stimulus (e.g. the highlighting of a target on a computer screen), the correct behaviour (e.g. the production of a negative SCP shift which moves a cursor toward the indicated target) is reinforced with a rewarding stimulus (e.g. money or points). Positive reinforcement leads to the more frequent occurrence of correct behaviour. Some authors have dealt with the question which specific mechanisms are involved in the acquisition of self-regulation of a physiological parameter. It was suggested that individuals have to “perceive” a physiological function in order to regulate it [8], [9], [10]: by presenting external feedback of the parameter in question contingent upon internal sensations evoked by self-regulation of this parameter, an internal “response image” is created whose activation leads to the production of the required response. However, the internal response image does not necessarily become conscious. Another theory states that different cognitive or behavioural strategies from the existing behaviour repertoire are tested until a strategy for self-regulation is detected [21], [22]. Recently, it was found that children used identical mental strategies to self-regulate their SCP into opposite directions [32]. This seems to indicate that conscious mental strategies do not influence the course of SCP. The results of Kotchoubey et al. [17] point at the same direction. Patients who learnt to self-control their SCP, could rate their performance correctly, but only after they had already acquired the self-regulation. Thus the conscious perception of a brain response follows its control and not vice versa.

In this study, a locked-in patient described his mental strategy to self-control his SCP. Since he has been trained over a period of 4 years, he has gained considerable insight into his technique. He described his mental strategies in detail by means of the thought translation device. Thus the complete text was written using self-regulation of SCP only. At no point in time other means of communication were added and no help of trainers or attendants was provided. This constitutes the first description of the mental strategies used to self-regulate SCP in a completely paralysed patient, thus excluding muscular or respiratory mediation of the learned response. The subjective mental strategy will be compared with the actual EEG pattern to examine if the neural activity during SCP self-regulation is reflected in the cognitive processes.