Brain–computer interfaces (BCI) allow control of computers or external devices with regulation of brain activity alone. Invasive BCIs, almost exclusively investigated in animal models using implanted electrodes in brain tissue, and noninvasive BCIs using electrophysiological recordings in humans are described. Clinical applications were reserved with few exceptions for the noninvasive approach: communication with the completely paralyzed and locked-in syndrome with slow cortical potentials, sensorimotor rhythm and P300, and restoration of movement and cortical reorganization in high spinal cord lesions and chronic stroke. It was demonstrated that noninvasive EEG-based BCIs allow brain-derived communication in paralyzed and locked-in patients but not in completely locked-in patients. At present no firm conclusion about the clinical utility of BCI for the control of voluntary movement can be made. Invasive multielectrode BCIs in otherwise healthy animals allowed execution of reaching, grasping, and force variations based on spike patterns and extracellular field potentials. The newly developed fMRI-BCIs and NIRS-BCIs, like EEG BCIs, offer promise for the learned regulation of emotional disorders and also disorders of young children.

A brain–computer interface (BCI) or brain–machine interface (BMI) activates electronic or mechanical devices with brain activity alone. BCIs and BMIs allow direct brain communication in completely paralyzed patients and restoration of movement in paralyzed limbs through the transmission of brain signals to the muscles or to external prosthetic devices. We differentiate invasive from noninvasive BCIs: Invasive BCIs use activity recorded by brain implanted micro- or macroelectrodes, whereas noninvasive BCIs use brain signals recorded with sensors outside the body boundaries.

History of BCI Research

Hans Berger, who discovered the human EEG, speculated in his first comprehensive review of his experiments with the ‘‘Elektrenkephalogramm’’ (1929) about the possibility of reading thoughts from the EEG traces by using sophisticated mathematical analyses. Grey Walter, the brilliant EEG pioneer who described the contingent negative variation (CNV), often called the ‘‘expectancy wave,’’ built the first automatic frequency analyzer and the computer of ‘‘average transients’’ with the intention of discriminating covert thoughts and language in the human EEG (Walter, 1964). Fetz (1969) published the first paper on invasive operant conditioning of cortical spike trains in animals. Only the recent development of BCIs, however, has brought us a bit closer to the dreams of these pioneers of EEG research. 

Invasive and noninvasive BCIs originate from different research traditions, though both have their roots in animal experiments. Invasive BCIs consist of implanted multielectrode grids inthe motor cortex of paralyzed patients (Donoghue, 2002), premotor cortex of monkeys (Carmena et al., 2003), or parietal motor command areas (Schwartz et al., 2001). They try to reconstruct intended skilled movements from neuronal firing patterns online. Based on ‘‘sparse coding’’ approaches to motor learning (Riehle & Vaadia, 2005) and directional coding vectors of motor neurons (Georgopoulos, Schwartz, & Kettner, 1986), automatized complex movements can be reconstructed online from relatively few motor neurons using simple algorithms:

Nicolelis’ group (Carmena et al., 2003) demonstrated in monkeys after extensive training of a reaching and grasping movement that firing patterns of 32 neurons are sufficient to execute that movement directly with an artificial limb.

Chapin, Moxon, Markowitz, and Nicolelis (1999) trained rats tomove a leverwith an artificial arm in a Skinner box for reward with extracellular firing of cortical cells without any actual movement. The neuronal firing pattern that used to precede and accompany the lever pressing response alone was able to operate on the lever delivering the reward.

Emotion and Quality of Life in ALS and Paralysis

Most ALS patients opt against artificial respiration and feeding and die of respiratory problems. In many countries, doctors are allowed to assist the transition with sedating medication to ease respiration-related symptoms. If doctor-assisted suicide or euthanasia is legal, as it is in the Netherlands and Belgium, very few patients vote for continuation of life. The vast majority of family members and doctors (usually neurologists) believe that the quality of life in total paralysis is extremely low and continuation of life constitutes a burden for the patient and that it is unethical to use emergency measures such as tracheostomy to continue life. The pressure on the patient to discontinue life is enormous. The facts on end-of-life issues and quality of life do not support hastened death decisions in ALS, however, and the scientific literature and our own studies challenge the pervasive myth of helplessness, depression, and poor quality of life in respirated and fed paralyzed persons, particular with ALS (Albert, Rabkin, Del Bene, Tider, & Mitsumoto, 2005; Quill, 2005). Most instruments measuring depression and quality of life such as the widely used Beck or Hamilton depression scales are invalid for paralyzed people living in protected environments because most of the questions do not apply to the life of a paralyzed person (‘‘I usually enjoy a good meal,’’ ‘‘I like to see a beautiful sunset’’). Special instruments had to be developed for this population (Ku¨ bler, Winter, et al., 2005). In studies by Breitbart, Rosenfeld, and Penin (2000) and by our group (Ku¨ bler, Winter, et al., 2005) only 9% of the patients showed long episodes of depression, most of them in the time period following the diagnosis and a period of weeks after tracheostomy. 

It could be argued that questionnaires and interviews reflect more social desirability and social pressure than the ‘‘real’’ behavioral–emotional state of the patient. The social pressure in LS, however, directs the patient toward death and interruption of life support. The data, therefore, may underestimate the positive attitude in these groups. This hypothesis is strongly supported by a series of experiments with ALS patients at all stages of their disease using the International Affective Picture System (IAPS; Lang, Bradley, & Cuthbert, 1999). Lule´ et al. (2005) and Lule´ et al. (in press) using a selection of pictures with social content, found more positive emotions to positive pictures and less negative ratings to negative pictures in ALS than in matched healthy controls. Even more surprising are the brain responses to the IAPS slides. FMRI measurement in 13 patients with ALS and controls demonstrated increased activation in the supramarginal gyrus and other areas responsible for empathic emotional responses to others comparable to the ‘‘mirror neuron network’’ identified first by Rizolatti and colleagues (Gallese, Keysers, & Rizzolatti, 2004). Furthermore, brain areas related to the processing of negative emotional information such as the anterior insulae and amygdala show less activation in ALS.

These differences become stronger with progression of the disease 6 months later. One is tempted to speculate that with progression of this fatal disease, emotional responding on the behavioral and central nervous system level improves toward positively balanced social cues, resulting in a more positive emotional state than in healthy controls! The positive responding and positive interaction of the social environment and caretakers to a fatally ill, paralyzed person may, in part, be responsible for the prosocial emotional behavior and for the modified brain representation predicted by social learning theory (Bandura, 1969). Taken together, the results on emotional responding and quality of life in paralyzed ALS patients suggest a more cautious and ethically more responsive approach toward hastened death decisions and last-will orders of patients and their families. The data reported here also speak pervasively for the usefulness and necessity of noninvasive BCI in ALS and other neurological conditions leading to complete paralysis.

The second major field of BCI research concerns restoration of movement in patients with paralysis,mostly spinal cord lesions, chronic stroke, and other movement disorders. It is certainly an attractive possibility to build a direct connection between voluntary movement command centers in the brain and the periphery isolated from these regions by acentral, spinal, or peripheral lesion.

Invasive ad Noninvasive BCIs for Restoration of Movement

Brain–computer interface research received its impetus from animal research reconstructing movement from microelectroderecorded spike trains or synaptic field potentials (Donoghue, 2002; Nicolelis, 2001). After extensive training and the implementation of learning algorithms (for an exception, where animals learned rapidly, see Serruya et al., 2002), monkeys move cursors on screens toward targets or an artificial hand moves in four directions directed by spike activity, demonstrating the possibility of translating cellular activity into simple movements online. After such training, even complex movement patterns can be reconstructed froman astonishingly small number of cells located in the motor or parietal areas (Musallam, Corneil, Greger, Scherberger,&Andersen, 2004;Nicolelis, 2001; Schwartz et al., 2001; Taylor et al., 2002). The plasticity of the cortical circuits allows learned control of movements directly from the cellular activity even outside the primary or secondary homuncular representations of the motor cortex (Taylor et al., 2002). A multielectrode array recording spike and field potentials simultaneously was implanted in a single quadriplegic patient’s motor hand area (2004) by Donoghue’s group (personal communication, April 2005). Within a few training sessions, the patient learned to use neuronal activity from field potentials to move a computer cursor in several directions comparable to the tasks used for multidimensional cursor movements in the noninvasive SMR-BCI reported byWolpaw and McFarland (2004). None of the invasive procedures allowed restoration of skillful movement in paralyzed animals or people in everyday-life situations. The animals studied in BMI research (Nicolelis, 2003) were all intact animals who learned to move an artificial device or curser for food reward without moving their intact arm in highly artificial laboratory situations. Any generalization from the invasive animal BCI approach to paralyzed people is premature.

Future Directions: The Metabolic Whole Brain BCI

Weiskopf et al. (2003) for the first time demonstrated convincingly that healthy persons are able to regulate BOLD (blood oxygen level dependent) responses from circumscribed cortical and subcortical brain regions using online functional magnetic resonance imaging (fMRI-BCI). These authors and others (DeCharms et al., 2005) demonstrated substantial effects of BOLD-response BCI training on behavior: Pain, emotional arousal, and memory were investigated and astonishingly strong effects on the behavioral variables after short training periods with fMRI-feedback training were shown. This is not surprising, considering that vascular changes in brain arteries and veins responsible for metabolic responses such as BOLD and brain blood flow may allow superior voluntary (operant) control because of the vascular-motor component of the physiological target response. Dilation and contraction of vascular changes are sensed by the brain and regulated by neural structures with closely coupled autonomic and somatic-motor functions, allowing access to voluntary control (Dworkin, 1993).

The results presented by Weiskopf et al. (2004), Weiskopf, Klose, Birbaumer, and Mathiak (2005), and Weiskopf, Scharnowski, et al. (2005) constitute the first step in the application of fMRI-BCI to emotional disorders: fMRI allows anatomically specific control of subcortical and cortical areas responsible for the regulation of emotions not as accessible to electrophysiological methods as EEG and MEG such as amygdala, limbic insular and cingulate regions, and anterior basal ganglia. Clinical application of fMRI-BCI is presently unrealistic and unlikely, considering the cost and technological difficulties involved in real-time fMRI. It will, at present, remain reserved for research purposes and experiments intending to demonstrate effects of learned local blood-flow changes on emotional and motivational behavior. A clinically more realistic new metabolic BCI system has been proposed and tested recently by Sitaram et al. (in press). These investigators used near-infrared spectroscopy (NIRS) andmeasured, with optical recording devices, changes in cortical oxygenation and deoxygenation. Using the reflection of light in living tissues with high circulation density such as the brain, NIRS is completely noninvasive (Coyle et al., 2004). NIRS devices are also relatively inexpensive (price equivalent to that of a multichannel EEG) and commercially available. Another virtue of NIRS is portability, allowing, for example, the training of young children. Sitaramet al. (in press) demonstrated inline operant control of sensorimotor brain areas in five healthy subjects and spelling of letters with NIRS-BCI with an accuracy of 70%–95% after only two training sessions and with information transfer speed comparable to EEG-BCI.