In addition to detecting brain activity externally, it is possible to implant devices directly into the brain. The advantage is that the electrodes are much closer to the source of electrical activity, so the signal-to-noise ratio is greatly improved and the localization provides a more direct idea of the origin of the activity (Vetter et al., 2003). A number of such devices have been developed (primarily in the United States), notably BRAINGATE, the Utah Probe, and the Michigan Probe (Archibald, 2005; Huys et al., 2011; Kipke et al., 2011). Such invasive probes or arrays of an EEG type of electrodes on the surface of the brain are contemplated only for patients who have severe disorders. The smallest probes currently available are much larger than individual neurons; serious damage is done to brain tissue on insertion, and the process is irreversible (He et al., 2006). It may be justified for severely disabled patients, but seems unlikely to be acceptable as an enhancement method in the near term. Major efforts are under way in a number of laboratories to integrate electrodes or meshes of electrodes with neurons (and other cell types) in culture, and this is potentially a less destructive approach (Lee et al., 2004). It has potential for repair of damaged neuronal tissue, including the possibility of using radiofrequency electronics so that signals can be read—or regions can be stimulated—remotely.
An alternative detection system is related more directly to physiology, particularly blood flow and glucose metabolism associated with activity in a particular region of the brain (Son and Yazici, 2005). That is the basis of functional magnetic resonance imaging (fMRI), positron-emission tomography (PET), and near-infrared functional (NIRF) imaging. Although fMRI and PET imaging are major imaging modalities in clinical medicine and brain research, neither can currently be implemented in a portable or even semiportable format (Son and Yazici, 2005).
NIRF imaging uses changes in the near-infrared absorption of deoxyhemoglobin vs. oxyhemoglobin to obtain information on changes in blood oxygenation (Cui et al., 2011; Hoshi, 2011). Blood oxygenation changes during high levels of activity, and substantial changes are observed in brain tissue. Arrays of near-infrared sources (for example, laser diodes) and detectors in combination can be used to generate low-resolution images of brain function. The penetration of near-infrared light through the skull of adults is modest (it is much higher in premature infants, and this technique is used in the neonatal intensive care unit).
Using brain signals to monitor cognition for use in training has demonstrated benefits. Some groups are pursuing augmented cognition whereby human performance is improved by designing for memorability. It is necessary for cognitive bottlenecks and firing patterns to be measured to detect cognitive states within some small amount of time (Son and Yazici, 2005). If that can be accomplished, perhaps behavior can be influenced.
Although a great deal of progress has been made in this realm, how the brain functions is still largely unknown (Son and Yazici, 2005; Hoshi, 2011). Even if neural events could be reliably correlated to the detection of sensory inputs or to the stimulation of specific motor activity, there is a fundamental issue of neuromuscular mismatch (Kirilina et al., 2012). The brain has evolved to control muscles, whereas BCI research strives to create interfaces between the brain and computers. Current neural implants are rather crude, and a more biological way to create an interface with the brain to enhance efficacy and decrease scarring is being sought and may be found in the next 20 years. The most promising direct applications of this research to human performance modification (HPM) are in developing therapies for existing disabilities. Even in those cases, the improvements do not approach the functionality of a healthy person. It is unlikely that human performance can be substantially enhanced via BCI in the near term.