A major concern in measuring brain function from active ambulatory people is loss of data because of unrecoverable contamination due to artifacts. Automated methods for artifact detection and decontamination have improved over the years [23], [24], [32]–[36] and are helpful in reducing human labor, improving consistency and recovering some of the data that would otherwise have to be discarded. However, such methods are still far from perfect and it is still essential that all data be subjected to labor-intensive expert manual review. In the current studies, data attrition increased markedly when recording under real world conditions. In Studies 1 and 2 in which participants were seated, stared straight ahead at a video monitor, made constrained limb movements and did not talk, ∼4% of the data to be analyzed was lost. By contrast, in Study 3, ∼40% of the data to be analyzed could not be recovered due to artifacts generated by chewing, swallowing, talking and moving around during the cocktail party. Yet, even so, it is encouraging that the majority of the data was valid under such conditions. Future improvements in automated artifact decontamination methods will hopefully produce higher rates of valid data.
