Fluctuations in Cognitive and Perceptual Activity

University Of Texas At Austin, Austin TX

Investigators

Abstract

Whenever a scientist measures human behavior there are always two levels of structure in the measurements. The first is generally referred to as the signal or the aspect of human behavior that is of interest. Measurement of the signal is generally the point of conducting an experiment. The second level of structure is the noise or error and it is usually not something that is purposefully sought and its analysis tends to be cursory. In standard statistical analysis, say for example in the general linear model, this aspect of the data is referred to as unexplained measurement variability. With NSF support Dr. David Gilden will pursue a fundamental and controversial understanding of this so-called noise or unexplained variability. Dr. Gilden begins with the hypothesis that the noise attending measurement may be of considerable interest in its own right and may contain important clues about what kind of thing human activity is. Previous work has shown that errors in production and fluctuations in reaction time may exhibit a particular kind of fractal structure known as 1/f noise. This particular noise is of intense interest in physics, biology, economics, and medicine for it is both generic and unexpected. In fact, at this time there is no agreement on how natural systems produce 1/f noise. The purpose of this proposal is to clarify the nature of this 1/f noise in measurements of human behavior. This will be done by a careful dissection of the components of psychological measurement into its perceptual demands, task demands, response mappings, and overt response. The second step is to characterize how the dynamic moves forward, whether it is tied to ongoing physiological fluctuations or whether it is uniquely activated on a measurement by measurement basis.