Keith Clayton, Vanderbilt University, Nashville, TN
Barbara Bruhns Frey, Vanderbilt University, Nashville, TN

Abstract: Successive tests of subjects' performance in reaction time tasks were treated as time series and submitted to spectrum analysis. Rather than revealing the "white" noise expected by the view that variability is due to random error (equivalent power across frequency), the power spectra revealed "colored" noise. The slopes of the power spectra did not vary much for tasks differing in memory requirements. Spectrum analysis of time series from the logistic map also showed colored noise in regions on the "edge" of chaos, showing that the presence of colored noise in cognitive data need not oblige a theoretical account based on a complex, high-dimensional, system.

Keywords: pink noise, spectrum analysis, choice tasks, mental measurement, memory