An investigation of the "randomness" of threshold measurements.

Abstract

Do thresholds vary randomly in time, or do they not? If not, what are the characteristics of their variations? To try to answer these questions, successive measurements of auditory, visual, and pain thresholds were obtained at 6-sec, 1 min., 3-min., and 1-day intervals on a series of human subjects. These psychophysical thresholds were subjected to 5 different kinds of analysis: autocorrelation, estimated power-spectrum transformation of the autocorrelation function, correlation of the thresholds of 2 modalities measured over the same period of time on the same subject, correlation of the spectra of the thresholds of 2 modalities taken over the same period of time on the same subject, and analysis of variance. The data obtained tended to confirm the hypothesis that variations of thresholds in time, under normal conditions, are not "random." This evidence is: (a) reliable changes in thresholds can be shown to occur in a subject from day to day; (b) changes in the thresholds of different modalities are slightly correlated; (c) successive threshold measurements are not independent; (d) some thresholds can be shown to undergo cyclical drifts; and (e) the patterns of variations of different modalities over same time interval in same subject are not unrelated. The relevance of these findings to biological measurements in general, and to an approach to the total organism is discussed.