Does the memory effect influence individual exponents in psychophysical power functions?

Abstract

Two groups of subjects made magnitude estimations of a series of eight circles, each of which had its own color. All subjects received the same number of presentations in both sessions, with half the subjects receiving the same size-color pairing of the first session, and the remaining subjects receiving a different size-color pairing, in the second session, one week later. Interrepetition correlations of exponents for the middle four adjacent presentations were computed. All of the correlations were significantly different from zero with the exception of the intersession correlation of the control group. The control group exhibited a significant decrease in correlation across sessions while the experimental group remained stable. These findings were totally unexpected. A test of difference between intersession correlations was not significant. The results can not easily be explained within the framework of either the response bias or memory hypotheses.