Sequential Predictors of Choice in Psychophysical Tasks
Gregory R. Lockhead Duke University Department of Psychology Durham, North Carolina
The responses people give to stimuli depend on prior events. In psychophysical judgment tasks, responses are positively correlated with the just previous stimulus or response or feedback. Too, those responses are either positively or negatively correlated with earlier events in the sequence. Which occurs depends, predictably, on details of the experiment. It is suggested here that these independent variables (prior stimuli, prior responses, and prior feedback) are not the source of the observed sequential effects. Instead, these variables are all correlated with one another and all seem to reflect an underlying memory scale for prior events. It is suggested that the manner in which people use and maintain this memory scale produces the persistent sequential effects. A model based on this possibility, which requires two assumptions, is shown to account reasonably well for the various observed contextual effects.
Human decision making takes place in the context of past experience as represented in memory. If this is a correct assertion, then different patterns of past experiences can result in different decisions. This means that if we knew how judgments were related to prior events we would not only have information about what was in memory but we might also be able to infer how information in memory was being used in making decisions. Some of the ways in which prior events predict choice are discussed in this article by examining trial-to-trial performance in simple psychophysical tasks.
Psychophysical tasks were chosen to examine the questions addressed here for several reasons: (1) Psychophysical methods are historically important pro-