Measurement effects in decision-making

Abstract

When participants are shown a series of stimuli, their responses differ depending on whether they respond after each stimulus or only at the end of the series, in what we call a measurement effect. These effects have received paltry attention compared with more well-known order effects and pose a unique challenge to theories of decision-making. In a series of two preregistered experiments, we consistently find measurement effects such that responding to a stimulus reduces its impact on later stimuli. While previous research has found such effects in noncumulative tasks, where participants are instructed only to respond to the most recent stimulus, this may be the first demonstration of these effects when participants are asked to combine information across either two or four stimuli. We present modeling results showing that although several extant classical and quantum models fail to predict the direction of these effects, new versions can be created that can do so. Ways in which these effects can be described using either quantum or classical models are discussed, as well as potential connections with other well-known phenomena like the dilution effect.