Choice changes preferences, not merely reflects them: A meta-analysis of the artifact-free free-choice paradigm.

Abstract

One of the prominent, by now seminal, paradigms in the research tradition of cognitive dissonance (Festinger, 1957) is the free-choice paradigm developed by Brehm (1956) to measure choice-induced preference change. Some 50 years after Brehm introduced the paradigm, Chen and Risen (2010) published an influential critique arguing that what the paradigm measures is not necessarily a choice-induced preference change, but possibly an artifact of the choice revealing existing preferences. They showed that once the artifact is experimentally controlled for, there is either no or very little evidence for choice-induced preference change. Given the prominence of the paradigm, this critique meant that much of what we thought we knew about the psychological process of cognitive dissonance might not be true. Following the critique, research using the paradigm applied various corrections to overcome the artifact. The present research examined whether choice truly changes preferences, or rather merely reflects them. We conducted a meta-analysis on 43 studies (N = 2,191), all using an artifact-free free-choice paradigm. Using different meta-analytical methods, and conceptually different analyses, including a Bayesian one, we found an overall effect size of Cohen’s d = 0.40, 95% confidence interval (CI) [0.32, 0.49]. Furthermore, we found no evidence for publication bias as an alternative explanation for the choice-induced preference change effect. These results support the existence of true preference change created by choice.

Transient Value Refinements during Deliberation Facilitate Choice

“…SoA is based on the observation that the difference in the subjective value ratings that people give to options in a choice pair is typically larger after the choice than before (i.e., the values "spread" apart). This effect is highly robust (e.g., Carlson & Russo, 2001;Chammat et al, 2017;Holyoak & Simon, 1999;Izuma et al, 2015;Izuma et al, 2010;Lee & Coricelli, 2020;Lee & Daunizeau, 2020Lee & Holyoak, 2021;Russo, Medvec, & Meloy, 1996;Salti et al, 2014;Sharot et al, 2012;Sharot, De Martino, & Dolan, 2009;Sharot, Velasquez, & Dolan, 2010;Voigt et al, 2019;Voigt, Murawski, & Bode, 2017; see Enisman, Shpitzer, & Kleiman, 2021 for a meta-analysis).…”

Cognitive dissonance from 2 years of age: Toddlers', but not infants', blind choices induce preferences

“…Similar blind choice-induced preferences have been found in adults (e.g., Enisman et al 2021) and older children (Egan et al 2010). They are considered a critical test of cognitive dissonance theory as the fact that these preferences arise even when participants are choosing blindly ensures that their choice and later preference could not have resulted from any actual initial preference for the chosen object (Chen & Risen, 2010;Izuma & Murayama, 2013).…”

“…Together with our findings of the emergence of blind-choice induced preferences around 2 years of age, these findings raise the question of what underlies this developmental change from known choice-induced preferences in infants to blind choice-induced preferences in 2-year-olds. If young infants' preferences are indeed already influenced by cognitive dissonance at 10 to 20 months, why then would they not show these choice-induced preferences when their choices were blind, as adults and older children do (Enisman et al, 2021;Egan et al 2010)?…”

“…It has been shown, however, that we do not only choose what we prefer, but we also prefer what we choose: We tend to adapt our preferences to what we have already chosen. Critically, we do so even if we had no initial preference for the chosen option, but were forced to choose one of them (Brehm, 1956;Festinger, 1957;Enisman, Shpitzer, Kleiman, 2021). Such choice-induced preferences have been argued to result from an effort to reduce cognitive dissonancean inner discomfort thought to arise when our choices are incongruent with our preferences or values (Festinger, 1957).…”

“…For example, when participants are forced to choose between two options that they initially rated as equally attractive, as a consequence of their choice, they devalue the discarded option and increase their rating of the chosen option (e.g., Brehm, 1956). Critically, it has been shown that this is even the case when their choice is blind (i.e., when the chooser did not know the options from which they were choosing, e.g., Sharot et al, 2010;Izuma & Murayama, 2013;Enisman, Shpitzer, Kleiman, 2021). The blind-choice experiments demonstrate that the observed preferences are a true consequence of the act of choosing itself, and cannot reflect any initial preference that might have guided the participant's choice (Chen & Risen, 2010;Izuma & Murayama, 2013).…”