| People are typically able to reach confident decisions despite high degrees of ambiguity, ambivalence, and contradiction among competing options. Prior research (Holyoak & Simon, 1999) indicates that preferences toward arguments that support one's decision are often adjusted over the course of reaching the decision. The result of this process is that people emerge with a coherent decision in which supporting arguments are biased toward their favored position relative to their original preference values. Three experiments are presented in which subjects must decide an ambiguous legal case. Participants' ratings of agreement toward the arguments of the case were recorded both before and after a verdict was reported. Experiment 1 tests whether a corrective decision strategy can attenuate the belief shift resulting in more objective decisions. Results indicate that the use of a strategy in which participants must consider the alternative position, both prior to deciding and prior to reporting post-verdict agreement, attenuated the level of decision-reinforcing bias among the agreement ratings. Experiment 2 tests whether the application of time pressure at the time of post-verdict ratings can attenuate attitude shifts. Results indicated that this manipulation attenuated decision-reinforcing coherence bias by a relatively small degree. Experiment 3 investigates the extent to which agreement shifts can be disrupted by working memory dual-tasks taxing the phonological loop component with articulatory suppression and the central executive component with a verbal trails switching task. Results demonstrated that the verbal trails dual-task disrupted coherence bias to a larger extent than articulatory suppression and control conditions. Additional analyses of the verbal trails condition revealed that participants who were successful at the dual-task showed a trend toward coherence bias, while those who were unsuccessful showed no coherence bias. These results indicated that retrieval of biased evaluations may be the most critical aspect of reporting coherence in argument evaluations. In sum the results indicate that coherence shifts are highly automatic, but can be attenuated by a strategy of considering the alternative position and by executive load conditions. |
