Advice Taking under Uncertainty: The Impact of Genuine Advice versus Arbitrary Anchors on Judgment

A major module of rational advice taking consists in the metacognitive ability to distinguish between credible advice and arbitrary anchors. Accordingly, we investigated the extent to which framing the very same information as either advice or anchor exerts a differential influence on quantitative judgments. Four experiments showed that although arbitrary anchors were given lower weight than advice, they nevertheless exerted a systematic impact on final judgments. Degree of integration was related to subjective confidence only in the advice condition, but not in the anchoring condition, suggesting that arbitrary anchors were not considered informative. Framing the source of advice as a human being versus as a computer did not affect our results. Only the aboutness of advice, that is, whether it targeted the focal judgment item, determined its influence on final judgments and on confidence. On the one hand, these findings speak to the (partial) sensitivity of human judges to the source and validity of advice under uncertainty. On the other hand, the persevering effect of arbitrary anchors demonstrates the dependence of judgments on unauthorized influences. Both findings together highlight the need to study advice taking from a metacognitive perspective.理性建议采纳的核心模块之一，是区分可信建议与任意锚点的元认知（metacognitive）能力。据此，我们探究了将完全相同的信息分别以建议或锚点的框架呈现时，会对数量判断产生何种差异化影响。四项实验结果显示：尽管任意锚点的权重低于建议，但其仍对最终判断产生了系统性影响。信息整合程度仅在建议实验条件下与主观置信度相关，而锚定实验条件下则无此关联，这表明任意锚点未被决策者视为有效信息。将建议来源设定为人类或计算机并未对实验结果产生影响。唯有建议的针对性（即是否指向核心判断项）决定了其对最终判断与置信度的影响。一方面，本研究结果印证了人类决策者在不确定性情境下，对建议来源与有效性具备（部分）敏感性；另一方面，任意锚点的持续影响效应，也证明了判断会受到未被授权的干扰因素影响。两项发现共同凸显了从元认知视角研究建议采纳行为的必要性。