人们在进行有关数值的判断时可能会受到事先考虑到的特定数值的大小影响，导致判断结果偏向该数值的现象被称为锚定效应（anchoring effect）。锚定效应是一种普遍存在且十分顽固的判断偏差，对人们生活中理性判断和决策有着重要影响。本研究采用了锚定效应的“比较-判断”研究范式探究了老年人和年轻人在人际间情绪判断中的锚定效应的特点。在实验中，让老年人和年轻人阅读情绪情景故事之后将故事主角感受到的情绪强度与锚定值做比较，然后再给出具体的情绪强度估计值，考察情绪强度估计值是否会随锚定值变化而变化，即是否出现锚定效应。

前两个研究基于锚定效应的选择通达模型，探讨了锚定信息的特征对锚定效应的影响。研究一关注了锚定信息的任务关联性对锚定效应的影响，研究采用2（年龄组：老年，年轻；被试间） × 2（锚定值：高锚，低锚；被试间）× 2（情绪效价：积极情绪，消极情绪；被试内）× 2（任务关联性：任务关联，任务无关联；被试内）混合设计；招募了65名老年人（年龄：M=64.62±3.82）和68名年轻人（年龄：M=22.21±2.89）；任务关联条件下比较问题与判断问题基于同一情景，锚定信息对于后续判断有关联，而任务无关条件下两个问题分别基于不同情景，比较问题中的锚定信息对于后续判断无关联。结果发现两个年龄群体在判断他人情绪时都受到锚定效应的影响（高锚组的估计值显著高于低锚组）；并且，任务关联条件下的锚定效应程度更大；此外，个体对消极情绪的判断也更容易受到锚定效应的影响。研究二整体采用了任务无关联的设计，并且关注锚定信息的情绪类别一致性对锚定效应的影响，研究采用2（年龄组：老年，年轻；被试间） × 2（锚定值：高锚，低锚；被试间）× 2（情绪效价：积极情绪，消极情绪；被试内）× 3（情绪一致性：情绪类型一致；情绪效价一致；情绪效价不一致；被试内）混合设计；招募了67名老年人（年龄：M=64.51±3.71）和66名年轻人（年龄：M=21.92±2.59）；情绪类别一致性是指比较问题情景中的情绪与判断问题情景中的情绪类别的一致性：两种属于同类型/同效价不同类别/不同效价。结果再次验证了即使是无关联的锚定信息也能对情绪判断产生锚定效应，且不受到情绪类型一致性的影响，但仍表现出对消极情绪的判断更受锚定效应的影响。研究三进一步对比了外部锚和内部锚两种类型的锚定效应对老年人和年轻人的影响，研究采用2（年龄组：老年，年轻；被试间） × 2（锚定值：高锚，低锚；被试间）× 2（情绪效价：积极情绪，消极情绪；被试内）× 2（锚定值来源：外部锚；内部锚；被试内）混合设计；招募了65名老年人（年龄：M=63.98±3.72）和63名年轻人（年龄：M=23.20±3.28）；在内部锚任务中，被试需要首先回顾自身经历判断自己在锚定情景（高锚/低锚情景）中的情绪强度，然后再判断一个类似的目标情景中故事主角可能的情绪强度。结果发现外部锚对个体情绪判断带来的影响比内部锚大。但是老年人相比于年轻人，更可能会受到内部锚的影响。

综上，三个研究的结果表明，不管对于老年人还是年轻人，情绪判断中都存在着较为强烈且稳定的锚定效应。也就是说，两个年龄群体的人在判断他人情绪时都很容易受到其他情绪信息的干扰和误导，从而判断产生偏误。而且即使是与判断目标毫无关联的情绪信息也可能会对个体的判断产生影响.此外，老年人和年轻人都表现出对消极情绪（主要是心烦和悲伤）有着更明显的锚定效应，我们对这类情绪的判断可能会更加不准确。最后，虽然来自外界干扰信息的影响往往更强，但是那些源于自身背景、经验的感受和情绪也会影响个体对他人的情绪判断，并且可能更多地影响老年人群体的判断。

The anchoring effect refers to the tendency that people’s numerical estimates are assimilated to a given anchor, which has been proven existed in many types of judgement and decision-making. This study investigated whether the anchoring effect existed in interpersonal emotion judgment for younger and older adults, using a traditional paradigm of “compare-judgment”. Participants were required to read the scenario which describe a brief emotion-related daily life story and then inferred the protagonist’s emotion intensity. Specifically, they have to judge and estimate the protagonist’s emotion intensity after compare it to a given anchor.

The first two studies focused on the impact of characteristics of anchor, which were based on the Selective Accessibility Model. The study 1 aimed to explore the impact of task-relevance of anchor on anchoring effect, by adopting a 2 (anchor: high/low; between-subjects) × 2(age group: younger/older; between-subjects) × 2(emotional valence: positive/negative; within-subjects) × 2(task relevance: task-relevant/task-irrelevant; within-subjects) mixed design. The participants included 65 older (age: M=64.62±3.82) and 68 younger adults (age: M=22.21±2.89). In task-relevant tasks, the compare and judgment phases shared identical scenarios; as for task-irrelevant tasks, the scenarios for compare and judgment were different, so the anchor in the previous phase was presumed not to offer a benchmark for later judgment. The results confirmed a robust anchoring effect (the estimates in high-anchor conditions were significantly higher). Additionally, the anchoring effect was greater in task-relevant tasks. Further, the anchoring effect was greater when judging negative emotion than that when judging positive emotion. The study 2 aimed to explore the impact of emotion-congruence of anchor on anchoring effect in anchor irrelevant tasks, by adopting a 2 (anchor: high/low; between-subjects)×2(age group: younger/older; between-subjects)×2(emotional valence: positive/negative; within-subjects)×3(anchor-emotion congruence: congruent-type/congruent-valence/incongruent-valence; within-subjects)mixed design. The participants included 67 older adults (age: M=64.51±3.71) and 66 younger adults (age: M=21.92±2.59). In congruent-type tasks, the anchor and target scenarios described identical emotion type; in congruent-valence tasks, these two scenarios emotions were same valence but different types; in incongruent-valence tasks, two scenarios emotions were opposite valence. The results repeatedly confirmed the anchoring effect, even the anchor was all irrelevant to judgment targets. Similarly, the anchoring effect was greater when judging negative emotion relative to positive emotion. And the anchoring effect was independent of anchor-emotion congruence. The study 3 further explored the effect of anchor source (experimenter-provided VS. self-generated) on anchoring effect, by adopting a 2 (anchor: high/low; between-subjects) × 2(age group: younger/older; between-subjects) × 2(emotional valence: positive/negative; within-subjects) × 2(anchor source: experimenter-provided/self-generated; within-subjects) mixed design. The participants included 65 older adults (age: M=63.98±3.72) and 63 younger adults (age: M=23.20±3.28). The tasks of experiment-provided were identical with anchor-relevant task in study 1; in the tasks of self-generated, participants were required to review their own experiences and judge their emotional intensity in anchoring scenarios (high or low anchor scenarios), and then to estimate protagonist’s emotional intensity in a similar target scenario. The results suggested the impact of experimenter-provided anchor was higher than self-generated anchor. But the older adults seem more vulnerable in self-generated anchor tasks, relative to younger adults.

In summary, the results showed that the anchoring effect in interpersonal emotion judgment is robust and stable, for both younger and older adults, which suggested both age groups are easily affected by biased context information when judging others’ emotions. And even the seemingly irrelevant anchoring information can affect individual’s emotion judgment. Besides, both younger and older adults showed a more pronounced anchoring effect on negative emotion judgments (but mainly in distressed and sadness emotion). Finally, although the effect from external emotion information was stronger, the misleading information from our own experiences and background can also biased our emotion judgment, and especially for older adults.