Predictive processing accounts of social perception argue that action observation is a predictive process, in which inferences about others' goals are tested against the perceptual input, inducing a subtle perceptual confirmation bias that distorts observed action kinematics toward the inferred goals. Here we test whether such biases are induced even when goals are not explicitly given but have to be derived from the unfolding action kinematics. In 2 experiments, participants briefly saw an actor reach ambiguously toward a large object and a small object, with either a whole-hand power grip or an index-finger and thumb precision grip. During its course, the hand suddenly disappeared, and participants reported its last seen position on a touch-screen. As predicted, judgments were consistently biased toward apparent action targets, such that power grips were perceived closer to large objects and precision grips closer to small objects, even if the reach kinematics were identical. Strikingly, these biases were independent of participants' explicit goal judgments. They were of equal size when action goals had to be explicitly derived in each trial (Experiment 1) or not (Experiment 2) and, across trials and across participants, explicit judgments and perceptual biases were uncorrelated. This provides evidence, for the first time, that people make online adjustments of observed actions based on the match between hand grip and object goals, distorting their perceptual representation toward implied goals. These distortions may not reflect high-level goal assumptions, but emerge from relatively low-level processing of kinematic features within the perceptual system. (PsycInfo Database Record (c) 2020 APA, all rights reserved).

中文翻译：

---

业务量匹配预测性地塑造了他人正在进行的动作的感知表示。

对社会知觉的预测处理说明认为，行动观察是一个预测过程，在该过程中，针对其他目标的推论是根据感知输入进行测试的，从而引发了细微的感知确认偏差，从而使观察到的行动运动学趋向于推断的目标。在这里，我们测试了即使没有明确给出目标但必须从展开动作运动学得出目标时，是否也会引起这种偏见。在2个实验中，参与者短暂地看到一个演员模糊不清地朝着一个大物体和一个小物体伸出手，或者是全手握把，或者是食指和拇指精确握把。在此过程中，手突然消失了，参与者在触摸屏上报告了其最后一次见到的位置。如预期的那样，判断始终偏向于明显的行动目标，这样，即使触角运动学相同，也可以感觉到力量抓地力靠近大物体，而精确抓地力则靠近小物体。令人惊讶的是，这些偏见与参与者明确的目标判断无关。当必须在每个试验（实验1）或不试验（试验2）中明确得出行动目标时，它们的大小相等，而且在各个试验中以及在各个参与者之间，明确的判断力和感知偏见是不相关的。这首次提供了证据，表明人们根据抓握力和物体目标之间的匹配程度对观察到的动作进行了在线调整，从而使他们的感知表示向隐含目标倾斜。这些失真可能无法反映高级目标假设，但会从感知系统内相对较低级别的运动学特征处理中产生。