Psychology Department, Liverpool John Moores University, Liverpool, United Kingdom.
PLoS One. 2024 Oct 10;19(10):e0312018. doi: 10.1371/journal.pone.0312018. eCollection 2024.
Bayesian reasoning (i.e. prior integration, cue combination, and loss minimization) has emerged as a prominent model for some kinds of human perception and cognition. The major theoretical issue is that we do not yet have a robust way to predict when we will or will not observe Bayesian effects in human performance. Here we tested a proposed divide in terms of Bayesian reasoning for egocentric spatial cognition versus allocentric spatial cognition (self-centered versus world-centred). The proposal states that people will show stronger Bayesian reasoning effects when it is possible to perform the Bayesian calculations within the egocentric frame, as opposed to requiring an allocentric frame. Three experiments were conducted with one egocentric-allowing condition and one allocentric-requiring condition but otherwise matched as closely as possible. No difference was found in terms of prior integration (Experiment 1), cue combination (Experiment 2), or loss minimization (Experiment 3). The contrast in previous reports, where Bayesian effects are present in many egocentric-allowing tasks while they are absent in many allocentric-requiring tasks, is likely due to other differences between the tasks-for example, the way allocentric-requiring tasks are often more complex and memory intensive.
贝叶斯推理（即先验整合、线索组合和损失最小化）已成为某些人类感知和认知的突出模型。主要的理论问题是，我们还没有一种强大的方法来预测何时会在人类表现中观察到贝叶斯效应，何时不会。在这里，我们根据自我中心空间认知与非自我中心空间认知（以自我为中心与以世界为中心）之间的贝叶斯推理进行了测试。该假设指出，当可以在自我中心框架内执行贝叶斯计算时，人们将表现出更强的贝叶斯推理效应，而不是需要非自我中心框架。进行了三个实验，一个是自我中心允许条件，另一个是非自我中心要求条件，但尽可能匹配。在先验整合（实验 1）、线索组合（实验 2）或损失最小化（实验 3）方面没有发现差异。之前的报告中存在对比，在许多自我中心允许的任务中存在贝叶斯效应，而在许多非自我中心要求的任务中不存在，这可能是由于任务之间的其他差异造成的，例如，非自我中心要求的任务通常更复杂且需要更多的记忆。
