Neural Correlates of Species-typical Illogical Cognitive Bias in Human Inference
Neural Correlates of Species-typical Illogical Cognitive Bias in Human Inference
复制标题
DOI：
10.1162/jocn.2009.21330
复制
发表时间：
2010-09
影响因子：
3.2
通讯作者：
Akitoshi Ogawa;Y. Yamazaki;K. Ueno;K. Cheng;A. Iriki
中科院分区：
文献类型：
--
作者：
Akitoshi Ogawa;Y. Yamazaki;K. Ueno;K. Cheng;A. Iriki
逻辑思考的能力是人类智力的一个标志，然而我们天生的推理能力却带有隐性偏见，常常导致不合逻辑的推理。例如，给定AB（“如果A，则B”），人们经常但错误地推断出相反的BA。这种被称为对称的推理模式在逻辑上是无效的，因为尽管它可能是真的，但它不一定是真的。给定一对条件关系，如AB和BC，人类反射性地执行两种额外的推理模式：及物性，其中一个（有效地）推断AC；与此形成鲜明对比的是，非人类动物可以处理及物性，但很少能获得对称性或等效性。在本研究中，人类受试者对抽象的、视觉呈现的图形之间的关系进行逻辑和非逻辑推理，同时用功能磁共振成像监测他们的大脑活动。在所有的推理模式中，前额叶、内侧额叶和顶叶内皮层都被激活。在传递性和等效性过程中观察到楔前叶和后顶叶皮层的额外激活，这可能反映了需要从记忆中检索中间刺激(B)。令人惊讶的是，非逻辑性推理和逻辑性推理的大脑活动模式非常相似。我们得出的结论是，观察到的推理相关的额顶叶网络适用于处理刺激之间的分类关联结构，而不是逻辑关联结构。在处理大量信息时，人类可能更喜欢分类而不是记忆逻辑结构，以尽量减少认知工作记忆负荷。
The ability to think logically is a hallmark of human intelligence, yet our innate inferential abilities are marked by implicit biases that often lead to illogical inference. For example, given AB (“if A then B”), people frequently but fallaciously infer the inverse, BA. This mode of inference, called symmetry, is logically invalid because, although it may be true, it is not necessarily true. Given pairs of conditional relations, such as AB and BC, humans reflexively perform two additional modes of inference: transitivity, whereby one (validly) infers AC; and equivalence, whereby one (invalidly) infers CA. In sharp contrast, nonhuman animals can handle transitivity but can rarely be made to acquire symmetry or equivalence. In the present study, human subjects performed logical and illogical inferences about the relations between abstract, visually presented figures while their brain activation was monitored with fMRI. The prefrontal, medial frontal, and intraparietal cortices were activated during all modes of inference. Additional activation in the precuneus and posterior parietal cortex was observed during transitivity and equivalence, which may reflect the need to retrieve the intermediate stimulus (B) from memory. Surprisingly, the patterns of brain activation in illogical and logical inference were very similar. We conclude that the observed inference-related fronto-parietal network is adapted for processing categorical, but not logical, structures of association among stimuli. Humans might prefer categorization over the memorization of logical structures in order to minimize the cognitive working memory load when processing large volumes of information.