Abstract: Deductive reasoning is an inference-making process that allows individuals to draw valid conclusions from a given set of premises. Conditional reasoning is an important form of deductive reasoning which involves making an inference with a given major premise and one of four possible minor premises. People have been shown to be prone to making logical errors in deductive reasoning and to be strongly influenced by the semantic content of logical arguments. This phenomenon is the famous ‘belief-bias effect’. That is, people tend to accept believable conclusions and reject implausible ones, regardless of logicality. Therefore, beliefs that are consistent with the logical conclusion facilitate one’s performance during the logical task; in contrast, beliefs that disagree with the logical conclusion are often inhibitory to the logical task. Take the conditional premise, “If someone is female, then raise your hand”, for example. The belief regarding the content will discourage the logical conclusion under the form (if a person is not female, then it is uncertain whether or not the person should raise his hand). However, in the conditional premise, “If a person is rich, then the person will be happy”, the logical conclusion seems to be ‘‘facilitated’’ in the form (if a person is not rich, then the conclusion is uncertain). Based on the inference-verification paradigm in which the major premise, minor premise and conclusion appear one after another, high-density event-related potentials (ERPs) were recorded to examine the electrophysiologic correlates of belief bias effect in realistically conditional reasoning in daily life. That is, the present study investigated spatiotemporal patterns of the brain in the performance of one conditional reasoning task (“If P then Q, not-P) with two conditions (facilitatory of belief condition and inhibitory of belief condition) using high-density (64 channels) ERP recording system. We aimed to clarify the temporal features of the belief-bias effect and determine ERP components that embody this bias. The ERP waveforms were time-locked to the onset of the major premises, the minor premises, and the conclusion, respectively. The averaged epoch for the major premises was 3200ms, including a 200ms pre-stimuli baseline. The averaged epoch for the minor premises was 3200ms, including a 200ms pre-stimuli baseline. The averaged epoch for the conclusion was 1200ms, including a 200ms pre-stimuli baseline. Only trials with correct judgment were taken into the ERP analysis. Results showed that, first, relative to the facilitatory of belief condition, the behavioral data reflected longer RTs and lower accuracy rates in the inhibitory of belief condition, verifying the belief-bias effect. Second, the components elicited by facilitatory of belief condition and inhibitory of belief condition were not significantly different neither the stages of inference of minor premise nor the stage of judgment of conclusion. However, the components elicited by the two conditions were significantly different during the stage of the semantic representation of major premise, most likely reflects the effect of belief-bias in conditional reasoning.

Key words: conditional reasoning, belief bias, event-related brain potentials (ERPs), late positive component (LPC)

摘要： 以日常生活中的条件推理语句为实验材料，采用大、小前提和结论依次呈现的“推断-判断”范式，利用事件相关电位(event-related brain potential, ERP)技术探讨了条件推理中否定前件下信念偏差效应的脑内时程动态变化。结果发现：在行为反应上，较信念促进，信念阻碍下的正确率更低反应时更长；在脑电上，两条件(信念阻碍和信念促进)诱发的ERP波形仅在大前提加工阶段出现明显的分离。这表明该推理下的信念偏差效应可能早在对大前提的语义表征阶段就已发生。

关键词: 条件推理, 信念偏差, 事件相关电位, 晚正成分