信息过滤与不确定决策:基于认知加工视角

Information Filtering and Decision-making: in the View of Cognitive Processing

个体决策总是伴随着信息筛选和聚焦的过程。以往针对信息数量影响风险决策的研究多是基于建模和优化,较少有学者从认知过程去探索信息对决策的影响。本研究设计了一个掩牌博弈游戏,利用事件相关电位技术(ERPs),比较了不确定情景下,个体在信息过滤前后的决策行为及脑电差异。行为数据显示,与信息过滤前相比,决策信息过滤后,个体的风险决策参与率更低;脑电结果表明,信息过滤后的决策,引发了个体更大的P2以及ERN振幅。上述结果表明,个体在不同的情境下,其信息处理和决策过程存在显著的差异。在信息数量比较多,并且时间短的情况下,决策者更倾向于调用启发式决策模式;而信息过滤后,信息量减少,信息属性更加明确,而且时间相对来说比较充裕,大脑会启动补偿式决策模式,去调用更多的认知资源来分析信息,此时,个体感知到的风险变大,行为会更加谨慎。

Decision-making is highly related to our daily lifebecause people need to make various decisions at all times. One of the most influencing factors of individual decision-making is the amount of information available. Information can reduce uncertainty;however, information overload may interfere the quality of decision-making. To process information quickly and make decision effectively, individuals are engaged in a constant filtering process based on screening and focalizing information. The process of information filtering allows individuals to efficiently process an overwhelming amount of information, but the reduction of information may lead to choice anxiety as a result of similar options. Answering problems about how information would affect individual decision-making can help understand decision-making mechanisms. Most previous studies on the role of information in decision-making adopted modeling and optimization approaches, while few studies discussed these issues from the cognitive perspective. Two decision-making patterns proposed by Kahneman, intuition and reasoning are widely accepted as the double system mode. After that, many studies about the cognitive process of decision-making emerged. With the discipline integration and technology development, some researchers have applied neuroscience research techniques for management science, such as ERPs （Event-Related Potentials）, to investigate the cognitive processes ofdecision-making. With these methods, investigators have found that some ERP components can represent the cognitive progress. For example, P2 component fluctuates with thevariation of attentional resources, and ERN is related to risk and the expectation of results. When faced with a larger risk, a lager ERN is produced. In addition, ERN is relevant to the goal of decision-making. When Systeml works, the heuristics decision pattern will invoke smaller amplitude compared with when we pursue for accuracy adopting System2. This study designed a gambling game, which simulates the dynamic process of information filtering. We applied ERPs to record the brain wave components before and after information filtering. Eighteen healthy college students （10M/8F, all right-handed） participated in this experiment. The stimuli in entire experiment consisted of 160 stimulus divided randomly into 4 blocks with 40 trials each. At the beginning of each trial, eight cards were presented. The valve of the first eight cards each was from ten to ninety generated by the system randomly. Followed by the first eight cards, five - cards stimulus was presented with other three covered randomly, and a fixed amount of gain emerged subsequently. At this moment, the subjects must choose to be risk seeking or to be conservative （take the fixed amount of gain and end this trial）. If he chose to be risk seeking, the system would continue to turn over another three cards randomly. Facing with two cards, participant needed to make the second decision： choose to be conservative （take the fixed gain） or to be risky （get the amount of money depending on either of the last two cards）. Applying Scan 4.5, we recorded the subjects＇ electrophysiological data while they were playing those games. ERPs were separately averaged for each condition and each subject. Behavioral data showed that participation rate （PR） of risky decision after information filtering was significantly less than that before information filtering. ERP results showed that two brain components were found. Risky decision-making after information filtering elicited significant larger P2 and ERN amplitudes than those before information filtering. These results suggested that information processing and decision-making under varied circumstances （before vs. after information filtering） were quite different. When faced with overloading information, decision-makers would pursue speed under this condition. However, after information filtering, they would allocate more cognitive resources to process the information and tend to be more cautious in decision-making to pursue accuracy. Therefore, larger ERN amplitude was manifested. Comparative with decision before information filtering, the lower PR and the larger ERN suggested that subjects perceive larger risk under risk seeking decision after information filtering. In summary, firstly, this study explores the cognitive processing of decision making from the information filtering perspective. It enriches current research in the fields of information updating theories and decision-making theories. Secondary, the results can help decision makers better understand their behavioral preferences so that they can avoid some impulsive behavior consciously and improve the efficiency of individual decision-making. Furthermore, our results can be applied into Internet businesses practice to design a better information layout ofhomepages and improve purchase conversion rate.