摘要
证据分析包含直觉机制与理性机制,其中理性机制的弱化助长了认知偏差。在偏向于整体主义的认知模式中,可信性的双重评价增加了证据分析的错误风险。证据分析主要受到前见、预断、混淆三种认知偏差的干扰。经验法则运用、理由说明及其检验、人工智能辅助三种认知偏差矫正机制各有优势与局限。矫正认知偏差的重点是充分发挥理性机制的检验功能以及准确识别证据之间的虚假印证。在此方面,相比于以经验法则为基础的证明力规则,以促进证据分析信息更新为目标的程序规则更为重要。理由说明与检验的本质是事实认定者的自我反思,其在矫正认知偏差方面的功能是不可替代的,需要相应加强证据分析方法的教学与培训。关于人工智能证据分析的技术开发,应当侧重于推理过程检验的模拟,而非计算结果的输出。
Evidence analysis encompasses both intuitive and rational mechanisms,however,the weakening of the rational mechanism exacerbates the cognitive biases.In the holistic cognitive pattern,the dual evaluation of credibility increases the risk of error in evidence analysis.Evidence analysis is primarily affected by three cognitive biases:pre-existing beliefs,prejudgments,and confusions.The three mechanisms for correcting cognitive bias—application of the rule of thumb,reasoning and its verification,and artificial intelligence assistance—each have its own advantages and limitations.The key to correcting cognitive biases lies in fully exerting the verification function of the rational mechanism and accurately identifying false corroboration.In this regard,compared with rules of weight based on heuristics(rules of thumb),procedural rules aimed at promoting the information updating for evidence analysis are of greater importance.The essence of reasoning and verification lies in the fact-finder's self-reflection,which plays an irreplaceable role in correcting cognitive biases.Thus,it is necessary to strengthen the teaching and training of evidence-analysis methods accordingly.In terms of technical development of AI-based evidence analysis,the research focus should be placed on simulating the testing of the reasoning process,rather than merely the output of computational results.
作者
陈苏豪
Chen Suhao(r,Shanghai Key Laboratory of Forensic Medicine and Key Laboratory of Forensic Science,Ministry of Justice,Shanghai 200063;College of Humanities and Social Sciences,Nanjing University of Aeronautics and Astronautics,Nanjing 211521)
出处
《证据科学》
2025年第5期544-558,共15页
Evidence Science
基金
上海市法医学重点实验室暨司法部司法鉴定重点实验室开放课题“科学证据解释中的认知偏差研究”(KF202525)阶段性研究成果。
关键词
证据分析
证人证言
可废止论证
认知偏差
人工智能
Evidence analysis
Witness testimony
Defeasible reasoning
Cognitive biases
Artificial intelligence
