Cognitive mechanisms of insight problem-solving and its computational model

今年，我们从经验上探讨了洞察力解决问题的个体差异，并在理论上将我们的方法与他人进行了对比。首先，我们使用几何谜题分析了洞察力解决问题的个体差异的来源。根据我们的理论，个体差异有三个来源：约束的初始强度，对解决问题尝试的评估，以及更新每个约束强度值的学习率。为了验证这些预测，我们将能在10分钟内解决几何洞察力问题的受试者与不能在10分钟内解决几何洞察力问题的受试者进行了比较。对求解过程和评价任务的微观分析表明，约束条件的初始强度和评价的充分性在两者之间是不同的，但它们的学习率没有差异。这些结果表明，有洞察力的问题解决者甚至在解决问题的初始阶段更经常地尝试非标准的尝试，并更准确地评估他们的尝试。讨论结果与其他方法的洞察力。其次，我们回顾了标准问题解决、记忆检索、问题空间、机会主义和约束方法等框架。从这些方法中获得的重要发现和启示是发展一种不涉及洞察力特定机制的理论，解释有用信息的时间敏感性，并将解决问题尝试的反馈机制纳入理论。理论分析表明，我们的理论对这些发现给出了连贯的解释，而其他方法却没有。我们进一步讨论了洞察力理论的问题，如对其他创造性思维的扩展，洞察力的有意识/无意识性质，以及神经学探索的可能性。

This year, we empirically explored individual differences in insight problem-solving and theoretically contrasted our approach with others.Firstly, we analyzed sources of individual differences in insight problem-solving, using a geometric puzzle. According to our theory there are three sources of individual differences: the initial strengths of constraints, the evaluation of problem-solving attempts, and the learning rate that updates the strength value of each constraint. In order to validate these predictions, we compared the subjects who could solve a geometric insight problem within 10 minutes with those who could not. A microscopic analysis of the solution processes and rating task revealed that the initial strengths of the constraints and the adequacy of evaluation are different between the two, but their learning rates are not. These results suggest that insightful problem-solvers try non-standard attempts more often even in the initial phase of their problem-solving and evaluate their attempts more precisely. The results were discussed in contrast with other approaches to insight.Secondly, we reviewed frameworks such as standard problem-solving, memory retrieval, problem-space, opportunistic, and constraint approaches to insight. The important findings and implications obtained from these approaches are to develop a theory that does not involve insight-specific mechanisms, to explain time-sensitiveness to useful information, and to incorporate to a theory a mechanism for the feedbacks from problem-solving attempts. Theoretical analysis revealed that our theory gives coherent accounts to these findings, while other approaches do not. We discussed further issues for theories of insight such as extensions to other creative thinking, conscious/unconscious nature of insight, and possibilities of neurological exploration.