Exploring the Unconscious Nature of Insight Using Continuous Flash Suppression and a Dual Task - eScholarship

Exploring the Unconscious Nature of Insight Using Continuous Flash Suppression and a Dual Task Hiroaki Suzuki (susan@ri.aoyama.ac.jp) College of Education, Psychology and Human Studies, Aoyama Gakuin University 4-4-25 Shibuya, Shibuya-ku, Tokyo 150-8366, Japan Haruaki Fukuda (haruaki@idea.c.u-tokyo.ac.jp) Graduate School of Arts and Sciences, The University of Tokyo 3-8-1 Komaba, Meguro-ku, Tokyo 153-8902, Japan Hiromitsu Miyata (miyata@hirc.aoyama.ac.jp) Human Innovation Research Center, Aoyama Gakuin University 4-4-25 Shibuya, Shibuya-ku, Tokyo 150-8366, Japan Keishi Tsuchiya (stk16.instagram@gmail.com) College of Education, Psychology and Human Studies, Aoyama Gakuin University 4-4-25 Shibuya, Shibuya-ku, Tokyo 150-8366, Japan Abstract Dynamic Constraint Relaxation Theory of Insight A growing body of evidence suggests that implicit information processing has considerable effects on the higher- order cognitive processes such as insight problem solving. Is such implicit information stored within the working memory system, or is it processed in a storage system other than working memory? To differentiate these two possibilities, the present study examined solution of the T-puzzle, an insight problem, after participants were or were not subliminally presented with the hint images by using the continuous flash suppression (CFS). A spatial tapping task, which is deemed to interfere with spatial working memory, was introduced during CFS. The two hypotheses each predicted deteriorated and maintained performance on the T-puzzle after the tapping task. Contrary to these hypotheses, participants tended to exhibit better solution performance and relaxation of constraints after having the tapping task, either with or without subliminal presentation of the hints. Mechanisms by which the secondary task may facilitate insight problem solving are discussed. Keywords: insight problem solving; working memory; implicit processing; T-puzzle; continuous flash suppression. Introduction Whereas explicit information processing is generally assumed to govern human higher thoughts, studies on implicit learning and memory have suggested that implicit information has considerable influence on our thoughts and behavior (Eagleman, 2011). For example, researchers have long assumed that conscious information processing including goal setting, planning, monitoring of actions, etc. plays a dominant role in human problem solving. However, a number of recent reports suggest that subliminally presented hint stimuli significantly facilitate subsequent performance on insight problems (Hattori et al., 2013; Suzuki & Fukuda, 2013). That is, information that is processed at the subconscious level may considerably influence higher-order cognitive processes such as insight. Insight problem solving has several unique characteristics. First, whereas problems typically used in psychological experiments are simple, it is quite difficult for solvers to attain solutions by themselves (Ohlsson, 1992). Second, solvers stick to the incorrect approaches and make the same errors repeatedly (Kaplan & Simon, 1990). During these impasses, they frequently ignore useful information that was accidentally found or generated (Suzuki et al., 2000). Finally, insight seems to come to the mind suddenly. A number of theories have proposed different mechanisms by which problem solving by insight is attained. We here adopt the ideas of the dynamic constraint relaxation theory (Suzuki, 2009), which has been developed under the strong influence of the notion of constraint (Knoblich et al., 1999) and Q-learning with softmax algorithm (Bridle, 1990). The theory assumes three kinds of constraints and a relaxation mechanism. The term “constraint” here refers to humans’ natural tendencies to select appropriate options and exclude inappropriate ones out of the huge amount of information. The object-level constraint reflects people’s natural preferences of how given objects are encoded. The relational constraint refers to solvers’ natural preferences of how given multiple objects are related to each other. The goal constraint evaluates a match between the current and the desired states, and gives feedback to the constraints responsible for generating the current states. At the initial stages of problem solving, the object-level and relational constraints jointly operate to lead solvers to an impasse. However, as solvers repeat manipulations, feedback provided by the goal constraint dynamically alters the strength values of the object-level and relational constraints. This increases the probabilities of constraint violation. At a certain stage of problem solving, solvers accidentally violate each constraint to attain correct solution. This theory