Symmetry bias is sometimes regarded as a culprit that results in fallacies and spoils our logically correct reasoning. We believe, however, that this bias is also indispensable if we are to obtain unfamiliar notions, to think creatively, and to infer heuristically. We devised an algorithm which can overcome the “exploration-exploitation dilemma” that is involved in the decision-making process under conditions of uncertainty, via a computer simulation based on the “2-armed Bandit Problem.” As a result, we confirmed that the agent-model, which contains both kinds of symmetry biases, displayed very flexible behavior, and that it achieved drastically higher scores than the other mechanical agents. In this paper, by drawing upon the theory of ‘Bi-logic’ as developed by Matte Blanco, we discuss why human-beings require this kind of illogicality, and how fundamental it is for human [un]consciousness.