Insight is a method for solving intellectual problems without any explicit algorithm.The analyses in the literature identified the areas in the brain that are related to insight but they have not elucidated the spaciotemporal structure of insight due to low time-precision.Moreover, they do not necessarily discriminate active areas related to insight from those related to the task itself.This study coped with the latter problem by employing an anagram task that has two ways to solve, insight and an algorithm.We measured EEG signals of participants during the task and made classifiers that predicted whether participants solved the problem by insight or not from the EEG signals.The better a classifier predicts an occurrence of insight, the more information on insight its input signal includes.Hence, we took two approaches from this standpoint.In one approach, we made classifiers each of which has the input vector that consists of band powers at a channel in a time-window and extracted good predictors, where the criterion is the area under the ROC curve or AUC.The result show that insight originates at P4 in the 10-20 system, propagates to Cz and T3 in 50 or 100 msec, and diffuse to other areas such as F4.In the other approach, we made only one classifier, that has the band powers of all the channels and the time-windows.The input vector was so high dimensional that we selected some using the sparse logistic regression or SLR.the result show that insight firstly appears in the gamma waves at P4 and T4, and secondly appears in the gamma wave at C4.Although the above results do not completely match each other, they are almost consistent with the existing results.