Chunk decomposition is a critical mechanism of insight problem solving and creative thinking. It refers to decomposing a holistic perceptual chunk into its components in order for creating new products or reorganization. Previous studies mainly investigated the neuro-cognitive mechanism of chunk decomposition related to the element-type-based chunk tightness, less is known on the neural correlates of insightful chunk decomposition associated with chunk tightness based on spatial intersecting relationship. The current study aimed to investigate the cognitive processing mechanism and the neural dynamics of insightful chunk decomposition by using the event-related potentials (ERPs) technique of high temporal resolution in a Chinese character decomposition task. Twenty-four volunteers (11 female; 20.29 ± 1.78) participated in this study. All of them were native Chinese, right-handed, and had normal and correct-to-normal vision. They reported no history of brain damage or psychiatry. Participants were asked to complete a character decomposition task in which they removed strokes or a character from a source character for getting a valid character. Two conditions were manipulated and compared according to the category relationship between the probe (the to-be-removed part) and the target (the left part). In the condition of same category, the to-be-removed part and the left part are both characters. In the condition of different categories, the to-be-removed part are strokes and the left part is a character. More critically, to ensure character decomposition occurs in an insightful way, we kept the to-be-removed part (strokes or a character) and the left part (a character) in a spatially intersecting relationship in the to-be-decomposed chunk. Meanwhile, ERPs were recorded after the onset of the source character when chunk decomposition occurs. We hypothesized that the condition of same category (vs. different categories) would induce greater perceptual conflict in the two stages of insightful chunk decomposition where the to-be-removed part (a probe) and the left part (a target) are spatially intersecting with each other in the source character. Specifically, in the stage of probe identification, same category (vs. different categories) would elicit greater perceptual conflict due to the distraction of the left part, which was reflected by greater N2 deflection. In the stage of perceptual transformation, same category (vs. different categories) would elicit greater perceptual conflict and make transformation more difficult, which was reflected by decreased LPC. The results supported the hypothesis by showing that the condition of same category elicited greater N2 amplitude during the time window of 250-400 ms and decreased LPC amplitude during the time window of 550-1000 ms, compared with the condition of different categories. Consistent with the EEG data, it was shown that chunk decomposition required significantly longer response times in the same category condition than in the condition of different categories. It is concluded that chunk decomposition might contribute to insight problem solving by breaking difficulty in two phases. In early stage of insightful chunk decomposition, the difficulty of probe identification might lead to perceptual conflict, eliciting greater N2 amplitude. In the later stage of insightful chunk decomposition, the difficulty of perceptual transformation might cause memory load, as reflected by decreased LPC amplitude. Ultimately, both difficulty in identification phase and transformation phase lead to longer reaction times in behavioral performances. The current study aimed to investigate the information processing mechanism and the neural dynamics of insightful chunk decomposition by using the event-related potentials (ERPs) technique of high temporal resolution in a Chinese character decomposition task. Twenty-four volunteers (11 female; 20.29 ± 1.78) participated in this study. All of them were native Chinese, right-handed, and had normal and correct-to-normal vision. They reported no history of brain damage or psychiatry. Participants were asked to complete a character decomposition task in which they removed strokes or a character from a source character for getting a valid character. Two conditions were manipulated and compared according to the category relationship between the probe (the to-be-removed part) and the target (the left part): in the condition of same category, the to-be-removed part and the left part are both characters; in the condition of different categories, the to-be-removed part are strokes and the left part is a character. More critically, to ensure character decomposition occurs in an insightful way, we kept the to-be-removed part (strokes or a character) and the left part (a character) in a spatially crossed relation in the to-be-decomposed chunk. Meanwhile, ERPs were recorded after the onset of the source character when chunk decomposition occurs. We hypothesized that the condition of same category (vs. different categories) would induce greater perceptual conflict in the two stages of insightful chunk decomposition where the probe and target are spatially intersecting with each other in the source character. Specifically, in the stage of probe identification, same category (vs. different categories) would elicit greater perceptual conflict due to the distraction of the left part, which was reflected by greater N2 deflection. In the stage of perceptual transformation, same category (vs. different categories) would elicit greater perceptual conflict and make transformation more difficult, which was reflected by decreased LPC. The results supported the hypothesis by showing that the condition of same category elicited greater N2 amplitude during the time window of 250-400ms and decreased LPC amplitude during the time window of 550-1000 ms, compared with the condition of different categories. Converging with the EEG data, it was shown that chunk decomposition required significantly longer response times in the same category condition than in the different categories condition. it is concluded that chunk decomposition might contribute to insight problem solving by breaking difficulties in two phases: in early stage of insightful chunk decomposition, the difficulty of probe identification might lead to perceptual conflict, eliciting greater N2 amplitude; in the later stage of insightful chunk decomposition, the difficulty of perceptual transformation might cause memory load, as reflected by decreased LPC amplitude. Ultimately, the two stages of difficulties lead to longer reaction times in behavioral performances.