Attention plays an important regulatory role in the visual feature binding. There are two possibilities that attention binds the suitable features together. Firstly, through paying attention to spatial location, features on the same location are bound. Namely, spatial location serves as binding clues. Secondly, attention binds different features according to temporal synchronization of features from the same object. Namely, binding is based on the temporal synchronization or adjacency. This study explored the mechanism of attention in feature binding which base on temporal synchronization or spatial location. Two paradigms, rapid serial visual presentation(RSVP) and feature dissociation were used to artificially mimic a condition of temporal synchronization. Three experiments were designed in this study. Experiment 1 and experiment 2 used RSVP paradigm, and they held two experiment conditions, fixing presentation and changing presentation. In former condition, 4 distractors and 1 target were rapidly presented on the location of fixation point in turn, and then to observe temporal binding effects on the same location. In latter condition, distractors and target appeared alternately above and below fixation point, and then to explore temporal binding effects on different location. Dependent variables are illusory conjunction(IC) of target's color. The main results are as follows: no matter what the distance between target and distractors, in fixing presentation conditions, color IC were more likely to take place in temporal adjacent events, t(34)=81.23，P＜.0001；In changing presentation conditions, color IC were more from the same location events, t(34)=－17.29，P＜0.0001，which revealed that features from the same location were more easily to combine than temporal adjacent features, F(1,34)＝142.1，P＜.001. Experiment 3 using feature dissociation paradigm, artificially separated attributes(letter and frame color) of objects on different location. There were three temporal synchronization conditions. The first one is the control condition, in which two letters and color frames simultaneously presented on different location. On this condition, the numbers of IC from frame color and letter are baseline. The second condition is temporal synchronization of different location, in which color frame in one position appeared together with letter in another position. The third condition is temporal synchronization of the same location, in which letters and color frame in the same location appeared in the meantime. The experiment will examine the difference of IC among three conditions. If the second IC numbers are higher than baseline level, but the third was lower, temporal synchronization theory will be proved. The result showed that there was a significant difference of IC among three conditions, F(2,58)＝25.103，P＜.0001. Temporal synchronization on different location and same location were both apparently higher than baseline level(P＜.0001), but difference of temporal synchronization between different location and same location was not significant(P＞.05). Although there was IC on second condition which seemed showing temporal synchronization, IC also appeared on the third condition. Both of ICs were higher than baseline and the difference between two types of IC was not significant. It shows that temporal synchronization assumption was not be proved. In conclusions: 1) When features were located in same spatial position, temporal adjacent features tended to be bound; Temporal adjacent features located in different position could not be bound. 2) When the distance between objects on different locations decreased, the feature binding was still based on same location instead of temporal synchronization. 3) There were more ICs on both temporal synchronization conditions of features on same and different locations, which shows no evidence to support temporal synchronization theory.