In the context of sensory processing, visual discrimination is a fundamental function that enables survival. Previous findings suggest that such discrimination function can be decoded from electroencephalographic brain responses, especially by using oscillation feature. However, how to evaluate the fast visual discrimination is still unclear. In this study, we hypothesize that brain’s oscillatory activity in a passive viewing condition can serve as a sensitive predictor of fast visual discrimination. A visual multi-feature paradigm which allowing investigation of several different change types was used to record both event-related potentials (ERPs) and behavioral responses. First, we investigated separating the behavioral hit rate as a function of reaction time (categorized from 200 ms to 1000 ms with step of 100 ms). In the subsequent step, we extract the slow theta component from ERP’s time frequency represents with time frequency principal component analysis (TF-PCA) and correlate its average power with behavioral performance. Our results showed that the significant detect window for different deviants’ level was from 400 to 600 ms, while the hit rates in such detect window showed a significant correlation with the averaged time frequency power in the slow theta band during 100-300 ms latency for the color and shape deviants. These findings suggest that the oscillation power, particularly in the slow theta range, of the brain responses is a predictor of fast visual discrimination.