Functional near-infrared spectroscopy (fNIRS) is well-suited for hyperscanning in naturalistic situations, offering significant potential for assessing social brain function in everyday life. Previous studies have reported inter-brain synchrony during social interactions and sought to explore its mechanisms by correlating behavioral events with brain signals. However, commonly used regression analyses, such as General Linear Models (GLM), rely on target events hypothesized as explanatory variables. This reliance introduces a dependency on the researcher's assumptions, which can compromise replicability in social neuroscience. While such dependency may be less problematic in strictly controlled experimental paradigms focused on specific hypotheses, it poses significant challenges in naturalistic experiments like social interactions, where numerous events and signals may serve as potential explanatory variables. To address this limitation, we introduced a new approach: signed-normalized mutual information (sNMI). This method enables a two-way analysis to evaluate relationships between event sequences and brain synchrony. Through simulations and real-data applications, we evaluated the performance of the proposed method. The results showed that sNMI analysis performed comparably to regression analysis in detecting both inter-brain synchrony and within-brain synchrony (i.e., brain connectivity). Moreover, applying sNMI to a tapping dataset revealed the expected patterns of synchrony between the lateral sides of the motor area. These findings validate the effectiveness of sNMI as a robust two-way analytical tool for studying brain synchrony.