The direction of gaze automatically and exogenously guides visual spatial attention, a phenomenon termed as the gaze-cueing effect. Although this effect arises when the duration of stimulus onset asynchrony (SOA) between a non-predictive gaze cue and the target is relatively long, no empirical research has examined the factors underlying this extended cueing effect. Two experiments compared the gaze-cueing effect at longer SOAs (700 ms) in Japanese and American participants. Cross-cultural studies on cognition suggest that Westerners tend to use a context-independent analytical strategy to process visual environments, whereas Asians use a context-dependent holistic approach. We hypothesized that Japanese participants would not demonstrate the gaze-cueing effect at longer SOAs because they are more sensitive to contextual information, such as the knowledge that the direction of a gaze is not predictive. Furthermore, we hypothesized that American participants would demonstrate the gaze-cueing effect at the long SOAs because they tend to follow gaze direction whether it is predictive or not. In Experiment 1, American participants demonstrated the gaze-cueing effect at the long SOA, indicating that their attention was driven by the central non-predictive gaze direction regardless of the SOAs. In Experiment 2, Japanese participants demonstrated no gaze-cueing effect at the long SOA, suggesting that the Japanese participants exercised voluntary control of their attention, which inhibited the gaze-cueing effect with the long SOA. Our findings suggest that the control of visual spatial attention elicited by social stimuli systematically differs between American and Japanese individuals.