Predation risk is a significant concern when social animals including humans engage in foraging. When people search for resources together, individuals often find themselves in a producer–scrounger game, in which some individuals bear the cost of risk monitoring while others can free ride on those efforts. A theoretically rational strategy is to mix foraging and risk monitoring randomly with the same probability across all members, but such uncoordinated action often yields inefficiencies of under- or over-supply of risk monitoring in a group. Here, we examined whether people could spontaneously develop a coordinated risk-monitoring system, alternating vigilance and foraging in a pair. Given that human cooperation is vulnerable to fear of exploitation and emotional arousal under risk, we hypothesized that such sources of anxiety would be potential disruptors to coordination. In a laboratory experiment, two participants worked on a “treasure hunt” task simultaneously, in which they chose between low or high vigilance against predators during foraging without verbal communication. If one chose high vigilance with personal cost, it yielded a spillover benefit to the other. Besides behavioral choices, each participant’s physiological arousal (skin conductance response) and cognitive effort (tonic pupil dilation) were measured during the task. Results showed that some pairs were actually able to develop a role-alternating system over time through tacit coordination, but coordinated action was also vulnerable to anxiety and mistrust among participants. Overall, these results imply that, besides the mutual behavioral control that often characterizes repeated interaction, cognitive control of emotional arousal may be a critical psychological factor for the emergence of coordinated cooperation.