Negotiated and reciprocal exchange structures in human-agent cooperation

Abstract

In light of increasing automation capability, Social Exchange Theory may help guide the design of automated digital interlocutors in human-agent teaming to enhance joint performance. The effect of two social exchange structures, negotiated exchange and reciprocal exchange, was assessed using a joint scheduling microworld environment. Negotiated exchange was operationalized as a bilateral requesting-and-accepting interaction in the microworld, and reciprocal exchange was operationalized as a unilateral resource-providing interaction in the microworld. Compared to the negotiated exchange structure, the reciprocal exchange structure led to increased resource-sharing by both the participant and agent, and thus higher joint performance in the microworld task. The reciprocal exchange structure thus enhanced performance by reducing communication overhead, facilitating proactive and increased resource-sharing. However, results also suggest participant resource-sharing was more mindless than strategic, and failed to consider agent needs. This study demonstrates a new approach to conceptualizing the social mechanisms of human-automation interaction for enhanced human-agent teaming.