Responding to the dearth of research on user-system interaction patterns in context-aware recommendation systems (CARSs), this study extends Social Interdependence Theory (SIT) to construct a research model that seeks to unveil how users’ perceptions of procedural and outcome accountability affect their interaction patterns with CARSs, which in turn dictate the outcomes to be gained from cooperative learning between the two parties. Distinguishing the usage of CARSs as decision aids from that of delegated agents, we posit interaction patterns as having an impact on cooperative learning outcomes in terms of perceived system evolvability and habitual usage. Furthermore, we postulate context- and user-oriented affordances as moderators influencing the relationship between individual accountability and interaction patterns. To empirically validate our hypothesized relationships, we outline our plans for conducting an fMRI experiment in conjunction with survey data collection methods.