There is broad awareness that the health of coworker relationships is often built on reciprocity and assessments of member exchanges, where relationships are pursued or terminated based on benefits received. Both social exchange theory and equity theory propose that, when an exchange relationship is no longer favorable (i.e., a negative reciprocity relationship), an individual should terminate it to prevent resource losses. While this is economically rational, it often is impossible or impractical to terminate a relationship in a work context. The objective of this dissertation is to address this apparent mismatch between theories of helping behavior and typical workplace dynamics. I do so by exploring three possible explanations for this mismatch. First, I argue that the one of the key assumptions of social exchange theory, that the relationships are voluntary, may not always hold in a work setting. Second, I argue that fluctuations in investment behavior, specifically helping, changes in a non-linear fashion over time. Finally, I examine the impact of reciprocity, perceptions of team member efficacy, and third-party investment on helping behaviors in a sustained, negative reciprocity relationship. Across two experiments and one field study, I found that helping behaviors change discontinuously over time, individuals will help a partner complete an interdependent task regardless of reciprocation, and that helping is driven by a combination of factors including partner performance and general perceptions of a partner's helpfulness.