Multi-agent systems are widely used to model components of a smart living environment as autonomous intelligent agents. Accordingly, its advantages to achieve the comfort and efficiency goals of smart home systems are well-documented in many studies. However, there is a clear lack of systematic investigation targeted at exploring the reasoning modules integrated into these systems. To close this gap, this paper examines the literature on multi-agent based smart home systems and provides a comprehensive overview of the essential requirements, assumptions, strengths, limitations, challenges and future research directions of their proposed reasoning systems. Moreover, it identifies the main technologies used to represent the home environment as a multi-agent system and the reasoning approaches utilized to bring decision-making ability into the smart living environment. As a result, this systematic literature review identifies the ability to learn, plan, predict, explain and reason with incomplete knowledge as the major elements of a smart home reasoning system. In addition, the findings of this work revealed the application of standard rule conflict resolution strategies and sensor data contextualization as principal solutions to address some of the problems caused by conflicting rules and agent goals. Further, it underlines the importance of utilizing hybrid reasoning approaches and the need to handle overlapping multi-inhabitant activities to realize the true potential of smart homes.