This chapter explores issues in memory and affect in connection with possible architectures for artificial cognition. Because memory and emotion are evolutionarily and developmentally rooted in social interdependence, a new understanding of these issues is necessary for the principled design of true intelligent systems. We treat emotion not as an optional extra or as a brief episode of feeling, but as the underlying substrate enabling the formation of social relationships essential for the construction of cognition. We treat memory not as the storage and retrieval of immutable data, but as a continuous process contingent on experience and never fully fixed or immutable. Three converging areas of research are identified that hold some promise for further research: social constructionism, reconsolidation memory theory, and memory models based on the nonlinear dynamics of unstable periodic orbits. We argue that the combination of these ideas offers a potentially more substantive approach to understanding the cognitive architecture.