Measuring memory integration: A metric tapping memory representation rather than inference
- Wangjing Yu, Columbia University, New York, New York, United States
- Margaret Schlichting, University of Toronto, Toronto, Ontario, Canada
- Katherine Duncan, University of Toronto, Toronto, Ontario, Canada
AbstractOur ability to link related events could be supported either by connecting their representations in memory, or by storing them separately but integrating their content when later drawing inferences. Here, we adapted classic memory contingency analyses to develop and validate an integration index designed to tap stored representations. We conducted three preregistered experiments adopting this metric. We found positive recall dependency for associations experienced both within the same and across different events. Compared to a conventional inference test, we found that recall dependency was more sensitive to a manipulation of memory integration. Leveraging recall dependency to investigate individual differences revealed that better memory for contextual detail was associated with faster inference judgments, consistent with high-fidelity representations of related memories—but only for people who tended to store memories separately. Our approach, thus, provides an important tool to illuminate how related events are represented in memory.