Developmental Differences in Hippocampal and Prefrontal Representations of Related Memories

• In Event: The Development of Memory Organization: Converging Insight From Multiple Perspectives

Abstract

Reasoning, which requires consideration of the relation among multiple events, improves into adulthood. Although children and adolescents form and retrieve memories for individual events, memory for the individual episodes is not sufficient to support inferring relationships among them (Bauer & San Souci, 2010; Schlichting et al., 2017). In adults, hippocampus (HPC) and medial prefrontal cortex (mPFC) work in concert to integrate new experiences with existing memories, resulting in memory codes that represent inferred knowledge (Zeithamova, Dominick, & Preston, 2012). Such integration changes the organization of individual events in memory, making related events more similar to one another (Schlichting, Mumford, & Preston, 2015). Because HPC and mPFC structure and function continue to develop through adolescence (Calabro, Murty, Jalbrzikowski, Tervo-Clemmens, & Luna, 2020; DeMaster & Ghetti, 2013; DeMaster, Pathman, Lee, & Ghetti, 2014), integration may be less evident at earlier ages. As a result, children and adolescents may be more likely to store related memories as individual, separate instances. Here, we tested whether overlapping events were represented differently in children and adolescents relative to adults.

Children (7-11 years), adolescents (12-15 years), and adults (18-23 years) learned an initial set of associations (AB) followed by overlapping (BC) and non-overlapping (XY) associations. Critically, overlapping associations (BC) could be integrated with initial pairs (AB) to promote inference about unobserved relationships (AC) (Figure 1). To assess age-related differences in overlapping memory representation, participants viewed indirectly related A and C items (e.g., Belle and purple object) before and after learning while fMRI activity was measured. Representational similarity analysis was used to quantify how the similarity among neural patterns evoked by A and C items changed as a function of learning, thus enabling estimation of whether indirectly related items became integrated (more similar) or differentiated (less similar) following learning.

Memory and inference performance were above chance in all groups (ps < .005). Moreover, memory for overlapping BC events and AC inference increased with age (ps < .03), indicating developmental improvements in learning and reasoning about related experiences. HPC and mPFC representations of indirectly related memory elements (A and C items) increased after learning in adolescents and adults, evincing preliminary evidence for integration at these ages. In contrast, children showed a qualitatively different representational profile, whereby related events were represented as less similar after learning, consistent with differentiation of related events in memory. Finally, in all age groups, greater evidence for HPC integration of overlapping memories predicted inference on a trial-by-trial basis (p = .03; Figure 2). These findings indicate that children organize competing memories differently than adolescents and adults. Whereas adolescents and adults integrate competing memories together in memory, children actively differentiate them, which may limit their ability to infer unobserved relationships among individual events. Our results further show that increases in integrated memory representation drives improvements in reasoning behavior across age. Together, these findings highlight how development supports the transformation of memory representation from a system that organizes related events separately to one that supports formation of integrated memories that facilitate the formation of inferred knowledge.

Authors

• Nicole Varga, University of Texas at Austin
  Presenting Author

• Alison Preston, University of Texas at Austin
  Non-Presenting Author