Interindividual Variability in Memory Performance Is Related to Corticothalamic Networks during Memory Encoding and Retrieval

Encoding new memories relies on functional connections between the medial temporal lobe and the frontoparietal cortices. Multiscan fMRI showed changes in these functional connections before and after memory encoding, potentially influenced by the thalamus. As different thalamic nuclei are interconnected with distinct cortical networks, we hypothesized that variations in corticothalamic recruitment may impact individual memory performance. We used a multiscan fMRI protocol including a resting-state scan followed by an associative memory task encompassing encoding and retrieval phases, in two independent samples of healthy adults (N1 = 29, mean age = 26, males = 35%; N2 = 108, mean age = 28, males = 52%). Individual activity and functional connectivity were analyzed in the native space to minimize registration bias. By modeling the direct and indirect effects of corticothalamic recruitment on memory using structural equation modeling, we showed a positive association between resting-state functional connectivity of the medial thalamic subdivision within the frontoparietal network and memory performance across samples (effect size R2 ranging between 0.27 and 0.36; p-values between 0.01 and 4 × 10−5). This direct relationship was mediated by decreased activation of the anterior subdivision during encoding (R2 ranging between 0.04 and 0.2; p-values between 0.05 and 0.006) and by increased activation of the medial subdivision during retrieval (R2 ranging between 0.04 and 0.2; p-values between 0.05 and 0.004). Moreover, three distinct clusters of individuals displayed different corticothalamic patterns across memory phases. We suggest that associative memory encoding relies on the distinct corticothalamic pathways involving medial thalamic recruitment and suppression of anterior subdivision to support the successful encoding of new memories.