The fornix supports episodic memory during childhood

Abstract

Episodic memory relies on the coordination of widespread brain regions that reconstruct spatiotemporal details of an episode. These topologically dispersed brain regions can rapidly communicate through structural pathways. Research in animal and human lesion studies implicate the fornix-the major output pathway of the hippocampus-in supporting various aspects of episodic memory. Because episodic memory undergoes marked changes in early childhood, we tested the link between the fornix and episodic memory in an age window of robust memory development (ages 4-8 years). Children were tested on the stories subtest from the Children's Memory Scale, a temporal order memory task, and a source memory task. Fornix streamlines were reconstructed using probabilistic tractography to estimate fornix microstructure. In addition, we measured fornix macrostructure and computed free water. To assess selectivity of our findings, we also reconstructed the uncinate fasciculus. Findings show that children's memory increases from ages 4 to 8 and that fornix micro- and macrostructure increases between ages 4 and 8. Children's memory performance across nearly every memory task correlated with individual differences in fornix, but not uncinate fasciculus, white matter. These findings suggest that the fornix plays an important role in supporting the development of episodic memory, and potentially semantic memory, in early childhood.

Keywords: fornix; memory development; tractography; uncinate.

Fig. 1

a) Schematic depictions of the memory assessments (left) and diffusion-weighted-imaging pipeline (right). The memory assessments comprised three tasks depicted in descending order: The CMS, the fact-source task, and the picture sequence task; b) Tactography for the uncinate (abbreviated UF) and fornix in a representative subject; and c) a graphic summary of the findings with each white matter index (legends) in each of the memory measures.

Fig. 2

Scatterplots depicting the relation between age and white matter indices including a) SL, b) FA, and c) surrounding FW. Additional scatterplots between memory performances (y-axes) and white matter SL (log-transformed, x-axes) from each task: d) the CMS, e) fact + source task, and f) temporal memory task. Note that in the scatterplots for (d, e, and f) the covariates of noninterest have been regressed out to illustrate the unique variance accounted for by the tract being depicted. Significance notation: ^**  P < 0.01, ^***  P < 0.001.