A Role for the Fornix in Temporal Sequence Memory

Abstract

Converging evidence from studies of human and nonhuman animals suggests that the hippocampus contributes to sequence learning by using temporal context to bind sequentially occurring items. The fornix is a white matter pathway containing the major input and output pathways of the hippocampus, including projections from medial septum, and to diencephalon, striatum, and prefrontal cortex. If the fornix meaningfully contributes to hippocampal function, then individual differences in fornix microstructure might predict sequence memory. Here, we tested this prediction by performing tractography in 51 healthy adults who had undertaken a sequence memory task. Microstructure properties of the fornix were compared with those of tracts connecting medial temporal lobe regions, but not predominantly the hippocampus: the Parahippocampal Cingulum bundle (PHC) (conveying retrosplenial projections to parahippocampal cortex) and the Inferior Longitudinal Fasciculus (ILF) (conveying occipital projections to perirhinal cortex). Using principal components analysis, we combined Diffusion Tensor Imaging and Neurite Orientation Dispersion and Density Imaging measures obtained from multi-shell diffusion MRI into two informative indices, the first (PC1) capturing axonal packing/myelin, the second (PC2) capturing microstructural complexity. We found a significant correlation between fornix PC2 and implicit reaction-time indices of sequence memory, indicating that greater fornix microstructural complexity is associated with better sequence memory. No such relationship was found with measures from the PHC and ILF. This study highlights the importance of the fornix in aiding memory for objects within a temporal context, potentially reflecting a role in mediating network communication within an extended hippocampal system.

Multimodal MEG and microstructure-MRI investigations of the human hippocampal scene network

“…Here, we expanded on these findings, linking scene oddity accuracy to more specific features of fornix microstructure. Redundancies in several microstructure measures were exploited to reveal two biologically interpretable components, which were similar to those described in previous studies that used similar PCA methods (Chamberland et al, 2019;Geeraert et al, 2020;Read et al, 2023). PC1 was most influenced negatively by MD, RD and positively by FR and MPF, and was interpreted as positively relating to a 'tissue restriction'…”

“…Density Index (NDI); and Orientation Dispersion (OD). This larger metric space was then reduced through PCA, a technique that has been shown to be effective in capturing biologically informative features of white matter microstructure (Chamberland et al, 2019;Geeraert et al, 2020;Henriques et al, 2023;Read et al, 2023). The results from the PCA (KMO: 0.66, sphericity: p<0.0001; Figure 3) showed that 94% of the microstructure data variance was accounted for by the first two principal components, PC1 and PC2.…”

“…Regarding the microstructure analyses, it should be noted that, while a PCA technique was applied to increase biological interpretability of microstructure features, tractography does not capture fibre properties directly, and interpretations of the PCA components have not been confirmed histologically (see Read et al, 2023 for detailed limitations of this technique).…”

“…Fornix, ILF and PHCB streamlines were generated using the protocols previously described in Read et al (2023). In short, 'way-point' ROIs were manually drawn onto whole-brain FA maps in the diffusion space of 18 subjects, using Explore DTI, to isolate individual tracts.…”

“…We next investigated the relationship between oddity performance and specific tissue properties of the fornix and ILF, alongside the parahippocampal cingulum bundle (PHCB) (Bubb et al, 2018). Tissue properties were obtained from principal components analysis (PCA)-based reduction of a range of indices derived from advanced biophysical (Neurite Orientation and Dispersion Density Imaging, NODDI; Composite Hindered And Restricted Model of Diffusion, CHARMED) and standard tensor-derived (DTI) models applied to multi-shell dMRI data, together with indices derived from quantitative Magnetization Transfer (qMT) imaging (Chamberland et al, 2019;Geeraert et al, 2020;Read et al, 2023). We predicted that tissue properties of the fornix would be associated with scene, but not face oddity discrimination, and vice versa for the ILF.…”