High-resolution structural and functional imaging of visual and memory systems

Primates, including humans, possess unique capabilities in visual functions, such as visually-guided grasping and object recognition. An accurate understanding of the evolution of the visual system is essential for revealing how brain evolution leads to primates’ specialization in these visual functions. Over the past several decades, visual neuroscientists have highlighted the roles of two distinct visual streams: the dorsal stream for spatial information processing and the ventral stream for categorical information processing (Ungerleider & Mishkin, 1982; Goodale & Milner, 1992). Recent diffusion-weighted MRI (dMRI) studies have renewed interest in the vertical occipital fasciculus (VOF), a white matter pathway connecting the human brain's dorsal and ventral visual cortices (Yeatman et al., 2014; Takemura et al., 2016). We examined the evolutionary trajectory of this pathway by analyzing dMRI data acquired from the brains of 12 mammalian species, including primates and non-primates (Takemura et al., 2024). Our findings suggest that the VOF is a prominent feature of the visual system preserved across all investigated primate species. In contrast, we did not find clear evidence for the VOF in all non-primate dMRI datasets. To complement the finding in non-primates dMRI data, we analyzed the rat brain's 3D-polarized light imaging dataset (Zilles et al., 2016; Schubert et al., 2016), providing micrometer-resolution visualization of fiber orientation. We discuss how the evolution of the VOF may be interpreted in the context of inter-species differences in the spatial organization of white matter pathways and primates' specialization of visual functions.  

The hippocampus is critical for a range of cognitive functions, particularly episodic memory, with well-documented functional differences along its anterior–posterior axis. These functional gradients are thought to arise from underlying patterns of anatomical connectivity between the hippocampus and the rest of the brain. However, much of what we know about these connectivity patterns comes from tract-tracing studies in rodents and non-human primates. Due to technical limitations, the anatomical connectivity of the in vivo human hippocampus has remained poorly characterized, with efforts to map these connections in sufficient detail only recently becoming feasible.
In this talk, I will present findings from two experiments that utilize our newly developed precision tractography pipeline to map the anatomical connectivity of the in vivo human hippocampus with unprecedented detail. I will describe two key studies. First, using high-quality diffusion MRI data from the Human Connectome Project, we systematically assessed and quantified anatomical connections between the hippocampus and the cortical mantle. Second, using a high-resolution diffusion MRI dataset acquired at 760 μm isotropic resolution, we mapped the anatomical connections more specifically between the entorhinal cortex (EC) and hippocampus with exceptional spatial precision, revealing the complexity of these connections.
For both studies, we performed meticulous manual segmentation of the hippocampus and EC to ensure anatomical accuracy and implemented a tailored DWI processing pipeline specifically designed to map hippocampal connectivity. This pipeline enabled us to track streamlines into the hippocampus and generate ‘endpoint density maps’ using Track Density Imaging. These maps allowed us to quantitatively assess, visualise, and map the spatial distribution and density of streamline endpoints within the hippocampus corresponding to each cortical area.
Our results revealed striking patterns of preferential connectivity along the anterior–posterior and medial–lateral axes of the hippocampus. Importantly, we observed distinct patterns of connections between different cortical areas and specific hippocampal subfields. These findings provide new insights into the neuroanatomical architecture underpinning hippocampal-dependent memory systems in the human brain and highlight the structural heterogeneity of hippocampal connectivity along its anterior-posterior axis. 

A hallmark of episodic memory is the ability to flexibly recombine information across episodes to form new associations and guide behaviour. This process, termed associative inference, relies on the hippocampus and surrounding medial temporal lobe (MTL) subregions. We previously found that cross-episode binding is improved when episodes are linked by scenes rather than by faces or objects. Here we tested whether differential recruitment of category-selective MTL subregions underlies these behavioural differences. Participants (n=31) completed study-test phases of the Associative Inference in Memory task, while undergoing high-resolution fMRI scanning at 3 tesla. During the study phase, participants encoded overlapping AB and BC pairs. A and C items were always objects, but the linking B item was either a face or a scene. At test, memory for the direct (AB, BC pairs) and indirect associations (inferred AC pairs) was tested. No face-selective regions were identified within the MTL based on an independent functional localiser scan. Within the MTL, the anterior hippocampal head, a region of interest which included the cornu ammonis 2, 3 and dentate gyrus (CA2-3/DG), anterolateral and posteromedial entorhinal cortex (alERC and pmERC), perirhinal cortex (PRC), and parahippocampal cortex (PHC) were identified as scene-selective (i.e., more active for scenes compared to faces). Although accuracy of the indirect inferences did not differ between associative pairs linked by faces and scenes, MTL cortex subregion recruitment differed across categories. Subregions in the MTL cortex (alERC, pmERC, PRC and PHC), but not the hippocampus (anterior hippocampal head, CA2-3/DG), were recruited to support associative inference for faces during encoding. These findings suggest regions in the MTL cortex identified as scene-selective here may be specialised for integrating disparate elements of episodes into coherent representations, and may be recruited when integration demands during encoding are high (e.g., during associative inference) even for non-scene stimuli. 

A crucial function of memory is to distinguish between similar experiences. Mnemonic discrimination paradigms that require participants to tell apart targets and highly similar lures can probe a brain region’s ability to resolve such interference and form high-fidelity memories. Based on these paradigms the hippocampal subfields dentate gyrus (DG) and CA3 were previously shown to engage in pattern separation (PS), a process to orthogonalize similar stimuli to reduce overlap. Cortical regions may additionally contribute to pattern separation in a category-specific manner with some regions being specialized for objects and others for scenes.
Ultra-high resolution 7-Tesla imaging was used to identify category-specific and -invariant neural correlates of mnemonic discrimination. This method could overcome limitations of prior 3T human functional imaging studies which were unable to distinguish signals from the DG and CA3 hippocampal subfields, did not carry out voxel-wise analyses to test for differences along the long axis of the hippocampus, and left out the amygdala despite its involvement in memory networks.
Young adults completed an object and scene mnemonic discrimination task in a 7-Tesla MRI scanner. Medial temporal lobe (MTL) subregions were manually delineated. Analyses identified regions with a pattern separation-like response of higher activation to lures compared to repeated targets. Such a response is indicative of a region treating a highly similar lure as a novel stimulus.
A voxel-wise analysis showed that across all trials, objects engaged visual ventral areas, perirhinal cortex (BA35+36), and the amygdala relatively more than scene trials. Scenes elicited greater responses in pmERC, subiculum, PHC, and parietal cortex. For the contrast that examined PS-like activity specifically, scene mnemonic discrimination engaged the precuneus, right perirhinal and parahippocampal cortex, and in the hippocampus the posterior DG, CA1, and subiculum subfields. For objects, this was the case for left perirhinal cortex, right anterior CA3 and bilateral amygdala and posterior DG. Analysis of individual trial-type responses showed that CA3 activity could not consistently distinguish between highly similar targets and lures. Consistent with pattern separation, signals in DG and amygdala could make this distinction for objects, while entorhinal cortex did so for objects and scenes. DG activity during mnemonic discrimination was associated with better memory performance, while CA3 and cortical activity were not.
These findings provide further support for a partial distinction between a posterior-medial and anterior-temporal network biased towards scene and object processing, respectively. Regions supporting mnemonic discrimination for scenes were strongly lateralized to the right hemisphere. PS-like responses were largely restricted to the posterior hippocampus. Crucially, several other MTL regions showed responses indicative of pattern separation, more so than in prior 3T studies. The use of 7-Tesla data also made it possible to identify distinct response patterns for DG and CA3 subfields, suggesting that the DG has a greater capacity to resolve feature interference. Importantly, PS-like activation specifically in DG, but not CA3, correlated with better memory performance. These findings provide a more fine-grained map of medial temporal lobe contributions to object and scene memory and highlight the behavioral relevance of dentate gyrus signals for memory fidelity.