A multilevel account of hippocampal function in spatial and concept learning: Bridging models of behavior and neural assemblies

Embedding emotion concepts in cognitive maps

Emotion knowledge is organized in a two-dimensional space known as the affective circumplex, which is thought to develop from core affective feelings and the co-occurrence of emotional events. Neural studies reveal that emotion concepts and cognitive maps of space and abstract concepts are represented in hippocampal-prefrontal systems. We propose that the circumplex is formed by learning the transitions between emotion concepts, a process mediated by a reciprocal network involving hippocampal cells that encode emotion concepts and grid cells in medial entorhinal and ventral prefrontal cortices that encode the relations between them. We anticipate that testing this hypothesis will shed light on the debate about whether emotions are biologically basic or constructed from core affective dimensions.

A unified neural representation model for spatial and conceptual computations

The hippocampus and entorhinal cortex encode spaces by spatially local and hexagonal grid activity patterns (place cells and grid cells), respectively. In addition, the same brain regions also implicate neural representations for nonspatial, semantic concepts (concept cells). These observations suggest that neurocomputational mechanisms for spatial knowledge and semantic concepts are related in the brain. However, the exact relationship remains to be understood. Here, we show a mathematical correspondence between a value function for goal-directed spatial navigation and an information measure for word embedding models in natural language processing. Based on this relationship, we integrate spatial and semantic computations into a neural representation model called "disentangled successor information" (DSI). DSI generates biologically plausible neural representations: spatial representations like place cells and grid cells, and concept-specific word representations which resemble concept cells. Furthermore, with DSI representations, we can perform inferences of spatial contexts and words by a common computational framework based on simple arithmetic operations. This computation can be biologically interpreted by partial modulations of cell assemblies of nongrid cells and concept cells. Our model offers a theoretical connection of spatial and semantic computations and suggests possible computational roles of hippocampal and entorhinal neural representations.

On the origin of memory neurons in the human hippocampus

The hippocampus is essential for episodic memory, yet its coding mechanism remains debated. In humans, two main theories have been proposed: one suggests that concept neurons represent specific elements of an episode, while another posits a conjunctive code, where index neurons code the entire episode. Here, we integrate new findings of index neurons in humans and other animals with the concept-specific memory framework, proposing that concept neurons evolve from index neurons through overlapping memories. This process is supported by engram literature, which posits that neurons are allocated to a memory trace based on excitability and that reactivation induces excitability. By integrating these insights, we connect two historically disparate fields of neuroscience: engram research and human single neuron episodic memory research.

Large language models surpass human experts in predicting neuroscience results

Scientific discoveries often hinge on synthesizing decades of research, a task that potentially outstrips human information processing capacities. Large language models (LLMs) offer a solution. LLMs trained on the vast scientific literature could potentially integrate noisy yet interrelated findings to forecast novel results better than human experts. Here, to evaluate this possibility, we created BrainBench, a forward-looking benchmark for predicting neuroscience results. We find that LLMs surpass experts in predicting experimental outcomes. BrainGPT, an LLM we tuned on the neuroscience literature, performed better yet. Like human experts, when LLMs indicated high confidence in their predictions, their responses were more likely to be correct, which presages a future where LLMs assist humans in making discoveries. Our approach is not neuroscience specific and is transferable to other knowledge-intensive endeavours.

Design of a Water-Soluble CD20 Antigen with Computational Epitope Scaffolding

The poor solubility of integral membrane proteins in water frequently hinders studies with these proteins, presenting challenges for structure determination and binding screens. For instance, the transmembrane protein CD20, which is an important target for treating B-cell malignancies, is not soluble in water and cannot be easily screened against potential protein binders with techniques like phage display or yeast display. Here, we use de novo protein design to create a water-soluble mimic of the CD20 dimer ("soluble CD20"). Soluble CD20 replaces the central transmembrane helix of CD20 with a water-soluble helix that dimerizes to form a coiled coil that structurally matches the dimer interface of native CD20 and presents the central extracellular loop of CD20 in a binding competent conformation. Unlike peptides derived from CD20, soluble CD20 binds tightly to monoclonal antibodies that recognize quaternary epitopes on the extracellular face of CD20. We demonstrate that soluble CD20 is easy to produce, remains folded above 60°C, and is compatible with binder screening via yeast display. Our results highlight the ability of computational protein design to scaffold conformational epitopes from membrane proteins for use in binding and protein engineering studies.

Differential reorganization of episodic and semantic memory systems in epilepsy-related mesiotemporal pathology

Declarative memory encompasses episodic and semantic divisions. Episodic memory captures singular events with specific spatiotemporal relationships, whereas semantic memory houses context-independent knowledge. Behavioural and functional neuroimaging studies have revealed common and distinct neural substrates of both memory systems, implicating mesiotemporal lobe (MTL) regions such as the hippocampus and distributed neocortices. Here, we explored declarative memory system reorganization in patients with unilateral temporal lobe epilepsy (TLE) as a human disease model to test the impact of variable degrees of MTL pathology on memory function. Our cohort included 31 patients with TLE and 60 age- and sex-matched healthy controls, and all participants underwent episodic and semantic retrieval tasks during a multimodal MRI session. The functional MRI tasks were closely matched in terms of stimuli and trial design. Capitalizing on non-linear connectome gradient-mapping techniques, we derived task-based functional topographies during episodic and semantic memory states, in both the MTL and neocortical networks. Comparing neocortical and hippocampal functional gradients between TLE patients and healthy controls, we observed a marked topographic reorganization of both neocortical and MTL systems during episodic memory states. Neocortical alterations were characterized by reduced functional differentiation in TLE across lateral temporal and midline parietal cortices in both hemispheres. In the MTL, in contrast, patients presented with a more marked functional differentiation of posterior and anterior hippocampal segments ipsilateral to the seizure focus and pathological core, indicating perturbed intrahippocampal connectivity. Semantic memory reorganization was also found in bilateral lateral temporal and ipsilateral angular regions, whereas hippocampal functional topographies were unaffected. Furthermore, leveraging MRI proxies of MTL pathology, we observed alterations in hippocampal microstructure and morphology that were associated with TLE-related functional reorganization during episodic memory. Moreover, correlation analysis and statistical mediation models revealed that these functional alterations contributed to behavioural deficits in episodic memory, but again not in semantic memory in patients. Altogether, our findings suggest that semantic processes rely on distributed neocortical networks, whereas episodic processes are supported by a network involving both the hippocampus and the neocortex. Alterations of such networks can provide a compact signature of state-dependent reorganization in conditions associated with MTL damage, such as TLE.

Hippocampal connectivity patterns echo macroscale cortical evolution in the primate brain

While the hippocampus is key for human cognitive abilities, it is also a phylogenetically old cortex and paradoxically considered evolutionarily preserved. Here, we introduce a comparative framework to quantify preservation and reconfiguration of hippocampal organisation in primate evolution, by analysing the hippocampus as an unfolded cortical surface that is geometrically matched across species. Our findings revealed an overall conservation of hippocampal macro- and micro-structure, which shows anterior-posterior and, perpendicularly, subfield-related organisational axes in both humans and macaques. However, while functional organisation in both species followed an anterior-posterior axis, we observed a marked reconfiguration in the latter across species, which mirrors a rudimentary integration of the default-mode-network in non-human primates. Here we show that microstructurally preserved regions like the hippocampus may still undergo functional reconfiguration in primate evolution, due to their embedding within heteromodal association networks.