A computational knowledge engine for human neuroscience

Beam, E.; Potts, C.; Poldrack, R. A.; Etkin, A.

doi:10.1101/701540

Home/Archive/bioRxiv

bioRxiv · 10.1101/701540

A computational knowledge engine for human neuroscience

Beam, E.·Potts, C.·Poldrack, R. A.·Etkin, A.

View original source ↗Export metadata ↓

Abstract

Functional neuroimaging has been a mainstay of human neuroscience for the past 25 years. The goal for this research has largely been to understand how activity across brain structures relates to mental constructs and computations. However, interpretation of fMRI data has often occurred within knowledge frameworks crafted by experts, which have the potential to reify historical trends and amplify the subjective biases that limit the replicability of findings. 1 In other words, we lack a comprehensive data-driven ontology for structure-function mapping in the human brain, through which we can also test the explanatory value of current dominant conceptual frameworks. Ontologies in other fields are popular tools for automated data synthesis,2, 3 yet relatively few attempts have been made to engineer ontologies in a data-driven manner.4 Here, we employ a computational approach to derive a data-driven ontology for neurobiological domains that synthesizes the texts and data of nearly 20,000 human neuroimaging articles. The data-driven ontology includes 6 domains, each defined by a circuit of brain structures and its associated mental functions. Several of these domains are omitted from the leading framework in neuroscience, while others uncover novel combinations of mental functions related to common brain circuitry. Crucially, the structure-function links in each domain better replicate across articles in held-out data than those mapped from the dominant frameworks in neuroscience and psychiatry. We further show that the data-driven ontology partitions the literature into modular subfields, for which the domains serve as generalizable archetypes of the structure-function patterns observed in single articles. The approach to computational ontology we present here is the most comprehensive functional characterization of human brain circuits quantifiable with fMRI. Moreover, our methods can be extended to synthesize other scientific literatures, yielding ontologies that are built up from the data of the field.

Explore related subjects

neuroscience Preprint

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Beam, E., Potts, C., Poldrack, R. A., Etkin, A.. 2019-07-14. A computational knowledge engine for human neuroscience. https://doi.org/10.1101/701540

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related preprints

bioRxiv2026

Next generation Human Neocortical Neurosolver model with biophysically realistic somatic, dendritic and synaptic dynamics

The Human Neocortical Neurosolver (HNN) is an open-source biophysical modeling tool for multiscale interpretation of human electroencephalography (EEG) and magnetoencephalography (MEG) signals. HNN's foundation is a pre-tuned canonical neocortical circuit model consisting of excitatory pyramidal neurons and inhibitory interneurons in supra- (L2/3) and infragranular (L5) layers. This model was originally developed based on studies in rodent and non-human primate cortices when dynamics of human neurons were poorly characterized. Yet recent advances in recordings from human tissue now provide unprecedented detail to update the model. Here, we present a significantly updated model for use in the HNN software, with electrophysiological properties and synaptic connections tuned to human tissue. Beyond reproducing biologically realistic somatic dynamics, the pyramidal cells simulate dendritic calcium (Ca2+) spikes in the L5 extratelencephalic-projecting (ET) neurons, and shorter dendritic spikes in the L2/3 pyramidal neurons. We demonstrate how the updated circuitry and new features in the HNN software can be used to disentangle somatic, synaptic, and dendritic contributions to human E-/MEG signatures. The model is distributed with workflows to simulate event-related responses and oscillations, and tutorials demonstrating how membrane potentials, Ca2+ dynamics, and synaptic currents can be recorded and visualized. These enhancements allow extensive model validations based on in vitro and in vivo electrophysiology, voltage and Ca2+ imaging, and provide users with novel insights on the mechanisms underlying E-/MEG signals.

neuroscience↗

bioRxiv2026

A biological light sheet links cellular voltage to mesoscopic brain dynamics

Membrane voltage is the fundamental currency of neural computation, yet cellular-resolution voltage imaging of large neuronal populations in vivo remains limited by trade-offs among imaging speed, field of view, and optical sectioning. We overcome this challenge by encoding a biological light sheet in the brain: Selective Plane Expression (SPress) restricts fluorescent indicator expression to a user-defined, 20-m-thick plane, suppressing background at its source. Following one-time two-photon photoactivation, SPress enables standard widefield microscopes to capture high-contrast kilohertz voltage dynamics from hundreds of densely labeled neurons across millimeter-scale cortical sheets and columns in vivo. These recordings reveal mesoscopic subthreshold waves that flexibly reorganize population-wide spike timing, together with cellularscale wavefront microstructure that preserves stereotyped local recruitment biases. By shifting optical sectioning from instrumentation to gene expression, SPress makes kilohertz cellular resolution voltage imaging accessible across mesoscopic scales.

neuroscience↗

bioRxiv2026

Persistent hippocampal traces of lifelong spatial memory

Memory presents a biological paradox: the brain changes continuously, yet a brief experience can influence behavior for a lifetime. How such memories are stored in brain circuits remains unclear. Here we show that two days in a binary-tree maze leave mice with spatial memories that survive repeated 90-day absences over as much as a year. Across return visits, mice explored the maze more efficiently than when naive and resumed direct runs to a reward location sooner, both by a factor of 3. Chronic calcium imaging of the same hippocampal CA1 neurons showed that although most neurons had changed their preferred locations, the population activity at a given maze location still resembled the pattern from 90 days earlier. Furthermore, even when a neuron's place fields changed location, they tended to remain at the same branching level of the binary tree, reflecting the local geometry of the maze. In return sessions where the neural tuning for branching level was preserved more accurately, the mouse explored the environment more efficiently. Population patterns were best preserved near the reward location, and more reliable neural activity there accompanied earlier direct runs to the reward. With repeated experience, spatial representations changed less, both within a visit and across the long absences. A brief experience thus left selective hippocampal traces that survived months of absence despite drift, grew more durable with each return, and were correlated with memory in navigation behavior.

neuroscience↗