Hypothesis-Driven Identification of Neural Algorithms With Dynamical Structure-Preserving Manifolds.
- 1Yale University
Abstract
Neuroscience aims to uncover the algorithms by which the brain builds and manipulates rich internal representations of objects, agents, and places. Existing reverse-engineering frameworks remain ineffective for identifying these algorithms. Here, we present Dynamical Structure-Preserving Manifolds (dSPMs)---a novel multilevel reverse-engineering framework that programs, without training, recurrent neural networks to enable hypothesis-driven identification of the brain's algorithms that generate symbolic, structure-preserving representations of the world around. We apply dSPM to identify an algorithm for physical scene prediction in the dorsomedial frontal cortex (DMFC) of macaques intercepting a ball on a Pong-like virtual gameboard. dSPM better explained DMFC than alternatives, made predictions confirmed in neural data, and revealed an algorithm of DMFC that we term 'manifold-based simulation': perceived initial conditions rapidly collapse neural activity onto a low-dimensional manifold whose geometry encodes a physics-based representation of the scene, supporting far-out predictions without traditional next-step simulation. dSPM supports effective exploration of the computational foundations of biological intelligence.
Related
- Structure-Aware Mapping with Dynamical Sources for Informer Weight Initialization in Time-Series Forecasting
- Point-Process Modelling of Cell-Type Interaction Structure in CosMx Glioma Data for Hypothesis-Driven Tumor Inference from scRNA-seq
- MorphoBranch: A Fine-Structure-Preserving Workbench for Morphometric Analysis of Branched Cellular Structures
- Similar Predictive Fit but Different Latent Dynamics: Characterizing Learned Dynamical Structure in Personalized Models of Brain Disorders
- scFLAME: a unified generative model for interpretable clustering, hierarchical structure discovery and marker-gene identification in single-cell RNA-seq data
BibTeX
@misc{calbick2026hypothesisdriven,
title = {{Hypothesis-Driven Identification of Neural Algorithms With Dynamical Structure-Preserving Manifolds.}},
author = {D. Calbick and J. Z. Kim and H. Sohn and I. Yildirim},
howpublished = {bioRxiv},
year = {2026},
month = oct,
doi = {10.64898/2026.10.09.755459},
url = {https://www.biorxiv.org/content/10.64898/2026.10.09.755459v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Hypothesis-Driven Identification of Neural Algorithms With Dynamical Structure-Preserving Manifolds.
AU - D. Calbick
AU - J. Z. Kim
AU - H. Sohn
AU - I. Yildirim
PY - 2026
DA - 2026/10/09
PB - bioRxiv
DO - 10.64898/2026.10.09.755459
UR - https://www.biorxiv.org/content/10.64898/2026.10.09.755459v1
AB - Neuroscience aims to uncover the algorithms by which the brain builds and manipulates rich internal representations of objects, agents, and places. Existing reverse-engineering frameworks remain ineffective for identifying these algorithms. Here, we present Dynamical Structure-Preserving Manifolds (dSPMs)---a novel multilevel reverse-engineering framework that programs, without training, recurrent neural networks to enable hypothesis-driven identification of the brain's algorithms that generate symbolic, structure-preserving representations of the world around. We apply dSPM to identify an algorithm for physical scene prediction in the dorsomedial frontal cortex (DMFC) of macaques intercepting a ball on a Pong-like virtual gameboard. dSPM better explained DMFC than alternatives, made predictions confirmed in neural data, and revealed an algorithm of DMFC that we term 'manifold-based simulation': perceived initial conditions rapidly collapse neural activity onto a low-dimensional manifold whose geometry encodes a physics-based representation of the scene, supporting far-out predictions without traditional next-step simulation. dSPM supports effective exploration of the computational foundations of biological intelligence.
ER -