Research map / Probabilistic and scientific modeling / Dynamical systems
A generative model of the hippocampal formation trained with theta driven local learning rules
Tom George, Kim Stachenfeld, Caswell Barry, Claudia Clopath, Tomoki Fukai
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Run the network in wake mode (theta=1) after training and compare apical prediction to basal input | Theta-gated Helmholtz network (linear, Nz=5, Np=50) · Artificial latent learning task | apical vs basal match (qualitative) | Improved |
| Run the trained network freely in sleep mode (theta=0, no input) and compare generated autocorrelation to the true latents | Theta-gated Helmholtz network · Artificial latent learning task | autocorrelation match (qualitative) | Improved |
| Train 30 minutes of simulated time with local wake-sleep learning | Theta-gated Helmholtz network · Artificial latent learning task | prediction error (apical minus basal) | Improved |
| Sensory lesion by switching to sleep mode for 20 seconds after 10 s of normal operation | Trained ring-attractor model · 1D circular track path integration | HPC position decoding error | Mixed |
| Use plastic learned recurrent and conjunctive weights, giving centre-surround recurrence and asymmetric conjunctive weights | MEC hidden layer · 1D circular track | emergence of ring attractor | Improved |
| Allow plasticity on both top-down and bottom-up hidden-layer synapses | MEC hidden layer · 1D circular track | fraction of active MEC neurons | Hurt |
| Use random non-Gaussian spatially tuned sensory inputs instead of Gaussian place-field inputs (ablation) | HPC sensory layer · 1D circular track | path integration learning | No clear effect |
| Use random instead of one-to-one connectivity from hidden units to conjunctive cells (ablation) | MEC conjunctive cells · 1D circular track | path integration learning | No clear effect |
| Shuffle HPC sensory inputs (remapping) with MEC recurrent weights frozen | Trained path-integrating model · Remapping experiment | path integration error after 1 s lesion | Hurt |
| Retrain only HPC<->MEC weights for 30 minutes after remapping, keeping ring attractor weights fixed | Trained path-integrating model · Remapping experiment | path integration error after 1 s lesion | Improved |
| Fix inference weights wgB to identity so MEC units are unimodal (ablation) | MEC in remapping experiment · Remapping experiment | MEC receptive field form | No clear effect |
Closest papers by meaning
- From Cortical Synchronous Rhythm to Brain Inspired Learning Mechanism: An Oscillatory Spiking Neural Network with Time-Delayed Coordination (NeurIPS 2026 · similarity 0.87)
- Learning the Plasticity: Plasticity-Driven Learning Framework in Spiking Neural Networks (NeurIPS 2025 · similarity 0.87)
- BioOSS: A Bio-Inspired Oscillatory State System with Spatio-Temporal Dynamics (NeurIPS 2025 · similarity 0.87)
- Discovering plasticity rules that organize and maintain neural circuits (NeurIPS 2024 · similarity 0.86)
- A Multi-Region Brain Model to Elucidate the Role of Hippocampus in Spatially Embedded Decision-Making (ICML 2025 · similarity 0.86)
- Neuroformer: Multimodal and Multitask Generative Pretraining for Brain Data (ICLR 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
