| name | dic-neuron-reconstruction-spike-times |
| category | neuroscience |
| trigger_words | ["DIC neuron reconstruction","dynamic input conductance","conductance-based neuron model","degenerate neuron population","spike time parameter inference","neuronal degeneracy","neuron model from spike times","biophysical parameter inference"] |
| description | Deep learning + Dynamic Input Conductances (DICs) methodology for fast reconstruction of degenerate conductance-based neuron populations from spike times alone, enabling scalable and interpretable inference from experimental recordings. |
| source | arXiv:2509.12783v2 |
| created | 2026-07-06T00:00:00.000Z |
DIC-Based Neuron Reconstruction from Spike Times
Source: arXiv:2509.12783v2 - "Fast reconstruction of degenerate populations of conductance-based neuron models from spike times" (Julien Brandoit, Damien Ernst, Guillaume Drion, Arthur Fyon, PLOS Computational Biology 2026)
Core Insight
Neuronal degeneracy (multiple distinct conductance sets yielding similar spiking patterns) is NOT a problem to solve but a feature to exploit. By combining deep learning with Dynamic Input Conductances (DICs) - a theoretical framework reducing complex CBMs to three interpretable feedback components - we can reconstruct entire degenerate populations from spike times alone in milliseconds.
What Are DICs?
Dynamic Input Conductances reduce any conductance-based model to three interpretable feedback components governing:
- Excitability - whether the neuron fires
- Firing patterns - how the neuron fires
- Recovery - how the neuron resets
This low-dimensional representation bridges experimentally observed activity and mechanistic models.
Methodology Pipeline
Step 1: Spike Times → DIC Densities
- Train a neural network to map spike times → DIC densities at threshold
- The NN learns a low-dimensional representation of neuronal activity
- Input: spike train timing data only
- Output: three DIC values (excitability, firing, recovery)
Step 2: DIC → Degenerate CBM Populations
- Use an iterative compensation algorithm to generate CBM populations
- Each CBM is compatible with the predicted DICs
- Ensures all generated models reproduce the observed firing patterns
- Works even in high-dimensional models
Step 3: Validation
- Test reconstructed models on noisy current injection (mimicking physiological stochasticity)
- Verify spiking AND bursting regime reconstruction
- Check robustness to variability
Practical Applications
Neuroscience Research
- Parameter inference from experimental spike recordings
- Population-level analysis of neuronal degeneracy
- Model validation against physiological data
- Fast screening of candidate neuron models
Computational Modeling
- Scalable CBM reconstruction from limited data
- Interpretable parameter reduction via DIC framework
- Degenerate population generation for uncertainty quantification
DIC-Based Pipeline Steps
- Collect spike times from experiment or simulation
- Preprocess into consistent spike train format
- Run DIC predictor NN to get three DIC densities
- Apply iterative compensation to generate CBM population
- Validate against original firing patterns
- Analyze degeneracy across the population
Key Advantages
- Speed: Milliseconds on standard hardware
- Interpretability: DICs are physically meaningful
- Scalability: Works with high-dimensional models
- Robustness: Handles noisy spike trains
- Completeness: Produces diverse degenerate populations
Pitfalls
- Ignoring degeneracy: Treating it as noise rather than a feature loses biological insight
- Using raw conductances: Without DICs, the inverse problem is intractable
- Single-model inference: One CBM doesn't capture the full range of compatible models
- Non-spiking data: Method requires spike times; subthreshold voltage data needs different approach