Empirical scaling laws in balanced networks with conductance-based synapses
Vicky Zhu, Gabriel Ocker, Robert Rosenbaum
TLDR
Recurrent neural networks with conductance-based synapses and spike correlations produce realistic membrane potential variability.
Key contributions
- Shows conductance-based synapses alone yield unrealistically low membrane potential variability.
- Demonstrates current-based models with spike correlations yield unrealistically high variability.
- Finds that combining conductance-based synapses and spike correlations produces realistic variability.
- Highlights that realistic dynamics require both modeling assumptions together.
Why it matters
This paper resolves a long-standing issue in neural network modeling by demonstrating how to achieve more realistic dynamics. It shows that combining conductance-based synapses and spike correlations is crucial for accurate predictions of cortical activity. This advances our understanding of neural network behavior.
Original Abstract
Strongly coupled, recurrent, balanced network models have been successful in describing and predicting many phenomena observed in cortical neural recordings. However, most balanced network models use current-based synapse models in place of more realistic, conductance-based models. Conductance-based synapse models predict unrealistically small membrane potential variability. On the other hand, introducing realistic levels of spike time correlations to models with current-based synapses predicts unrealistically large membrane potential variability. We use computer simulations to show that these two effects can cancel: Recurrent network models with conductance-based synapses and spike time correlations produce more realistic, moderate levels of membrane potential variability. Consistent with recent work on feedforward networks, our results show that including more realistic modeling assumptions produces more realistic dynamics, but only if when two modeling assumptions are included together.
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.