Search Papers | Poster Sessions | All Posters
Poster C19 in Poster Session C - Friday, August 9, 2024, 11:15 am – 1:15 pm, Johnson Ice Rink
Partial observation can induce mechanistic mismatches in data-constrained RNNs
William Qian1, Jacob Zavatone-Veth1, Benjamin Ruben1, Cengiz Pehlevan1; 1Harvard University
One of the central goals of computational neuroscience is to understand how the dynamics of neural circuits give rise to their observed function. A popular approach towards this end is to train recurrent neural networks (RNNs) to reproduce experimental recordings of neural activity. These trained RNNs are then treated as surrogate models of biological neural circuits, whose properties can be dissected via dynamical systems analysis. While recent advances in population-level recording technologies have allowed simultaneous recording of up to tens of thousands of neurons, this represents only a tiny fraction of most cortical circuits. Here we show that partial observation can create mechanistic mismatches between a simulated teacher network and a data-constrained student, even when the two networks have otherwise matching architectures. In particular, we show that partial observation of models of working memory in cortex based on functionally feedforward or low-rank connectivity can lead to surrogate models with spurious attractor structure.
Keywords: recurrent neural networks short-term memory dynamical systems data-driven modeling