Efficient neural decoding of self-location with a deep recurrent network

Abstract

Place cells in the mammalian hippocampus signal self-location with sparse spatially stable firing fields. Based on observation of place cell activity it is possible to accurately decode an animal’s location. The precision of this decoding sets a lower bound for the amount of information that the hippocampal population conveys about the location of the animal. In this work we use a novel recurrent neural network (RNN) decoder to infer the location of freely moving rats from single unit hippocampal recordings. RNNs are biologically plausible models of neural circuits that learn to incorporate relevant temporal context without the need to make complicated assumptions about the use of prior information to predict the current state. When decoding animal position from spike counts in 1D and 2D-environments, we show that the RNN consistently outperforms a standard Bayesian approach with either flat priors or with memory. In addition, we also conducted a set of sensitivity analysis on the RNN decoder to determine which neurons and sections of firing fields were the most influential. We found that the application of RNNs to neural data allowed flexible integration of temporal context, yielding improved accuracy relative to the more commonly used Bayesian approaches and opens new avenues for exploration of the neural code.

A Probabilistic Framework for Decoding Behavior From in vivo Calcium Imaging Data

“…In particular, including pairwise (Pillow et al, 2008;Meshulam et al, 2017) or temporal (Naud and Gerstner, 2012) correlations of neuronal activity could reduce decoding error. Note that these temporal correlations would also be taken into account when using long shortterm memory (LSTM) artificial neural networks (Tampuu et al, 2019), thus increasing reconstruction accuracy at the expense of interpretability. Rather than proposing a sophisticated analysis pipeline, the methods presented here have the advantage of remaining simple, requiring only few data points, and are easily interpretable using metrics that can facilitate the communication of results along with significance and confidence intervals, making it an appropriate tool for exploration of calcium imaging data in conjunction with behavior.…”

Decoding speech from spike-based neural population recordings in secondary auditory cortex of non-human primates

“…Previous work has shown that increased neuronal firing rates correlate strongly with feature importance when using LSTM RNN models for neural decoding 35 . Therefore, to investigate the effect of channel count on decoding performance, we ordered neural channels according to the highest neural activity (i.e., highest counts of threshold crossings) over the training data set.…”

On the open-source landscape of PLOS Computational Biology

“…However, the required skills can be developed within labs in a short time frame. As an example, we took two recently published papers in PLOS Computational Biology [ 3 , 4 ] and customized the code provided by the authors to create NeuroLibre-style Jupyter Books with interactive figures using Plotly ( Example 1 and Example 2 ). As part of an experimental collaboration between NeuroLibre and PLOS Computational Biology , the journal team is seeking user feedback on these examples to further understand the value of these features to the journal’s community.…”