Learning the Synaptic and Intrinsic Membrane Dynamics Underlying Working Memory in Spiking Neural Network Models

Abstract Recurrent neural network (RNN) models trained to perform cognitive tasks are a useful computational tool for understanding how cortical circuits execute complex computations. However, these models are often composed of units that interact with one another using continuous signals and overlook parameters intrinsic to spiking neurons. Here, we developed a method to directly train not only synaptic-related variables but also membrane-related parameters of a spiking RNN model. Training our model on a wide range of cognitive tasks resulted in diverse yet task-specific synaptic and membrane parameters. We also show that fast membrane time constants and slow synaptic decay dynamics naturally emerge from our model when it is trained on tasks associated with working memory (WM). Further dissecting the optimized parameters revealed that fast membrane properties are important for encoding stimuli, and slow synaptic dynamics are needed for WM maintenance. This approach offers a unique window into how connectivity patterns and intrinsic neuronal properties contribute to complex dynamics in neural populations.

[1]  Eimantas Ledinauskas,et al.  Training Deep Spiking Neural Networks , 2020, ArXiv.

[2]  B. Connors,et al.  Intrinsic firing patterns of diverse neocortical neurons , 1990, Trends in Neurosciences.

[3]  Xiao-Jing Wang Decision Making in Recurrent Neuronal Circuits , 2008, Neuron.

[4]  S. Thorpe,et al.  Spike times make sense , 2005, Trends in Neurosciences.

[5]  Guangyu R. Yang,et al.  Training Excitatory-Inhibitory Recurrent Neural Networks for Cognitive Tasks: A Simple and Flexible Framework , 2016, PLoS Comput. Biol..

[6]  Terrence J. Sejnowski,et al.  Simple framework for constructing functional spiking recurrent neural networks , 2019, Proceedings of the National Academy of Sciences.

[7]  R. Romo,et al.  Neuronal correlates of parametric working memory in the prefrontal cortex , 1999, Nature.

[8]  Quoc V. Le,et al.  Searching for Activation Functions , 2018, arXiv.

[9]  Timothée Masquelier,et al.  Deep Learning in Spiking Neural Networks , 2018, Neural Networks.

[10]  Wulfram Gerstner,et al.  Neuronal Dynamics: From Single Neurons To Networks And Models Of Cognition , 2014 .

[11]  Devika Narain,et al.  Flexible sensorimotor computations through rapid reconfiguration of cortical dynamics , 2018 .

[12]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[13]  R. Tremblay,et al.  GABAergic Interneurons in the Neocortex: From Cellular Properties to Circuits , 2016, Neuron.

[14]  Kazuyuki Aihara,et al.  Chaotic simulated annealing by a neural network model with transient chaos , 1995, Neural Networks.

[15]  E. Halgren,et al.  Spatiotemporal dynamics of neocortical excitation and inhibition during human sleep , 2012, Proceedings of the National Academy of Sciences.

[16]  Kaushik Roy,et al.  Going Deeper in Spiking Neural Networks: VGG and Residual Architectures , 2018, Front. Neurosci..

[17]  Paul J. Werbos,et al.  Backpropagation Through Time: What It Does and How to Do It , 1990, Proc. IEEE.

[18]  C. Petersen,et al.  Cell-Type-Specific Sensorimotor Processing in Striatal Projection Neurons during Goal-Directed Behavior , 2015, Neuron.

[19]  Hilbert J. Kappen,et al.  Learning Universal Computations with Spikes , 2015, PLoS Comput. Biol..

[20]  Wenrui Zhang,et al.  Spike-Train Level Backpropagation for Training Deep Recurrent Spiking Neural Networks , 2019, NeurIPS.

[21]  Tobi Delbrück,et al.  Training Deep Spiking Neural Networks Using Backpropagation , 2016, Front. Neurosci..

[22]  Y. Ermoliev,et al.  The Minimization of Semicontinuous Functions: Mollifier Subgradients , 1995 .

[23]  Emre Neftci,et al.  Surrogate Gradient Learning in Spiking Neural Networks: Bringing the Power of Gradient-based optimization to spiking neural networks , 2019, IEEE Signal Processing Magazine.

[24]  Thomas Miconi,et al.  Biologically plausible learning in recurrent neural networks reproduces neural dynamics observed during cognitive tasks , 2016, bioRxiv.

[25]  L. Abbott,et al.  Stimulus-dependent suppression of chaos in recurrent neural networks. , 2009, Physical review. E, Statistical, nonlinear, and soft matter physics.

[26]  Christopher D. Harvey,et al.  Recurrent Network Models of Sequence Generation and Memory , 2016, Neuron.

[27]  Valentin Dragoi,et al.  Efficient coding in heterogeneous neuronal populations , 2008, Proceedings of the National Academy of Sciences.

[28]  Sommers,et al.  Chaos in random neural networks. , 1988, Physical review letters.

[29]  Geoffrey E. Hinton,et al.  Learning representations by back-propagating errors , 1986, Nature.

[30]  Christopher Kim,et al.  Learning recurrent dynamics in spiking networks , 2018, bioRxiv.

[31]  W. Newsome,et al.  Context-dependent computation by recurrent dynamics in prefrontal cortex , 2013, Nature.

[32]  Shyan-Shiou Chen,et al.  Chaotic Simulated Annealing by a Neural Network With a Variable Delay: Design and Application , 2011, IEEE Transactions on Neural Networks.

[33]  Wilten Nicola,et al.  Supervised learning in spiking neural networks with FORCE training , 2016, Nature Communications.

[34]  R. Douglas,et al.  Recurrent neuronal circuits in the neocortex , 2007, Current Biology.

[35]  M. Medalla,et al.  Strength and Diversity of Inhibitory Signaling Differentiates Primate Anterior Cingulate from Lateral Prefrontal Cortex , 2017, The Journal of Neuroscience.

[36]  Yoshua Bengio,et al.  Gradient Flow in Recurrent Nets: the Difficulty of Learning Long-Term Dependencies , 2001 .

[37]  Francesca Mastrogiuseppe,et al.  Linking Connectivity, Dynamics, and Computations in Low-Rank Recurrent Neural Networks , 2017, Neuron.

[38]  P. Goldman-Rakic Cellular basis of working memory , 1995, Neuron.

[39]  Thomas Miconi Biologically plausible learning in recurrent neural networks reproduces neural dynamics observed during cognitive tasks , 2017 .

[40]  L. F. Abbott,et al.  Generating Coherent Patterns of Activity from Chaotic Neural Networks , 2009, Neuron.

[41]  Terrence J. Sejnowski,et al.  Gradient Descent for Spiking Neural Networks , 2017, NeurIPS.

[42]  C. Petersen,et al.  State-dependent cell-type-specific membrane potential dynamics and unitary synaptic inputs in awake mice , 2018, eLife.