Noise in Neurons and Synapses Enables Reliable Associative Memory Storage in Local Cortical Circuits

Abstract

Neural networks in the brain can function reliably despite various sources of errors and noise present at every step of signal transmission. These sources include errors in the presynaptic inputs to the neurons, noise in synaptic transmission, and fluctuations in the neurons’ postsynaptic potentials (PSPs). Collectively they lead to errors in the neurons’ outputs which are, in turn, injected into the network. Does unreliable network activity hinder fundamental functions of the brain, such as learning and memory retrieval? To explore this question, this article examines the effects of errors and noise on the properties of model networks of inhibitory and excitatory neurons involved in associative sequence learning. The associative learning problem is solved analytically and numerically, and it is also shown how memory sequences can be loaded into the network with a biologically more plausible perceptron-type learning rule. Interestingly, the results reveal that errors and noise during learning increase the probability of memory recall. There is a trade-off between the capacity and reliability of stored memories, and, noise during learning is required for optimal retrieval of stored information. What is more, networks loaded with associative memories to capacity display many structural and dynamical features observed in local cortical circuits in mammals. Based on the similarities between the associative and cortical networks, this article predicts that connections originating from more unreliable neurons or neuron classes in the cortex are more likely to be depressed or eliminated during learning, while connections onto noisier neurons or neuron classes have lower probabilities and higher weights.

Random noise promotes slow heterogeneous synaptic dynamics important for robust working memory computation

“…Our findings are closely related to the previous studies that reported the benefits of random neural noise ubiquitous in the cortex in memory recall and associative learning [Zhang et al, 2021, McDonnell and Ward, 2011]. For example, recent experiments showed that a high level of noise and randomness in the olfactory system (i.e., random and seemingly unstructured networks in the piriform cortex) allows for not only flexible encoding of sensory information but also maintenance of the encoded information [Caron et al, 2013, Pashkovski et al, 2020, Litwin-Kumar et al, 2017, Zhang et al, 2019].…”

R-MaS3N: Robust Mapping of Spiking Neural Networks to 3D-NoC-Based Neuromorphic Systems for Enhanced Reliability

“…To execute SNNs on neuromorphic hardware, neurons and synapses must be mapped to their neuronal and synaptic circuits. However, the brain's neuronal circuit could be susceptible to faults from various sources, such as noise in synaptic transmission and fluctuations in post-synaptic potentials [12], eventually leading to errors in neuronal circuit outputs. Likewise, neural circuits in neuromorphic hardware may fail due to electrical interference and cross-talk between neurons [12] [13].…”

Does increased gait variability improve stability when faced with an expected balance perturbation during treadmill walking?

“…The opposing interpretation argues that movement variability is an exploratory behaviour, which allows for rich sensory input from the environment [4]. To ensure the reliability of neural networks, it is best to optimize errors and noise, which are unavoidable, and learn to retrieve motor memories in their constant presence [5]. Thus, rather than attempting to supress errors and noise, the brain learns to exploit them to reliably store memories [5].…”

“…To ensure the reliability of neural networks, it is best to optimize errors and noise, which are unavoidable, and learn to retrieve motor memories in their constant presence [5]. Thus, rather than attempting to supress errors and noise, the brain learns to exploit them to reliably store memories [5].…”