Biology subjects

Caze, R. D.

Publications and source records attributed to Caze, R. D..

3 recordsLinked to original sources

bioRxiv2023

Demonstration that sublinear dendrites enable linearly non-separable computations

Theory predicts that nonlinear summation of synaptic potentials within dendrites allows neurons to perform linearly non-separable computations (LNSCs). Using Boolean analysis approaches, we predicted that both supralinear and sublinear synaptic summation could allow single neurons to implement a type of LNSC, the feature binding problem (FBP), which does not require inhibition contrary to the XOR. Notably, sublinear dendritic operations enable LNSCs when scattered synaptic activation generates increased somatic spike output. However, experimental demonstrations of scatter-sensitive neuronal computations have not yet been described. Using glutamate uncaging onto cerebellar molecular layer interneurons, we show that scattered synaptic-like activation of dendrites evoked larger compound EPSPs than clustered synaptic activation, generating a higher output spiking probability. Moreover, we also demonstrate that single interneurons can indeed implement the FBP. We use a biophysical model to predict under what conditions a neuron can implement the FBP and what leads to failures. Experimental results agree with the model-determined conditions and hence validate our protocol as a solid benchmark for a neuron to implement linearly non-separable computations. Since sublinear synaptic summation is a property of passive dendrites we expect that many different neuron types can implement LNSCs.

neuroscience↗

bioRxiv2021

An experimental prediction: granule cells can perform linearly non-separable computations

Multiple studies show how dendrites might extend some neurons computational capacity. Past works most often focus on pyramidal cells which have an extended dendritic tree where voltage can vary independently; pyramidal neurons dendrites emit local spikes turning dendritic branches into local non-linear subunits. However, these studies leave a large fraction of the nervous system unexplored. Here, we study a neuron with a modest dendritic and non-spiking dendrites. Granule cells dendrites do not spike and these cells membrane voltage remain constant over the neuron. This study uses a conjunction of Boolean algebra and biophysical modelling to predict that Granule cells can perform linearly non-separable computations. In a previous study, we found a linearly non-separable Boolean function possible to implement without dendritic spikes, we coined this computation the feature binding problem. In the present work, we use the intrinsic saturation of synaptic conductance to implement this computation in a biophysical model. An integrate and fire, cannot perform such type of computations. Confirming our prediction would change how we understand the nervous system.

neuroscience↗

bioRxiv2020

Dendrites decrease the synaptic weight resolution necessary to compute

In theory, neurons modelled as single layer perceptrons can implement all linearly separable computations. In practice, however, these computations may require arbitrarily precise synaptic weights. This is a strong constraint since both, biological neurons and their artificial counterparts, have to cope with limited precision. Here, we explore how the non-linear processing in dendrites helps overcoming this constraint. We start by finding a class of computations which requires increasing precision with the number of inputs in a perceptron and show that it can be implemented without this constraint in a neuron with sub-linear subunits. Then, we complement this analytical study by a simulation of a biophysical neuron model with two passive dendrites and a soma, and show that it can implement this computation. This works demonstrates a new role of dendrites in neural computation: by distributing the computation across independent subunits, the same computation can be performed more efficiently with less precise tuning of the synaptic weights. We hope that this works not only offers new insight into the importance of dendrites for biological neurons, but also paves the way for new, more efficient architectures of artificial neuromorphic chips. Author SummaryIn theory, we know how much neurons can compute, in practice, the number of possible synaptic weights values limits their computation capacity. Such a limitation holds true for artificial and synthetic neurons. We introduce here a computation where the required means evolve significantly with the number of inputs, this poses a problem as neurons receive multiple thousands of inputs. We study here how the neurons receptive element-called dendrites-can mitigate such a problem. We show that, without dendrites, the largest synaptic weight need to be multiple orders of magnitude larger than the smallest to implement the computation. Yet a neuron with dendrites implements the same computation with constant synaptic weights whatever the number of inputs. This study paves the way for the use of dendritic neurons in a new generation of artificial neural network and neuromorphic chips with a considerably better cost-benefit balance.

neuroscience↗