What is it about?
We don't really know how the neurons in the brain process information. One idea is that they detect coincidences in their input (they answer whenever they see two or more stimulus pulses at the same time), another idea is that they integrate over their input (the speed of their answer pulses reflects the speed of the stimulus pulses they see). This paper suggests a way to measure how much of integration and how much of coincidence detection a neuron is doing.
Featured Image
Photo by Robina Weermeijer on Unsplash
Why is it important?
The discussion about whether neurons are detecting coincidences or rather integrating has been going on for a long time. Many people have written about the subject, but each publication measured this in a different way, many of them very informally, and have come up with different results. To make such studies comparable, our paper proposes a mathematically rigid definition of coincidence detection and integration. We also added a new mode - gap detection - and we found that a neuron can integrate and detect coincidences at the same time, on different parts of its stimulus.
Perspectives
Writing this paper was great fun, in particular as the collaboration with my co-authors was a great pleasure. After long discussions, numerous tests, and several major re-writes of the text, I think we have come up with a new way of seeing what neurons are doing, and I hope our results will form the basis of many more interesting studies — ultimately leading to our understanding how that large computer that is our brain actually works.
Jacob Kanev
Technische Universitat Berlin
Read the Original
This page is a summary of: Integrator or Coincidence Detector: A Novel Measure Based on the Discrete Reverse Correlation to Determine a Neuron’s Operational Mode, Neural Computation, October 2016, The MIT Press,
DOI: 10.1162/neco_a_00875.
You can read the full text:
Resources
Integrator or Coincidence Detector: A Novel Measure Based on the Discrete Reverse Correlation to Determine a Neuron’s Operational Mode
Original Article in Neural Computation
Source Code of C++ Library used for Simulations in the Paper
NeuroLab is a C++ class library which offers simulation of stochastic differential (integral) equations, as well as recording and display of data. Intended audience are neuroscience researchers who want to run simulations of single or multiple connected neurons, analyse the results, and display / save to various file formats.
Neurolab Documentation
Online documentation of the neurolab C++ library.
Contributors
The following have contributed to this page
