Single neurons can detect sequences

Single neurons can detect sequences
A neuron in the visual cortex of the mouse was filled with a fluorescent dye so that the dendrites could be visualised. A laser was targeted to small spots on single dendrites to activate groups of inputs in different orders. The electrical response of the neuron was recorded and was found to be be different for each of the input sequences. Credit: Tiago Branco/Hausser Lab: UCL

Single neurons in the brain are surprisingly good at distinguishing different sequences of incoming information according to new research by UCL neuroscientists.

The study, published today in Science and carried out by researchers based at the Wolfson Institute for Biomedical Research at UCL, shows that single neurons, and indeed even single dendrites, the tiny receiving elements of neurons, can very effectively distinguish between different temporal sequences of incoming information.

This challenges the widely held view that this kind of processing in the requires large numbers of neurons working together, as well as demonstrating how the basic components of the brain are exceptionally powerful computing devices in their own right.

First author Tiago Branco said: "In everyday life, we constantly need to use information about sequences of events in order to understand the world around us. For example, language, a collection of different sequences of similar letters or sounds assembled into sentences, is only given meaning by the order in which these sounds or letters are assembled.

"The brain is remarkably good at processing sequences of information from the outside world. For example, modern computers will still struggle to decode a rapidly spoken sequence of words that a 5 year-old child will have no trouble understanding. How the brain does so well at distinguishing one sequence of events from another is not well understood but, until now, the general belief has been that this job is done by large numbers of neurons working in concert with each other."

Using a mouse model, the researchers studied neurons in areas of the brain which are responsible for processing sensory input from the eyes and the face. To probe how these neurons respond to variation in the order of a number of inputs, they used a laser to activate inputs on the dendrites in precisely defined patterns and recorded the resulting electrical responses of the neurons.

Surprisingly, they found that each sequence produced a different response, even when it was delivered to a single dendrite. Furthermore, using theoretical modelling, they were able to show that the likelihood that two sequences can be distinguished from each other is remarkably high.

Senior author Professor Michael Hausser commented: "This research indicates that single neurons are reliable decoders of temporal sequences of inputs, and that they can play a significant role in sorting and interpreting the enormous barrage of inputs received by the brain.

"This new property of and dendrites adds an important new element to the "toolkit" for computation in the brain. This feature is likely to be widespread across many brain areas and indeed many different animal species, including humans."

Explore further

Key mechanism in the brain's computation of sound location identified

Citation: Single neurons can detect sequences (2010, August 12) retrieved 20 October 2019 from
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only.

Feedback to editors

User comments

Aug 13, 2010
What I am surprised about, is that despite having that huge computational power in our brain - we still perform very poorly when we need to remember large and discrete data sets, like remembering many long passwords.

Aug 13, 2010
Actually this behavior of neuron is well known for many years. By Hebbian theory of synaptic plasticity this aspect of soliton behavior is represented by principle "cells that fire together, wire together". When one cell repeatedly assists in firing another, the axon of the first cell develops synaptic knobs (or enlarges them if they already exist) in contact with the soma of the second cell. In this mechanism the principle of associative learning and long-term memory is realized.
.. we still perform very poorly when we need to remember large and discrete data sets...
It depends. I'm remembering a content of roughly 20.000 scientific articles, for example (I've a database of them). Of course many articles are related mutually, so that the total volume of information may be lower. Human brain wasn't required to remember large discrete value sets during his evolution - instead of it it was required to recognize fractal shape of countryside, faces of people and so on...

Aug 13, 2010
@ Tahoma, this is not, in fact, what the study shows. While the wiring up of the network is most likely activity dependent, as many years of research have suggested, Hebbian plasticity is not the mechanism that explains the ability of dendrites to distinguish patterns of synaptic input. Instead it relies purely on the location of synapses on the dendritic tree, a gradient along the dendrites of the local voltage change to synaptic input and the presence of NMDA receptors at the activated synapses. Since the NMDA receptors are voltage-sensitive, throughout each sequence they will be opened to a different extent depending on the location in the dendritic tree of the previously activated synapses in the sequence.

Aug 13, 2010
.. Hebbian plasticity... relies purely on the location of synapses on the dendritic tree..
it doesn't: Hebb's principle is a mechanism of alteration of the weights between neurons without relying to their actual structure at all.


Aug 13, 2010
You misunderstood me, I said that the study 'relies purely on the location of synapses on the dendritic tree...etc', NOT that Hebbian plasticity relies on this. Maybe, if you can access Science, all will become clear from reading the actual paper.

Aug 13, 2010
After then your comment has no meaning for me at all.

The finding of study presented in article is in good agreement with Hebb's theory, which is based on many previous experiments, though. Actually there exist some artificial models/prototypes based on this principle, already.


Aug 13, 2010

Spend a few years writing software on the CPU, then on the GPU. You'll get a very clear idea of why one is drastically better than the other depending on the problem you're working to solve.

Also consider that fundamentally real analysis and abstract visualization are entirely different ways to analyze information.

When someone says a system is very powerful it's meant to be interpreted within a specific scope, not globally.

One step further then that, the mental capacity to store information for some given period of time is distinct from processing information itself.

There are many subtle, yet distinct, lines between all the components that together encompass intellect and intelligence.

Aug 13, 2010
This seems like good news for Jeff Hawkins hypothesis in "On Intelligence".

Aug 13, 2010
The article doesn't cite the actual mechanism that allows a single neuron to distinguish between different temporal sequences of input. Hebbian plasticity doesn't address this issue because it is about how collections of neurons form networks, as in 'form follows function'.

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more