SciBar #23: Talking Mathematical Neuroscience in The Vat and Fiddle

For April's event, SciBar presents Professor Stephen Coombes from the University of Nottingham to talk about "Mathematical Neuroscience - A New Discipline for Understanding the Brain". For years mathematics has been known for its use in fields such as physics and engineering but more recently, it has become a tool for biology too, especially neuroscience - the study of the cells in the brain.

There are 10,000 million neurons in your head and those cells that are in your cerebellum want to fill space, just like a tree. The branch structure that they create allows them to grab inputs from other cells in the brain. We have known about these structures for over 100 years and have been able to make line drawings of them based on observations in a microscope. But what do these cells actually do? We can track electrical activity in the brain but modern technology such as x-ray, CT, MRI and fMRI allow us to track individual cells and the whole brain all of which are less evasive than firing electrons into the brain.

In the fifties, Hodgkin and Huxley described the mathematical model of the electrical impulses in the brain. This was proved to be a good model through testing on squids. Without computers they had to evaluate the model using a mechanical calculator. Around the same time EEG technology started to improve allowing investigation into large scale brain rhythms using recording electrodes stuck to the outside of the head. This method can detect brain waves over time such as the Alpha wave of around 8-10 cycles per second which is produced when the brain is at rest. These rhythms are created by the interaction between "excite" and "inhibit" cells in the brain.

In the seventies, neural mass models were developed. These models predicted how epileptic brains would behave under EEG observation, this meant that there was now a model that could be tested instead of experimenting on epilepsy sufferers. Functional connectivity structures between those excite and inhibit cells help differentiate between normal brains and people with issues.

By the eighties, neural field models were developed, these are whole brain equations but had some slight drawbacks as they model the brain as spherical. This was also a time when the idea of neural networks started to gain traction. These artificial networks could be trained to solve real world problems. Unfortunately, they couldn't solve all problems and "neural networks" became a dirty phrase.

These days we have fMRI machines where we can actually look at oxygen use inside the brain. Deep brain stimulation has also been developed to help neurons fire in the right order. This can help with Parkinson's, where all the neurons want to fire at the same time. The treatment was discovered by accident but now around 100,000 people have an electrode in their brain to help with their Parkinson's. It has also been used to help with other conditions such as depression.

The next development is the discipline of neural engineering - steering away from maths towards engineering. The big things coming out of this will be brain computer interfaces which would help to control prosthetic limbs. Also in the works is the human brain project - 0.5 billion Euros have been put aside to build a computer model of the brain. It will be a biological neural network, although if it doesn't work it could lead to a similar backlash to that against neural networks in the eighties.

SciBar returns to The Vat and Fiddle on Wednesday 25 May at 7:30pm where Dr Michael Loughlin from Nottingham Trent University will be talking about antibiotic resistance. 

Nottingham SciBar website