The magic of math

Decoding stimulus class and onset time from a continuous data stream in single trials: Illustration of two electrodes and the continuous classifier using 2 broadband features. Credit: University of Washington. PLOS Computational Biology

It is now hundred and forty years since the discovery that a brain generated electrical signals (1875, a rabbit brain...) and 90 years since the first recording of a human brain electrical activities. 
In these years many progresses have been made in the understanding of the biological machinery producing those signals, their location and general meaning.  Pinpointing specific meaning is still very difficult. We have seen progress in these last ten years in correlating specific signals in the brain motor area to the movement of a limb and these are now being used to control prosthetic limbs with growing precision.

Progress has been made also in decoding perception of visual stimuli, although so far it has been more kind of ON-OFF, like seeing an image or not seeing it.

The understanding of what a person is actually thinking is still somewhere in the future (and, as I have observed in some previous posts on this topic, once we will get there we will open a can of worm.... that will be very difficult to manage).

A paper published by Washington University researchers is bringing us a bit closer in understanding how certain visual stimuli are processed and managed in the brain.

They have found a way to use math to interpret the electrical signals captured by electrodes inserted in the temporal lobes of epilepsy patients (the electrodes were inserted for a week in each patient to determine the area of origin of incontrollable epilepsy seizures). 

For the first time the study has shown that it would be possible, in real time, to analyse the electrical patterns and decode what the person is actually seeing. In the experiment the patients were presented with images of houses or faces, each image flashed for just 400ms followed by a grey screen.

The mathematical analyses was able to correctly interpret the signals with a 95% accuracy and with a delay of just 20ms. In previous works researchers were able to detect the processing of visual stimuli but not in real time.

Clearly, distinguishing between the processing of a face or a house image is a long shot from being able to tell what the person has seen (one should be able to differentiate among millions of different objects, and in most cases one is seeing several objects at the same time). 

One could just laugh at this achievement when compared to the normal "complexity" of what is going on in a brain. 

However if one takes a look at the representation of the electrical activity monitored it seems impossible that out of that mess one could detect anything significant. Besides, we have to recognise that there is a continuous progress in our capability of looking at brain electrical activity and making sense out of it.

Reading the mind will not happen anytime soon, but it will happen.

Author - Roberto Saracco

© 2010-2018 EIT Digital IVZW. All rights reserved. Legal notice