Understanding the origins of thought: big data on brain images
Specialised in the analysis of brain images, Gaël Varoquaux is a member of the Parietal team at the Inria Saclay – Île-de-France research centre. On 29 September 2016, he was the guest at a "Unithé ou café" meeting organised by Saclay centre.
Gaël Varoquaux is looking to understand cognitive processes, for example those that occur in face recognition. His research combines the neurosciences, cognitive psychology, computer science and mathematics. The researcher uses both brain activity images - obtained by MRI in particular - and big data processing tools. "Computer science and statistical learning enable us to extract bilateral links between brain activity and certain representations of cognition. My aim is therefore to associate the perception of the concepts and signals emitted by the human brain."
PARIETAL team has analysed the brain activity of subjects looking at photos and watching videos. Based on image recognition tools, convolutional networks, they draw characteristics from the photos and demonstrate that brain activity is indeed explained by them. As a result, it is possible to draw a parallel between the cognitive mechanisms used by the brain in order to understand images and convolutional networks. Moreover, equipped in this way with a good description of what is contained within the stimuli, it is possible to understand what, within brain activity, predicts to which stimuli the brain is reacting. These predictions work for simple tasks such as, for example, the vision of a face. The aim of Gaël Varoquaux and his team is to devise statistical tools in order to produce predictions during much more complex cognitive experiments.
Click here to see the video of the presentation.