Mind-reading competition

brain image interpretation.jpgDon’t worry, this isn’t about telepathy and doesn’t involve Uri Gellar.

No, it’s about a team of three Italian researchers who won $10000 in a brain-activity interpretation competition organised by the University of Pittsburgh earlier this year.

Entrants were provided with the fMRI data and behavioural reports recorded when four people watched two movies. The competitors’ task was to create an algorithm that could use the viewers ongoing brain activity to predict what they were thinking and feeling as the film unfolded. The crunch test came from a third film. This time the competing researchers were shown the viewers’ brain activity only, and they had to predict the behavioural data – what the viewers had reported seeing and feeling during the film on a moment-by-moment basis. The full rules are here.

The Italians – Emanuele Olivetti, Diego Sona, and Sriharsha Veeramachaneni were the most accurate, achieving a correlation of .86 for basic features, such as whether an instant of the film contained music. The full results are here.

I heard about this from the latest Nature Neuroscience editorial. They discuss the competition in the context of the increasing trend for researchers to see if they can predict what people are thinking based on their overall brain activity (this often gets discussed in relation to lie detection), rather than the more traditional correlational/localisation approach of seeing what brain activity occurs where, when people are thinking certain things.

The Nature Neurosci. editors welcome the shift:

Neuroimaging’s obsession with localization has often led to accusations that it is little more than phrenology. By using population responses across the whole brain to ask how rather than where information is processed, neuroimaging may be starting to come of age.

Link to the competition website.

Post a Comment

Required fields are marked *
*
*

Follow

Get every new post delivered to your Inbox.

Join 23,930 other followers