Decoding and predicting decisions from brain signals

John-Dylan Haynes
Bernstein Center for Computational Neuroscience, Charité Berlin

The recent emergence of pattern classification in neuroimaging has made it possible to study the 
representation of specific cognitive variables in the human brain. In a number of studies we have used 
classification techniques to unravel the encoding of decisions in the human brain. In one line of work we 
studied free choices for specific action plans. We found that it was possible to decode which specific 
intention a person was covertly holding. In another group of experiments it was possible (at least to 
some degree) to decode and predict the free choices for actions even before people believed to be 
making up their mind. Such a predictability of seemingly free choices raises the question whether certain 
beliefs regarding "free will" are scientifically warranted. Finally, we performed several studies on 
decoding of choices in more realistic scenarios. It was possible to decode preferences for objects 
(consumer products/cars and politicians/parties). In these cases decoding was based on implicit brain 
signals while people were distracted from thinking about the objects with an attention-demanding 
secondary task. Taken together, our results unravel how the brain processes decisions outside of 
conscious awareness and attention.