John-Dylan Haynes, director of the Berlin Center for Advanced Neuroimaging, is something like the chief mind reader among brain researchers.

He was one of the first to use machine learning algorithms to recognize patterns in magnetic resonance imaging (fMRI) images.

Thinking and decision-making processes should be better understood.

In their book he and co-author Matthias Eckoldt summarize around fifteen years of research in this field.

The subtitle promises a lot: "How our thoughts arise and how you can read them."

The idea is not new.

When the Jenenser professor Hans Berger developed electroencephalography (EEG) almost a hundred years ago, it was thought that one would soon be able to write letters to oneself in “cerebral writing”.

Then, in the 1950s and 1960s, great advances were made in the field of electrical brain stimulation.

During the Cold War, this went hand in hand with ideas about mind reading and mind control.

For some it was a curse, others saw it as a blessing, for example in the fight against crime.

The authors are now also playing with similar thoughts: Could the terrorist attacks of September 11th have been prevented with a mind control at the airport?

Restrictions are quickly forgotten

The majority of the book conveys scientific principles: How does fMRI work? What is pattern recognition? What are brain-computer interfaces? How do the relevant experiments work? In particular, Haynes' own studies are discussed. Numerous illustrations clarify the abstract concepts, comics loosen up the scientific matter. In this sense, the book is really successful and suitable for a broad readership.

Research must be limited to what is experimentally feasible and ethically justifiable. But Haynes quickly forgets these restrictions as soon as he interprets his data. This is also very clear in the new book. For example, the self-experiment of a television reporter gets its own chapter with the heading “Computers crack the thought code”. The woman brought ten pictures - of people, animals, sights and objects - which she looked at in the magnetic resonance tomograph. As is usual in such experiments, the algorithm learned in a first phase to differentiate between the measurement data. In the second phase, the images were viewed again, but the algorithm now had to determine what the subject saw. Conclusion:The computer achieved “a hit rate of one hundred percent in recognizing the journalist's thoughts”.

Myth formation on the subject of free will

Nothing was "cracked" here and no thought was read.

The significance - that the Brandenburg Gate has been seen - is given by the experimenter alone, not the computer.

The reporter could have seen, thought or done something completely different, as long as the data patterns in the first and second phases are sufficiently similar.

"Brainless brain research" called it the psychology professor Scott Lilienfeld, a critic of brain imaging research and its exaggerations.