About ten years ago, a method today considered a classic, named functional nuclear spin to examine the brain, caused a stir. Back then, neuroscientists succeeded to recognize whether their test persons were looking at a house, a face or a chair at that moment by activation patterns in the sectional images of magnetic resonance imaging. For the first time, the following publication of the tests made the topic ‘mind-reading by MRI’ open to a larger public.
Glance into a virginal brain
Nonetheless it was not the great break through back then: The whole thing only worked because the scientists had taken a good look at how each and every test person’s brain reacted exactly to the various pictures. It was recognition of patterns with prior announcement which is not exactly what would pass as real mind reading in spiritual circles. But now there is a new work getting a whole lot closer to real mind reading. The psychologist Russell Poldrack at the Institute for Behavioral Neuroscience, the department for psychiatry of University of California in Los Angeles/USA, and a team of scientists are expected to report in the October 2009 edition of the journal Behavioral Neuroscience about a test series with 130 test participants. The scientists made an fMRI-scan of the brain while the participants were working on one of eight ‘brain-teasers’, including reading words out loud, recognizing rhymes, counting of tunes, pushing buttons on certain sign stimuli and making financial decisions. The scientists analyzed the fMRI-scans of 129 test persons. They developed an algorithm for predicting which of the eight tasks were solved by brain no. 130.
About six times better than coincidence
The whole procedure was done for each of the 130 participants. Different from other examinations here the algorithm was applied on a ‘virginal brain’, a brain where no information was available while formulating the algorithm. The results are astonishing. Statistically eight tasks result in a random hit ratio of just 13 percent. Poldrack and his colleagues though achieved 80 percent with their algorithm. In other words: In four out of five brains the algorithm provides the true answer to the ‘what are you right now thinking about-question’. “For sure it’s not perfect, but it’s still pretty good”, says Poldrack. “With functional MRI we can say quite a bit about what a person is thinking at that moment although we never saw his or her brain before.” Among others, one of the fascinating things about the California data is the realization that human brains are very similar. “Quite often we are concentrating on the alleged differences between single brains. But our study shows that the brains of most people function quite similarly – otherwise it wouldn’t work”, emphasizes Poldrack.
The more chaotic you think, the harder it gets
In many respects reassuring though is the fact that – despite all mathematics – they did not succeed 100 percent. “We are not nearly good enough with it to for example get admitted to court”, says Poldrack. He also points out that the 80 percent success rate is only possible if the number of potential solutions is limited to eight exactly defined thoughts. “If with eight very different mental processes we succeed only by 80 percent we definitely are very far away just yet to figure out what a person is thinking if millions and millions of thoughts are involved.”
The limits of the Californian attempt of mind reading are also obvious at a completely different point. The mathematics of the used algorithm are based on originates of the world of machine learning. It is a very similar technology like those used by internet portals such as for example Amazon to make shopping profiles of customers which serve as base for targeted product suggestions. If you observe these things you find out that the efficiency of such functions leaves a lot to be desired. Somehow comforting, isn’t it?!