Up until very recently, if you wanted to have your mind read, you had to visit a psychic or a mentalist who would, for a small fee, pretend to do so. These people could tell you things about yourself despite never having met you, convey messages to you from loved ones who’d passed on, or even discover where you’d hidden an item on stage just by touching you. The tools necessary to accomplish all this were surprisingly few—a crystal ball, a beautiful assistant, and a cynically wielded knowledge of human nature. But now there’s another option for those who want to have the contents of their mind exposed: You can go to the Gallant Lab in Berkeley and a team of scientists can read it for you … and all they need is an fMRI scanner, a couple of movie trailers, and eighteen million seconds of footage from YouTube.

***

MRIs See Minds in Motion 

Currently the best method for measuring human brain activity, functional magnetic resonance imaging (“fMRI”) doesn’t directly measure the activity of neurons, those cells that process and transmit information—rather, fMRI measures the changes in blood flow, blood oxygenation, and blood volume caused by neural activity. Because neural activity occurs much faster than the corresponding hemodynamic changes, previously it had been thought impossible to use fMRI to decode dynamic activity in the brain. But the team of researchers at the Gallant Lab devised a two stage encoding process that allowed them to model the relationship between the information recorded by the brain and the signals measured with fMRI.

First, a library of visual energy and motion filters is built, spanning a range of specific directions, speeds, and positions with each tied to specific neural activity. Second, each element of the library is processed through an equation that describes how neural activity then influences hemodynamic activity. Because fMRI measures that hemodynamic activity using volumetric pixels (or “voxels”) of 2.0 x 2.0 x 2.5 millimeters, each voxel represents the simultaneous activity of hundreds of thousands of neurons. Through regression analysis, the researchers create an encoder able to map the motion and energy of a visual action to the corresponding voxels measured by fMRI.

With this encoder in place, the researchers tested it by turning the tables and using it as a decoder. Shown a series of trailers for movies like The Pink Panther 2, Star Trek, and Madagascar, subjects in an fMRI scanner had their brain activity recorded. Those records were then processed back into a collection of visual actions. To test the accuracy of the data, a small video library comprising approximately eighteen million seconds of randomly selected YouTube footage was broken down and sorted into the same types of distinct motion directions, positions, and speeds used in the first stage of the encoding process. The hundred clips whose predicted activity was most similar to the observed brain activity were averaged into a “reconstruction” and then compared to the original movie trailer viewed by the subjects. The accuracy of the result, according to the Gallant Lab scientists, is far greater than mere chance and “demonstrate that dynamic brain activity measured under naturalistic conditions can be decoded using current fMRI technology.”

Although the aim of the experiment was to create a computational model of brain activity relating to moving images, possible future applications could include the development of a brain-machine interface for a neural prosthesis (which could work at the prompting of brain activity), or decoding the visual content of mental processes like dreaming, imagining, or remembering.

***

Even Frauds are Scientific About Their Trade

It all seems a far cry from the world of hucksters and con men, doesn’t it? In fact, scientific attempts to “read” subjects through different approaches, such as Gallant Lab’s breakthrough in neuroimaging, share more things in common with the psychic offering you advice from your deceased Uncle Pete than you might think.

After all, while the goal of science is to expand the greater body of knowledge with hypotheses proven through replicable testing, conmen and fakes need a replicable way to separate the money from a mark. There are three important considerations shared by scientists and charlatans alike.

1. Assemble The Library

Data is crucial. Just as researchers at Gallant had to create dictionaries of positioning and movement, Paul Ekman and Wallace Friesen developed the Facial Action Coding System to taxonomize human facial expressions. By breaking facial expressions into specific Action Units, Ekman and his collaborators created an index that paved the way for the science of microexpressions popularized in the TV show Lie to Me (based on Ekman’s work). Momentary expressions, unconscious actions on the face before a person has a chance to paste on an expression that allows them to lie, can be recognized and read—allowing the reader to see what’s really in a person’s head.

Similarly, Ricky Jay is widely considered to be one of the best performers of sleight-of-hand and misdirection not only because of his considerable physical dexterity but because he has accrued arguably the greatest collection of writings on stage magic and trickery ever assembled. As the magician Michael Weber explained in a New Yorker profile on the man, “There are two ways you can expand your knowledge [in magic]—through books and by gaining the confidence of fellow-magicians who will explain these things. Ricky to a large degree gets his information from books—old books—and then when he performs for magicians they want to know, ‘Where did that come from?’ And he’s appalled that they haven’t read this stuff.”

By gathering and studying data, scientists (and swindlers) can process related information more quickly, allowing them to make jumps at an intuitive level. As the theoretical physicist Henri Poincaré wrote, “It is through science that we prove, but through intuition that we discover.”

In an interview with David Frost, Orson Welles talked about the occupational hazard of phony psychics: becoming a “shut-eye,” which was the term for when a fake fortune teller began to believe they had actual psychic powers. Welles quit his brief career as a fake fortune teller after surprising both his client and himself by correctly announcing she had just lost her husband: “Undoubtedly, it’s not psychic,” Welles told Frost. “Undoubtedly, there was evidence of a tragedy. There were all kinds of things that went into the [mind] and got processed without me crookedly thinking what I’m going to say to her.”

2. Know the Value of Good Code

Converting a piece of information from one form to another is the essence of coding. And according to Gallant Labs, one of the keys to their experiment’s success lay in their construction of the encoding model that allowed them to predict brain activity response to “arbitrary novel movie inputs.” To a much lesser extent, other scientific attempts to read minds or mental states search for the way a person’s thoughts or feelings are encoded in unconscious movements, whether that’s the incomplete shrug or eyebrow twitch studied in microexpressions.

The phenomenally successful mind-reading act of the early 20^th century, John and Eva Fay, also knew the value of good code: Like many tag-team mentalist performers, the Fays constructed a cipher made up of ordinary words and phrases that, when used properly, allowed the assistant to tell the blindfolded psychic what they held in their hand, where an audience participant was from, or any number of arcane details, all under the disguise of asking a question. The psychic then announced these details to the astonishment and delight of the audience.

3. Let the Subject Do The Work

One potentially controversial element to the Gallant Labs experiment is that all three of the subjects were co-authors of the paper. In order to build the proper energy motion encoding model for each subject, it was necessary they spend several hours in an MRI watching footage. In order for the encoding and decoding processes to be as accurate as possible, the subjects had to remain alert and focused for long periods of time, chores best done by those fully invested in the research. Because the experiment focused on the early part of the visual system which is mostly free of influence from intention or prior knowledge, Gallant Labs insists it is highly unlikely any cheating or bias skewed the results of their work.

But the best judge of that will be how easily other researchers can recreate these results using different subjects: A difficulty in doing so has injured the reputation of both microexpressions (after receiving such criticism, Paul Ekman no longer publishes his work in peer-reviewed journals, claiming that the work could be replicated by countries the United States views as a potential threat).

By contrast, “cold reading” counts on exactly the sort of subject bias true scientific experiments try to avoid: The cornerstone of pseudo-psychic mind-reading, cold reading is the art of using statements applicable to a wide range of people as if they were personally and specifically directed to the swindler’s mark. Sometimes called Barnum statements (because, as P.T. Barnum claimed about his shows, “they have something for everyone”), they elicit the listener’s trust in the charlatan’s “psychic” ability because the listener doesn’t realize the extent to which the statements are true for everyone. “You say, ‘between the ages of thirteen and fifteen, you had a great change come into your life,’” Welles told David Frost. “Well, that happens in everybody’s life. ‘You’ve got a scar on your knee.’ Everybody fell down and has a scar on their knee!”

With the subject’s trust in place, the psychic uses the conceit of a varying psychic connection that allows them to “hear” several spirits (some of whom the psychic claims are confused or reluctant to speak) to make guesses about their mark—for example, asking if the listener knows someone deceased whose name begins with an “M”—and refining or dismissing them based on the mark’s responses. (If the psychic correctly guesses the name of the deceased person is “Michael,” the mark believes the psychic can really communicate with the spirits. If the psychic guesses wrong, they can say they misunderstood the spirit, and the spirit is actually concerned about someone named Michael—does the subject know who they might mean?) With the use of cold reading and Barnum statements, most marks will believe psychics have told them a tremendous amount of information, when in fact it is the mark who has actually told the psychic.

Because of the way human consciousness works, we believe our knowledge of the world to mostly be based on observed facts, instead of biases and inferences that influence the “facts” we record. As scientists try to read the human mind, they must work doubly hard to ensure their results show what the subject sees, not what they themselves want to see. To stand Poincaré’s statement on its head: “Through intuition, we discover—but it is through science that we prove.”