(The following article was originally published in Swedish in the Summer 2023 edition of Forskning & Framsteg magazine)
Magic and science might seem like an unlikely pairing, but scientific interest in magic illusions can actually be traced back to the earliest days of experimental psychology, when natural scientists sought to investigate claims about apparently miraculous spiritual phenomena. In fact, the earliest studies of magic involved researchers empirically demonstrating that healthy honest observers could be fooled into mistaking illusions for ‘genuine’ supernatural phenomena. Today, the so-called ‘Science of Magic’ has arguably been undergoing a renaissance, with researchers around the world increasingly using tools developed for magic performances to study human (and in some cases animal) cognition. Magic trick methods have been demonstrably effective tools for overtly and covertly manipulating participants’ experiences in laboratory settings. I myself have used it to build and use fake mind-reading machines together with colleagues at McGill University in Canada. I have an education in experimental psychology and neuroscience, but have also worked for many years as a semi-professional magician .
Pick a card!
One the best things about researching magic is it lends itself excellently to practical demonstrations! Take a look at the above image of 6 different play cards. Select any card, and concentrate on your selection. Try and picture your card’s image as vividly as possible in your mind. It might help to close your eyes for a moment. Believe it or not, I’m going to eliminate the card that you are thinking of. Scroll to the bottom of this article and, you will see below that there are now only five cards left, and your card is gone! I’ve apparently read your mind and removed the card that you were merely thinking of!
Beyond illustrating how magic tricks can exploit counterintuitive psychological principles to induce seemingly ‘impossible experiences’, this trick can also exemplify how differences in framing and presentation can create diverse illusory experiences. The original version trick involved physical cards, and it could be presented as a feat of dexterity. The performer might falsely claim to have snatched the target card with speed that was ‘quicker than the eye.’
But the trick can also be presented as an act of mind reading or even mind influencing. Perhaps the magician divined your thought-of-card by sensing your unspoken thoughts. Or maybe the magician manipulated your free will and influenced you to select the card of their choice?
Failing to See (and failing to think that we cannot see)
This particular trick has even been used by contemporary psychiatrists as a quick and reliable way of inducing a non-distressing anomalous experience of ‘thought interference’ in both clinical and non-clinical populations. Of course, nothing ‘magical’ actually occurred. This trick operates on a principle that psychologists now call ‘change blindness.’ Change blindness occurs when people fail to detect changes in visual scenes when these changes are accompanied by a visual disruption. In this case, I’m able to remove the card that you were ‘merely thinking of’ because I actually remove all the cards! This can feel magical because most of us don’t believe that they could possibly fail to notice such a dramatic switch. Magicians had been using this principle to perform tricks since at least the early 1900’s. The basic idea behind this trick was developed and by a magician named E.A. Parsons, who performed under the name Henry Hardin (1849- 1929). It’s worth noting that change blindness was only formally recognized by the scientific community in the 1990’s, when a team of researchers led by Ron Rensink developed a simple visual paradigm to demonstrate that healthy observers have a great deal of difficulty identifying sometimes dramatic changes to photographs. Ironically, the Rensink and his team struggled to publish their findings because other visual scientists who peer-reviewed their work initially believed that it was impossible that viewers would take so long to detect the changes!
Change Blindness Demo: This gif an approximation of the stimulus images from used in Rensink, O'regan, & Clark's (1997) ‘Flicker Paradigm’ . There are actually three images- a photo, a black screen, and modified photo. How many differences can you detect between the two photos?
Magical Seeming Innovations
Today, change blindness is an established field of research. In combination with a bit of sleight-of-hand tricks, it has actually led to the discovery of so-called choice blindness , and to the founding of the research team where I’m currently employed. When new technological innovations emerge, it people perceive them as almost magical. Historical examples include wireless telegraphy and television. At the moment, artificial intelligence has taken on a similar role. The science fiction writer Arthur C. Clarke once noted that “any sufficiently advanced technology is indistinguishable from magic.” The premise of our research is that the reverse can also be true: Magic tricks and illusions can be made to be indistinguishable from advanced technology.
Partially thanks to recent developments in AI and machine learning, genuine neurotechnology is advancing at rapid pace. Machines are becoming increasingly capable of decoding human brain activity. While impressive, such procedures remain relatively limited: Information decoded from brain activity is often rudimentary and requires cooperation from participants. Furthermore, the designing and operating such machines still requires highly expensive equipment and a high level of scientific expertise. But nonetheless, brain reading technology has the potential to not only become more powerful, but also more commonplace. Many reasonable people agree that we should start taking steps to prepare for this potential future, and it would be ideal to if we could anticipate how people will react to such developments. But this presents a problem. We’d be asking people imagine their own future reactions to speculative, currently science-fictional, technology. Past psychological research has demonstrated that can people struggle to predict even their own feelings and behavioral response to future events.
Using Mentalism Magic to Simulate Neurotechnology
The fake set-up at McGill University that utilised a dummy fMRI scanner to help create the illusion of advanced neuroscientific technology (Photo by Madalina Prostean)
Myself and my colleagues in the Choice Blindness Lab, along with our collaborators at McGill University, have developed a novel solution to this methodological problem: We’ve created an experimental paradigm where we use magic illusions to trick participants into thinking that they are interacting with real technology. To ‘prove’ the machines are effective, we borrowed methods from a particular branch of performance magic known as ‘mentalism.’ Mentalism is a genre of magic that involves creating the appearance of psychic or supernormal powers, like reading someone’s mind or predicting the future.
Such methods have traditionally been used for entertainment purposes by magicians who openly acknowledge that they are using trickery. Historically, they have also been used to by fraudulent mystics looking to create the impression that they have genuine supernatural abilities.
Participants sign up to our studies believing that they are taking part in research involving cutting edge experimental neurotechnological systems. We always reveal the illusory nature of the machines to the participants. But we don’t tell them until after the experiment has concluded. So even though the technology itself is illusory, when participants believe that it’s real, we can then effectively measure and record genuine behavioral responses. Paradoxically, we need to deceive our participants to ensure that we receive honest responses.
We recently published a paper describing how we led 59 participants to believe that (fake) neurotechnological machines could infer their unspoken personal preferences even and detect their deep-seated attitudes. Participants in our experiments were generally very accepting of our illusory machines- no one expressed any overt suspicion about the veracity of the machines nor the ‘evidence’ of their efficacy. We had participants complete traditional paper/pencil style surveys about their attitudes, and we then compared these survey responses with supposed outputs produced by our mind reading machines. Throughout the experiment, we used various magic trick methods to secretly peek at participants’ written responses so that we could manually adjust our ‘outputs’ match them at other times we would use slight-of-hand sneakily write the outputs after participants had explicitly revealed them. For the participants, this created a compelling illusion that our technology was able to ‘read their brains’ with uncanny precision.
A simulated version of our simulated set-up at Lund University, the EEG cap plugs into nothing and the apparent ‘biometric facial analysis’ that participants saw is actually a Snap filter (Photo by Peter Westrup)
Genuine reactions to fake results
In a way, ‘performing’ the tricks under such conditions was actually easier than doing so in the context of a magic show. When people go to see a show, they know that a magician is going to be try to misdirect them, but our participants initially had no reason to suspect that we were using trickery. And because we were cheating, we were always able to make it seem as if the machine’s outputs corresponded with participants’ attitudes.
At points of the procedure, we deliberately presented participants with readings that we knew contradicted their survey responses. This is perhaps one of the more fascinating aspects of the study: When confronted with fictional outputs from the fake machines that contradicted their own stated preferences, most participants accepted that the machine was more accurate at determining their attitudes then they were! One participant rationalized that the machine had accurately detected what they felt “in [their] heart.” In effect, our participants seemed to be willing defer to a sham neurological measurement even when it directly contradicted their own introspection. Overall, I consider this experiment to be a promising proof of concept for our method of convincingly simulating advanced technologies using magic methods. We’re now running a series of follow-up experiments to further explore the paradigm’s potential.
I hope the insights gleaned from our methods can help stimulate discussion about the impacts emerging technologies. Allowing people to experience the effects of technology can provide valuable information about people’s reactions to prospective technologies before such technologies are feasible may be of use to ethicists and policy makers who are grappling with the potential societal implications. And we’re also planning on more closely exploring psychological the mechanisms that lead participants to believe our deceptions. This knowledge could help us develop new tools to promote better critical thinking and healthy skepticism.
Haunted Histories and the Science of Magic
Richard Hodgson & ‘Friend’- An example of a fraudulent spirit photograph featured in Hereward Carrington’s (1930) The Story of Psychic Science
While I’m excited about the future directions of our paradigm, I’m also immensely pleased with the way our study fits into the history of the magic and psychology. The idea of using discretely magic methods to simulate extraordinary experiences arguably dates back to one of the first ever empirical studies of magic. The study in question, titled, ‘The possibilities of mal-observation and lapse of memory from a practical point of view’, was originally published in 1887 in the Proceedings of the Society for Psychical Research. It wasn’t conducted by experimental psychologists, but rather by a pair of ‘psychical’ researchers: Richard Hodgson and S.J. Davey. They were investigating the reliability of eyewitness testimony as it related to alleged paranormal phenomena. Specifically, they were interested in the reports of seances that were being conducted by contemporary spiritualist mediums. Proponents of spiritualism, who included eminent scientists such as William Crookes, Oliver Lodge, and even Alfred Russell Wallace (who co-developed the theory of natural selection along with Charles Darwin) argued that witness testimonies of miraculous seeming events should be considered powerful evidence in favor of the genuineness of mediums. For example, one of the popular séance phenomeon at the time was ‘spirit slate writing.’ Mediums would use small chalkboards that were common in school classrooms at the time, and written messages would seem to appear on on the boards even when it seemed that no living person could have possibly produced the writing. On the other hand, skeptics argued that witnesses might be being deceived by elaborate illusions and tricks that were akin to methods used by performing magicians.
To empirically investigate this question, Hodgson and Davey conducted a series of hoax seances using magic trick methods. Participants in the study were invited to spend an evening with a ‘medium,’ Davey, and observe his abilities. Davey used slight-of-hand techniques and misdirection to secretly write on the slates himself (or to sneakily swap blank slates for pre-written ones), and then pretended that the writing had come from spirits. [PS4] The participants weren’t explicitly told that they were seeing a genuine display of mediumship, nor were they informed that they were seeing a magic show. Unlike most professional mediumistic demonstrations or magic shows, participants weren’t financially charged. However, their ‘price of admission’ was that they were asked to write letters to Hodgson detailing the events of the evening in as much detail as they could remember. Hodgson and Davey collected 27 distinct accounts from 17 separate performances. Because the ‘séance’ was scripted and choreographed, the researchers were able to directly assess the accuracy of the witness statements compared to actual events.
The results were dramatic and controversial. Hodgson and Davey wrote that witnesses failed to notice (or remember) critical aspects of the performances. For example, many witnesses forgot or failed to notice moments when Davey himself was handling the slates- they testified that the slates had never left their sight, when, in reality, Davey had not only held them, but had managed to substitute entirely different ones. So even if someone knew how the tricks were accomplished, there was no way to reconstruct the genuine methods based on the witness testimonies- the stories that participants told were genuniely impossible. Furthermore, many participants also confidently reported seeing events that never actually happened. Some participants’ testified that they had specifically examined the slates themselves and even cleaned them to ensure that they were blank, when in actual fact they had not been allowed to touch the slates, which weren’t blank, and had writing on them all along. One participant was amazed that the spirits had somehow ‘known’ his secret childhood nickname “Boorzu” when he’d actually been presented with the sloppily written word “book.” The results demonstrated that magic methods lead cause healthy honest observers to genuinely experience extraordinary phenomena, even when these phenomena are produced by trickery.
Not only did this experiment represent a revolutionary way of integrating magic methods into a study of perception and memory, but the results have been borne out by subsequent cognitive psychology experiments into phenomena like inattentional blindness and reconstructive memory.
In developing our fake mind reading machines, we used magic methods that were directly derived from the methods used by Davey to fraudulently produce ‘spirit writing.’ Today, we’re effectively operating as mediums. Except, instead attributing our powers to spiritual entities, we pretend our powers are derived from a fictional neurotechnological system. The exact same magic trick methods that deceived Victorian-era observers are, with the proper framing, just as viable today.
I don’t think this necessarily means that people are gullible or unobservant. Rather, it illustrates that our minds are often much weirder than we imagine them to be. This discrepancy between how people think their minds work and how they actually work can be exploited for entertainment, for fraud, or for empirically exploring the behavioral impacts of speculative technology.
