“Applying Insights from Magic to Improve Deception in Research: The Swiss Cheese Model”, 2020-11-10 (; similar):
Researchers generally receive little training in experimental deception.
Drawing on the field of magic, we present a novel model of effective deception.
First, deception should have many “layers” rather than a single cover story.
Second, these layers should be subtle rather than explicitly stated.
We provide strategies for improving deception and thus the reliability of research.
Social psychologists, placebo scientists, and consumer researchers often require deception in their studies, yet they receive little training on how to deceive effectively. Ineffective deception, however, can lead to suspicion and compromise the validity of research. The field of magic offers a potential solution; magicians have deceived audiences for millennia using a variety of robust techniques.
As former professional magicians, we propose the Swiss cheese model of deception and argue that deception should be subtle yet elaborate. Subtle deception involves techniques such as fake mistakes, planted assumptions, and convincers. Elaborate deception involves layering many of these techniques rather than relying on a single cover story.
We have demonstrated the potency of these principles by making participants believe implausible ideas, such as that a machine is controlling their mind or that the placebo they consumed was a psychedelic drug.
These principles can help researchers reduce demand characteristics, improve blinding, and increase the generalisability of studies that require deception.
[Keywords: deception, suspicion, magic, placebo, blinding, ethics]
1.1: Deceive elaborately with many layers: Co-author A. R. used to perform an act in which he would appear to read the mind of an audience member. The secret was simply that the audience member he selected for the demonstration was a paid confederate; the apparently impromptu mind reading was actually a scripted exchange. In the middle of one show, a man in the theatre stood up and shouted, “I was here last week and he chose the same woman. She’s a stooge!” After some commotion and hesitation, the magician invited the heckler onto the stage and then proceeded to read his mind instead. The act was powerful for the audience and particularly so for the initial confederate. The magician later “confided” to her that he could indeed genuinely read minds, but it was cognitively taxing for him, which is why he hired her as a confederate. The confederate was so impressed that she praised his magical powers in front of friends and colleagues for years after the performance. As it turns out, the heckler was the magician’s uncle—yet another confederate.
This additional layer of deception was intended to fool the audience as well as the initial confederate.
[Because who would expect two layers? ‘Magic’ is doing more work than any reasonable person would expect…]
Magicians often use such elaborate forms of deception ( et al 2014; 2012). Audiences may suspect stooges in a magic show, but they are less likely to suspect one stooge to cover up another. In other cases, magicians may show up at a restaurant hours before a performance to stick playing cards under each of the tables, one of which will be used in a casual magic trick over dinner. Or, the spouse of a magician may pretend to not understand English in order to discreetly eavesdrop and signal information undetected from the audience.
Such elaborate acts, requiring considerable time, money, or effort, can be difficult for lay audiences to imagine and are thus particularly deceptive (2012).
In research, deception is often confined to a few layers, such as a bogus device or a false explanation of what a task is measuring ( et al 1995), though adding more layers may increase the effectiveness of the deception. In one study ( et al 2016), we had to convince educated participants that a (sham) MRI scanner could both read their mind and insert thoughts into their head; we were testing whether the delusion of thought insertion could be reproduced in a non-clinical population. To do so, we used various layers to strengthen the deception. The first 30 min of the protocol included fake MRI safety screenings, a lab technician (surrounded by scientific paraphernalia) describing the complex workings of the machine, and a sham calibration procedure. As in magic, such deception can lead participants down one explanatory path (eg. that a novel technology will control their mind), making them less likely to discover the underlying “secret” (2016). These many layers constitute costly signaling: the effort involved in the procedure was specifically intended to make participants less likely to infer that it was all a sham (2018). In a replication, removing one of the key layers of deception made the procedure less convincing (Pailhès et al in progress). Related studies of machine mind reading and thought insertion that used fewer layers of deception have also resulted in higher rates of suspicion or somewhat weaker effects ( et al 2014; 2013).
Elaborate deceptive methods are occasionally required in placebo research. In a study applying the Swiss cheese model, we used a dozen researchers in lab coats, a security guard, a handful of confederates, sham blood pressure feedback, and fake drug information sheets to convince participants that the placebos they consumed were actually psychedelic drugs ( et al 2020). Accordingly, some of the participants reported alterations in consciousness similar to what one would expect from a moderate dose of the actual drug. In a study of placebo alcohol, et al 2016 also used various layers of deception: confederates made off-hand comments about friends who got drunk while previously completing the study, the researchers sprayed the room with an alcohol scent, and the (non-alcoholic) drinks had real alcohol rubbed along the rim for subtle taste cues.
…When guessing 3 people’s chosen playing cards, they [mentalists] will intentionally get the last one slightly wrong (eg. guessing the Seven of Diamonds rather than the Seven of Hearts) to make the situation appear more plausible and lead people to believe it is telepathy rather than a trick (1983, Intimate Power). This trickery is effective because it is more difficult for audiences to imagine that such seemingly costly mistakes would be carefully planned to improve the show (2018).