There is a fascinating review in Nature Reviews Neuroscience this month about the cognitive science of magic tricks -- authored by both scientists and practicing magicians (sadly behind a subscription wall). The article attempts to list and describe in neuroscientific terms the techniques that magicians use to trick their audiences. The authors break down these into "visual illusions (after-images), optical illusions ('smoke and mirrors'), cognitive illusions (inattentional blindness), special effects (explosions, fake gunshots, et cetera), and secret devices and mechanical artifacts (gimmicks)." The use of visual illusions to study perception is certainly nothing new, but the emphasis on cognitive illusions -- illusions that trick higher order perceptions like attention and judgment -- is novel.

There is some interesting stuff particularly on inattentional blindness. Inattentional blindness is when you don't see something right in front of you because you aren't paying attention. The categorical example of this effect is shown in the YouTube video below:

Did you see the moonwalking bear? Most people don't the first time because they are too busy counting the passes. (Video from here.) (I am pretty sure that wasn't the original video that Simon and Chabris used when they first performed this experiment, but I think this one is funnier.)

I was struck by this passage that discusses the relative weight of different types of illusions in manipulating the audiences perception. A few studies exist that have compared the effects of attention with other visual illusions. They found that the inattentional blindness has a much larger effect than the misdirection of looking in the wrong place:

Kuhn and Tatler measured the eye movements of observers during the presentation of a magic trick (a magician made a cigarette 'disappear' by dropping it below the table). To our knowledge, this is the first study to have correlated the perception of magic with any physiological measurement. The goal of the experiment was to analyse the scan paths of subjects to determine whether observers missed the trick because they did not look at it at the right time or because they did not attend to it (irrespective of the position of their gaze). The results showed that the detection (or not) of the cigarette drop could not be explained at the level of the retina. That is, detection rates were not significantly influenced by blinks, saccadic movements or how far the cigarette was from the centre of vision at the time of the drop. The authors concluded that the magician primarily manipulates the spectators' attention rather than their gaze, using similar principles to those that are used in inattentional-blindness studies. Thus, to overcome the magician's misdirection, spectators should reallocate their attention -- rather than their gaze -- to the concealed event (that is, the cigarette drop) at the critical time. (Emphasis mine. Citations removed.)

Interestingly enough, while the conscious perception is tricked by inattention the oculomotor system that moves the eyes is not.

A recent study of the Vanishing-Ball Illusion further supports the conclusion that the manipulation of gaze position is not critical for effective covert misdirection. In the Vanishing-Ball Illusion, a ball thrown by the magician vanishes mid-flight. To achieve this effect, the magician begins by tossing the ball straight up in the air and catching it several times without event; then, on the final toss, the magician only pretends to throw the ball. The ball is in reality hidden in the magician's hand, but most spectators perceive it ascending and then vanishing mid-flight. During the execution of this trick, the magician's head and eyes follow the trajectory of an imaginary ball being thrown upwards. Kuhn and Land found that the magician's use of such social cues was critical for making the spectators' perceive the illusion (that is, the ball vanishing mid-flight). However, observers did not direct their gaze to the area in which they claimed to have seen the ball vanish, suggesting that the oculomotor system is not fooled by the illusion. Instead, the illusory effect is presumably caused by covert redirection of the attentional spotlight to the predicted position of the ball. This result is consistent with previous studies that suggested that there are separate mechanisms for perception and visuomotor control. For instance, the eye movements of blindsight patients are biased towards stimuli that the patients do not consciously perceive. Kuhn and Land further proposed that in the Vanishing-Ball Illusion the covert redirection of the attentional spotlight to the predicted position of the ball might be related to "representational momentum." (Emphasis mine. Citation removed.)

I find experiments like this one with the Vanishing-Ball illusion so fascinating because they allow us to dissociated the conscious perception of the object from the operation of the oculomotor system. Read the whole thing.

Anyway, the authors core argument is that the types of illusions created by magicians form valuable tools for studying cognition and perception. I couldn't agree more. If you have access to the article, there is also a great video of a magician tricking philosopher Daniel Dennett.