0
Would visual salience measures help in understanding the results?

To understand more about how the magician controls the viewer's gaze, it might be interesting to compute automatic visual salience measures on the video frames. Visual salience might explain some of the viewer's eye movement, particularly why they're looking in the 'wrong' place (from the perspective of trying to see the trick). This might also be an interesting dataset on which to explore both static and dynamic measures of visual saliency.

waiting for moderation
2 Answers
1
Accepted answer

Yes!! This is a great idea, and of central interest to our research. The fundamental reason that eye movements are critical is that we can only see well in the fovea of our retinas... which is about 0.1% of your visual field (about the size of your thumbnail at arms length). Everywhere else, you are legally blind. So our visual systems choose where we put our central gaze points very carefully, and your internal model of the world around you is built from your eyes only actually seeing just a few points in the world, then your brain builds a simulations that you call your conscious perception. So how it is that the brain targets eye movements is critically important to visual neuroscience.

Magicians, if you think about it, could be described as performers who control your eye movements. And they have an intuitive feeling for what draws your eye movements best. Visual neuroscience (and us, specifically) is also trying to solve this problem in parallel by doing exactly what you suggest. We currently have a paper in review in which we have looked at what people choose to look at in a scene, and how they utilize their oculomotor system to target those areas and to optimize acquisition of critical information at those "interesting" (and salient) points of the scene. This study will be coming out of Susana's lab once published so keep your eyes peeled on her website for its release: http://smc.neuralcorrelate.comm

waiting for moderation
0
Accepted answer

Nice idea, and definitively worth implementing in our upcoming magic research. One thing to keep in mind, though, is that, in addition to "bottom-up" salience determined by the low-level spatio-temporal features of the scene, the magician often has vise-grip control of "top-down" salience, through attentional misdirection, emotional manipulation, and even the implantation of false memories in the audience.

waiting for moderation
0

That's a very good point about top-down measures --- I was definitely thinking bottom-up. These videos are pretty "low key" compared to some of the magic shows I've seen, so I would guess that the bottom-up approach has a better than usual chance here. Maybe I will find out in the next paper!

-
waiting for moderation
0

Yes, the videos are pretty basic snapshots of what happens in an actual magic show, where there's a contextual narrative, story line, rapport with the public, etc. That's why the paper is entitled "Perceptual elements...". The hope is that we can understand the individual elements that integrate a magic trick before tackling the whole. So we're going for the low-hanging fruit first.

-
waiting for moderation

Ask me anything journal club

- On The Neuroscience of Illusion

Who: Dr. Stephen Macknik, Director of the Laboratory of Behavioral Neurophysiology at Barrow Neurological Institute, Dr. Susana Martinez-Conde, Director of the Laboratory of Visual Neuroscience, and Mr. Hector Rieiro, graduate student at the Barrow Neurological Institute.

What: Drs. Macknik and Martinez-Conde and Mr. Rieiro have studied a magic trick filmed in magician duo Penn & Teller’s theater in Las Vegas, to illuminate the neuroscience of illusion. Their results advance our understanding of how observers can be misdirected and will aid magicians as they work to improve their art.

Drs. Macknik and Martinez-Conde and Mr. Rieiro will be answering your questions live from 8am-10am PST, regarding their article, magic illusions, or any other topic of relevance to visual neuroscience.

When: December 17, 2013 08:00 am PST

Where: Ask me anything - On The Neuroscience of Illusion