It’s a rolling shutter effect of a CCD / CMOS image sensor being polled sequentially by the controller.

Though I am intrigued by ian_b’s explanation.

http://www.globalnerdy.com/2009/01/13/cheap-camera-interesting-shot/

The reflected image is a slight degree off the line needed for a perfect mirror (the top of the cabinet seems to be angling slightly down and away), so you’re getting a slight bit more of a view from under the eyes (perhaps more lashes visible?). Then, the overall reflected image is slightly darker, and the color registration is different, making darker areas look larger (or more shadowed). Note how the light in the reflected window is more blue than the light from the window itself. It may be the blue wash that seems to saturate the reflection at some level.

Photoshop.

Note how the secondary reflections (the reflection in the monitor of the reflection in the stove) are not quite positioned the same. That is physically impossible. An image of an image has to be located at the corresponding point.

The 666 on his head is a little trickier to see, however.

It would be interesting to see the EXIF data. The complete absence of IPTC and EXIF data is either a normal BB procedure, or done deliberately by the submitter.

The sky through the window is blown out, but the tree is visible, the tree is mostly in focus, the interior of the house is in focus from the closest buttons on the remote to the window. The photographer was using a small aperture. I don’t see a lot of noise in the shot either indicating a lower ISO number was used (there is apparent noise and noise-reduction in the deeper midtones).

The average duration of a blink is 300 to 400 milliseconds, so a 1 second exposure give the subject enough time to blink, given that the subject and his reflection are at opposite sides of the image. Given 1 second, there is roughly half the frame between the two sets of eyes. That’s 500 milliseconds or less, depending on the shutter gap.

On the other hand, the shutter speed might have been faster than one second, with the effect developed through practice. There is some movement with the mouth in the right half of the frame. I would expect blurring and more subject motion with slower shutter speeds. There are some subtle differences in posture in the two images of the subject, indicating some time has passed.

The small aperture and portrait format leads me to think this photo was staged to demonstrate this effect.

In any case, a shutter isn’t all or nothing, it’s almost always a mechanical progression across the
focal plane (except in the case of leaf and various high-speed and special purpose shutters).

It does appear to be a different angle, since I notice in the reflection his head stick outs several inches from beyond the cabinet, however on the left his ear is flush with the cabinet.

From open to closed the eyelid takes 0.175 seconds. The slowest shutter speed setting that could be used without blurring is likely around 1/30 or 0.033 seconds. At this speed, the shutter is way faster than they eyelid. This isn’t rolling shutter.

A number of commentators mention the rolling shutter effect of airplane props. These props are making around 33 complete revolutions in one second. Constant eye blinking taking 350ms a blink would produce 3 blinks in one second. The props are moving much faster than the eyelid.

This is not rolling shutter. This is not rolling shutter.