A Study of the Poggendorff and Related Illusions


First posted Feb 2003, most recent revision August 2012

The discussion here is a bit involved.

Are you looking for an introduction to the Poggendorff illusion,

or to illusions in general?

Try these sites!

Michael Bach’s site - with a brilliant introduction to the Poggendorff Illusion

My illusion blog - with a Poggendorff category

Introduction and overview for this site

The Poggendorff illusion, usually shown as a geometric figure, can readily be demonstrated with a rod at an oblique angle with a plank in front of it.  The ends of the rod appearing either side of the plank are are aligned, yet appear misaligned.  The effect was first noticed just over 150 years ago, yet there is still no consensus as to its cause.  Some researchers attribute the effect to confusion between alternative three dimensional configurations that the figures might present.  Others invoke misjudgements of two-dimensional components in the figure, such as angles or extents.

I argue for two-dimensional effects.  There is one component of the illusion that is easy to overlook because it is invisible:  the trajectory across the gap between the ends of the oblique segments.   I follow researchers who have suggested that the slope of this invisible  trajectory is slightly rotated, due to interfering effects from cardinal axes projected into the visual field by the brain, combined, according to my account, with artificially strong axial and orientation cues presented by these geometric figures.

These dominant orientational cues, projected in and discovered in the figures, may be combined in everyday vision, I suggest, to fix gravitational vertical in the visual field, as the image on the retina swings with changing head orientation.  Suppose that the directional cues are represented in the brain by signalled but unrealised eye movements, and that these are also involved in judging alignment across gaps:  alignment might then be misjudged because of conflicting signal streams.

That’s a bit involved, but the justification for it comes from measurements of the way that apparent misalignment seems to vary as Poggendorff figures are rotated through 360 degrees - so when they revolve like the arms of a wind-turbine seen head on.  We can model the variation by summing effects attributed to cardinal axes, angle axes in the figures, and dominant edges in the figures.  The models rather successfully match measured patterns of misalignment in the experimental literature, as both the orientation of the Poggendorff figures and the angles between the test arms and long parallels in the figures vary. 


Welcome to the site,

But if you have a visual perception course component,


This is not peer reviewed, professionally endorsed stuff.

Don't believe a word of it!

You might not get credit for using it in coursework.