In some of the old-time sense data literature you’ll read people saying that if you look at a coin held at an angle to you, it looks elliptical. It’s not clear what relevance this would have to debates about perception even if it were true, but the fact is that for most people it isn’t true. A coin held at an angle to you looks like a circular object held at an angle to you. It takes quite a bit of effort, if it is possible at all, to see it as elliptical. The general point is that the kind of correction that is needed for a coin to look circular (even though its retinal impression is elliptical) happens very early on in processing, so we have little or no conscious access to any elliptical impression the coin makes.

I’m sure there’s actual data on this, but it’s interesting to test how far this extends. The other day I was staring at the (circular) logo on a Starbucks cup and I noticed something rather odd. If you hold the cup with the logo facing towards you, tilt the cup towards you (as if you were drinking from it) and look down at the logo, I think it won’t look circular. To my eyes at least the bottom half of the logo looked like a semi-circle (being viewed at an odd angle) while the top half looked very flattened. The overall impression was sort of like a shield with curved corners.

Now there’s two features of the Starbucks cup that could be causing this distortion. First, the cup is curved. Second, the cup tapers in towards the bottom. (So overall it is, I think, a cross section of a cone.) I wonder which of these is responsible for the distortion effect here. Careful inspection of a Nantucket Nectars bottle (which is cylindrical in the crucial section) makes me think it is the first.

Evolutionary explanations for why we are able to make one kind of correction but not the other would be welcome. I think it’s evidence that we did have two dollar coins in the savannah, but we didn’t have Starbucks then.