February 20th, 2012

beartato phd

(no subject)

Saw a talk by this guy Denis Pelli who did some neat research in the mid 90s about reading texts in the face of noise. Now while it makes perfect sense that a letter shape should have a somewhat localized fourier spectrum --- to someone who knows what a fourier transform is --- it's possible to get across the point with just one lovely image:

Opacity of the letters varies up and down, and spatial frequency of noise varies right and left. You can read letters on the left and right sides --- where the noise is either very high or low frequency --- but in the middle you can't.

Anyway, the talk was only so-so. The guy didn't talk particularly slowly but it felt like the rate of communication was sloooow slow slow. The audience questions got kind of out of hand, too.