Mumford's first work that I know of after his mathematical peak was on machine vision, in the 1980s. I know this because I went back to the math department to visit and saw him running around with a video camera. I got the impression he was interested in how the actual brain works even back then.

This would be around 30 years ago.