"Marc J. Zeitlin" wrote:
At high frequencies (above 1000-2000 Hz), the wavelength of the sound
starts becoming very small.

The wavelength at ~2000 Hz is still about ~6 inches.

This implies that for the microphone to
measure (and hence cancel) the noise that the EAR is hearing, the
microphone would have to be essentially inside the ear, or else it
would be canceling something other than what the ear perceives.

Two limiting factors would seem to be the distance between the external
microphone and the headphone, and the frequency response of the feedback
electronics. If your feedback response is slow, it might actually pay to
design it so the external microphone is farther away from the headphone.

Since sound approaches from all directions, and the microphone and
headphone must accomplish their tasks in only a one to two-dimensional
space, a three-dimensional "cone of silence" device is best. Fortunately
such a device was invented back in the 1960s; here's some pictures:

http://www.oneeyedman.net/cone_of_silence.jpg
http://www.hojohnlee.com/weblog/wp-c...of-silence.jpg