The way that general anesthetics work is still somewhat mysterious. Here’s one intriguing hypothesis.
Anesthetics increase conduction velocity in myelinated fibers. Perhaps they disrupt the carefully calibrated timing of axonal transmission. This may selectively interfere with spike timing-based computation, while leaving rate-based computation intact. If consciousness (or some general class of higher-order functions) is spike timing based, but lower-order functions are rate-based, this would explain why anesthetics selectively affect higher-order functions.
Swindale, Nicholas V. Neural Synchrony, Axonal Path Lengths, and General Anesthesia: A Hypothesis. Neuroscientist 2003 9: 440-445
I didn’t read the article yet (I don’t have access privs to that journal), but it looks cool.