I imagine that to be divided into at least two processes, though (just 20/20 hindsight speculation.)
Brainwork, wherein previous memory/impressions cause you to crunch some ideas of road-surface, geometry and such to make a soon-decision
--is the slowest and I think, most easily tied to aging plus physical condition.
Muscle-memory seems to me to be less age-debilitated and also the place where 'skill' resides mostly, (along with the undefinables like
--a mood of intelligent fearlessness?/relaxing-intoa rhythm, with confidence) ... this doesn't work too well in words, I see. Ex:

A particular S-bend on a familiar group Sunday Morning Ride™: I recall a probably 98%? day, when I just did approach this spot 'relaxed'.. let the meat-machine go
for the airborne rise at whatever "V" seemed right. It was glass-smooth, in 3D. I could not have willed that trajectory sequence (nor 'explained' anything to a following Brit rider)
who noticed the dance, quipping, "thought we were practising on the Island, did we?"

But competent racers have to be in that groove, at 99% perpetually--and this was memorable because: I (knew) I can't Do that. (The ones who too-often try for 105% of their skill-set?
never learned creeping-gradualism 101, I wot: ego is such a retard; always forgets too, that pain hurts.)

(Hammond had no muscle-memory (or vehicle memory either!) so it sounds as if he did an exemplary job of it.) For me, the decisive deterioration is within the er, 'fluid in the inner-ear gyroscope?'
which, exactly like the analogy, behaves as if the viscosity is thicker==too slow to trust :-/
So I won't be tempted to imagine that muscle-memory would SYA, er SMA.

Hmmm.. Kepler and I share similar defects (as did Hubbell before the fix--I'm in good company), even if it has to be small-V vrooom :-)