I've always been of the opinion that attaching yourself to the pedals is more likely to result in landing on your head. Well not always - I formed that opinion after waking up one nice sunny afternoon in the middle of the road and no idea how I got there, and eventually coming to that conclusion.
Let's say you are doing 50km per hour downhill (abt 30mph)
That's 50,000m per hour, or
50,000/3,600m per sec = 13.89m per sec
According to the Human BenchMark site the median for clicking a mouse is 215 milliseconds. I suspect unclipping from spds would take a mite longer, but let's use that figure.
So distance travelled before unclicked is 13.89 x 0.215 = 2.98metres.
Let's assume you have hit something that stops your front wheel so now you are rotating around it.
Where will that 2.98m put your head by the time you unclick or let go the bars?
(I assume letting go the bars would happen about the same reaction time unless a decision was made to hold tight.)
Presumably it will be a complex action, and I'm trying to work out the scenarios.
1. Your body will pivot around the ankles until it hits the bars because your arms aren't strong enough to resist the forward motion, then the bike will pivot around the front axle if your brakes are on, or will it pivot around the contact patch? Anyhow maybe you have released by this point but your body is rotating.
2. Your arms are stiff enough to resist the impact (calculation suggests unlikely), so the pivoting starts immediately.
Anyone care to model this? I don't have the mathematical capacity - I've run out of fingers & toes on my Irish digital calculator (even assigning binary to them).