Think of roll moments and roll inertias. Similar to linear F=ma but angular rates and angular inertias. I don't have my text book handy so won't remember the exact equations or units (I hear cheers of gratitude) but....

The roll acceleration is the rolling moment due to ailerons, Clda x da (rolling moment due to ailerons times aileron deflection) divided by the roll inertia (with proper care taken of the units which I didn't). The roll angular deceleration is due to roll damping, the Clp x p, again the rolling moment that is due to the roll rate (there is a differential angle of attack on each panel due to the roll rate and a corresponding lift differential on each panel) times the roll rate. The roll moment in ft-lbs is those two Cl times a bunch of stuff.

So you start out with the initial ailerons deflections causing a constant angular (roll) acceleration. As the roll rate increases there is an opposing roll moment due to the roll rate. That causes an angular deceleration.

This keeps up until there is a steady state roll rate achieved with the aileron deflection held in - the input moment is equal to the damping moment. Remember that the moments in ft/lbs are a function of rotation speeds, etc.

When the aileron control is released you now only have roll damping and the roll rate.

In a perfect world the deceleration would be initially high but would asymptotically approach 0 - it would take a long time to get to zero - resulting in the need for a little reverse aileron to stop the final little bits of roll rate.

In the model airplane low roll inertia real world the deceleration takes place very quickly and we get to a point where the roll rate is insignificant very quickly. You don't have to use any reverse stick. Usually a bump in the air comes along to jiggle things to require a stick input before the residual roll rate is a problem. So the result is that it looks like all you have to do is release the stick and the roll stops. Also the residual roll isn't a problem as you are turning or doing something else.

In the old days we would calculate on big sheets of paper the roll time history of something like a big transport, not a good way for an engineer to spend his time. There is an approximation to the steady state roll rate that I don't remember but it doesn't tell you how you got to the steady state roll condition - which is important. Now we use a computer. I ran those programs for a couple of years, not all that much fun either as you had to hand plot the results at that time.

Aside from the last bit of gripe stuff is this any help?