Did you mean that even at the very low voltage levels, (uV to mV range) reverse biased capacitors does not behave as a capacitor. That is to say there is no charge stored at the capacitor plates.
No, I did not mean that. In this special case of electrolytic capacitors, the dielectric is basically a thin oxide layer and a electrical double layer formed on that.
I do not know exactly but at low reverse voltages (<1-5V, I presume) the oxide layer is intact and the electrical double layer undergoes changes and that affects the capacitance.
At high reverse bias voltages, the oxide layer develops faults (somehow it becomes leaky) and the defects apparently never heals.
In essence, if you have a 25V 10uF capacitor (Al electrolytic), it will not get damaged if you apply 1-2V in the reverse way. But 25V reverse voltage will kill the oxide layer and the capacitor will become leaky (not hold charge).
If you apply reverse voltage (low value; say 1-2V) and measure the capacitance it will find anything from 5-7uF and the readings will be unstable.
Of course high voltage electrolytic capacitors are more sturdy and they can perhaps take more abuse.
- - - Updated - - -
I doubt that you can substantiate the statement for mV bias.
No, I do not mean that with mV reverse potential. I do not recall accurately but the voltage drop across the electrical double layer is close to a volt and hence mV bias will not show much effect. But rarely a capacitor is used at mV potentials.