with uncertainty that takes measurements, those measurements should serve as a correction base on the degree to which the measurement is believed to be statistically accurate.

For example, in a robot system, you can drive for a little while with only accelerometers and gyros, but after a while the noise in those sensors cause the estimated position to drift. Those are high frequency estimates. Then, when a GPS measurement comes in at a slower rate, you correct your estimate base on the higher accuracy and higher trustworthiness of the measurement.