All variables have an initial value, which is the mean value of this variable in a normally functioning, physiological human. Mathematically, the error from normal of the controlling variable is calculated, and a gain value is used to calculate the desired value of the target variable, as a deviation from its normal value. The target variable's value is then gradually moved towards this desired value, at the given rate. The rate is measured as a fraction of the desired change per minute; i.e. 1.0 gives instantaneous correction, and 0.5 will decay towards the desired value with a half-life of 1 minute.
There are several types of curve that can be used to calculate desired values from errors - these correspond to different gain functions:
desired = target_normal + (controlling_current - controlling_normal) * gain + constant . If 'limited' is selected, the desired value is confined to the normal range.
desired = target_normal + constant * (error ^ gain)
desired = target_normal * (controlling_value / controlling_normal)^gain + constant