Figure 2.

Sample probability mass diagrams for two predictors $S_1$ and $S_2$ to a given target T. Here events in the two different predictor spaces provide the same amount of pointwise information about the target event, ${log}_2\raisebox{1ex}{$4$}\!\left/ \!\raisebox{-1ex}{$3$}\right.$ bits, since $P(T|s_1^1)=P(T|s_2^1)$, although each excludes different sections of the target distribution $P(T)$. Since they both provide the same amount of information, is there a way to characterise what information the additional unique exclusions from the event $s_2^1$ are providing?