When an increment (S+) or a decrement (S-) is added to an ongoing pedestal in a detection task, performance is far better than when all trials have a signal and the task is to identify its valence. This is the opposite of predictions from a standard detection theoretical analysis. It has been suggested that the detection task reflects sensitivity to transients associated with stimulus change while identification task reflects analysis of the sustained level, i.e., the energy. To be complete, this must argue that detection is of absolute change, regardless of its sign. To test this, detectability was measured with bidirectional signals (S+) and (S-) was compared that with only a single signal type, either (S+) or (S-). This was done in paradigms: (1) with the pedestal constant and (2) with gaps in the pedestal on either side of the signal that prevented the use change per se. Results show the hypothesis, showing that monodirectional signals were equally detectable as bidirectional signals when there were no gaps but with gaps, they were better.