In terms of what the pulse sequence is doing to the magnetisation, nothing.
It's to do with how the signal is displayed in the image.
You can represent the signal from an inversion recovery experiment with its true -M0 to +M0 recovery (where M0 is the equilibrium magnetisation) as a range of grey-scale values from black (-M0) to white (+M0), where zero magnetisation - the null point - is grey. This is true inversion recovery, sometimes called Phase Sensitive Inversion Recovery, or PSIR. PSIR is only used in specific circumstances that I can think of (neonate brains, some cardiac scans).
Or you can do what we normally do with the MR signal, and that is make a magnitude image
with it, discarding the phase information but gaining an SNR bump. This would mean that the image grey values go from white (-M0) through grey to black at the null point and back to white again (+M0). This can lead to ambiguity where a particular grey scale value may indicate two different T1s. Where that ambiguity is important, PSIR is used.