WISE  [3.4]  –  [4.6]  color (red stars),  [3.4]  –  [12]  color (black diamonds) and  [3.4]  –  [22]  color (blue triangles), plotted against the 3.4 μm absolute magnitude. Both the  [3.4]  –  [12]  and the  [3.4]  –  [22]  color show the expected trend (they become redder for late-type stars) and can be fitted with a polynomial. The  [3.4]  –  [4.6]  color is expected to show a similar trend and furthermore be smaller than the other two colors for any spectral type. This is clearly not the case for the bright end of our sample (above magnitudes of  ~6 or absolute magnitudes of  ~8–9). We therefore omit the second channel in our analysis.