Where do early-type void galaxies come from?

(Figures 1+2 of Croton & Farrar, arXiv:0801.2771)

Not from anywhere particularly inspiring, apparently. The authors
investigate red and blue galaxy populations in voids using the
Millennium simulation. On the left are the predicted Millennium
galaxy luminosity functions (solid lines) and observed 2dF LFs for
the global sample (solid points) and voids (open circles). The right-
hand plot shows the predicted and observed LFs of red and blue void
galaxies, with galaxies at the centers of their halos (i.e. non-
satellites) shown as a dotted red line.

A "radio-mode" feedback model, efficiently quenching star formation
above a threshold halo mass but independent of environment, was
employed and tuned to fit only the global 2dF luminosity function.
As the left panel shows, this more or less automatically matches the
void galaxy LF as well, and reproduces the relative red and blue void
populations reasonably well (right panel). The authors conclude
there is no unusual physics going on in void galaxy formation, and
further that the presence of "red and dead" void galaxies is not
actually a big problem for galaxy formation models. As in denser
environments, red galaxies are simply those that occupy the most
massive halos.