We present luminosity functions derived from a spectroscopic survey of AGN selected from Spitzer Space Telescope imaging surveys. Selection in the mid-infrared is significantly less affected by dust obscuration. We can thus compare the luminosity fun
ctions of the obscured and unobscured AGN in a more reliable fashion than by using optical or X-ray data alone. We find that the AGN luminosity function can be well described by a broken power-law model in which the break luminosity decreases with redshift. At high redshifts ($z>1.6$), we find significantly more AGN at a given bolometric luminosity than found by either optical quasar surveys or hard X-ray surveys. The fraction of obscured AGN decreases rapidly with increasing AGN luminosity, but, at least at high redshifts, appears to remain at $approx 50$% even at bolometric luminosities $sim 10^{14}L_{odot}$. The data support a picture in which the obscured and unobscured populations evolve differently, with some evidence that high luminosity obscured quasars peak in space density at a higher redshift than their unobscured counterparts. The amount of accretion energy in the Universe estimated from this work suggests that AGN contribute about 12% to the total radiation intensity of the Universe, and a high radiative accretion efficiency $approx 0.18^{+0.12}_{-0.07}$ is required to match current estimates of the local mass density in black holes.
We present preliminary results on fitting of SEDs to 142 z>1 quasars selected in the mid-infrared. Our quasar selection finds objects ranging in extinction from highly obscured, type-2 quasars, through more lightly reddened type-1 quasars and normal
type-1s. We find a weak tendency for the objects with the highest far-infrared emission to be obscured quasars, but no bulk systematic offset between the far-infrared properties of dusty and normal quasars as might be expected in the most naive evolutionary schemes. The hosts of the type-2 quasars have stellar masses comparable to those of radio galaxies at similar redshifts. Many of the type-1s, and possibly one of the type-2s require a very hot dust component in addition to the normal torus emission.
