The previous discussion of AGN host galaxy properties approaches the second question - the question of whether the energy liberated by black hole growth actually affects the host galaxy - only indirectly by observing that most AGN host galaxies appear to be stable disk galaxies with, at least locally, no evidence for recent enhancement or quenching of star formation.
This raises the question of where we actually see AGN terminating, or at least modifying, star formation properties in their host galaxies. More or less the only unambiguous cases are the giant ionized outflows observed in powerful radio galaxies by Nesvadba et al. (2006), Nesvadba et al. (2008). Even in the case of dramatic examples such as the powerful molecular gas outflows seen in Mrk 231 by Fischer et al. (2010), a starburst-driven wind cannot be entirely ruled out.
The commonality of black hole accretion phenomenology and physics exhibited by X-ray binaries (XRBs) and AGN has been discussed extensively in the literature (e.g. Maccarone et al. 2003, McHardy et al. 2006, Körding et al. 2006) and is particularly interesting since some XRBs are known to put out large amounts of kinetic energy, directly impacting their environment (e.g. Gallo et al. 2005, Pakull et al. 2010). If, as has been hypothesized, AGN can undergo rapid accretion state transitions similar to XRBs, could the radiatively efficient (quasar) phase be less important for feedback work? And could radiatively inefficient, kinetic outflows be how a large fraction of feedback work by black holes actually occurs? If so, then how do we find these radiatively inefficient - i.e. dim - AGN?
One case where this might be occurring is the nearby (z = 0.05) galaxy IC 2497, which features a light echo of a recent (> 200,000 yr) powerful quasar outburst preserved on a giant external atomic hydrogen cloud (Lintott et al. 2009, Józsa et al. 2009, Rampadarath et al. 2010, see Figure 3). The nucleus of IC 2497, which hosted a Lbol ~ 1046-47 erg s-1 less than 200,000 years ago, is now at least four orders of magnitude dimmer (Schawinski et al. 2010c). The rapidity of this `switch off' makes IC 2497 the best and most accessible place to test the hypothesis that quasars can undergo state transitions into a radiatively inefficient, kinetic mode.
Figure 3. Hubble Space Telescope image of the galaxy IC 2497 (top) and Hanny's Voorwerp (green, [OIII] 5007). The Voorwerp is a light echo from a past quasar episode of the dormant black hole in IC 2497. The light travel time between the nucleus and the Voorwerp implies a time delay of less than 70,000 years, preserving a record of a powerful quasar phase in IC 2497 in the recent past (credit: NASA, ESA, W. Keel (University of Alabama), and the Galaxy Zoo Team).
There are now other examples of similar 10-100 kyr timescale variability in local AGN (Keel et al. 2011, Marecki & Swoboda 2011a, Marecki & Swoboda 2011b), giving further support to the XBR-AGN analogy and raising the prospect that radiatively inefficient accretion modes are an important missing piece in the feedback puzzle.