Knock me over with a feather. I really didn't expect this. I learned about this first over at Phil Plait's blog, but given the title of my blog I think I really ought to address it.
First, some background. My obsession with interacting galaxies goes back to my first year of grad school, when, in 1991, I started working with Tom Soifer. This wasn't too long after the publication of the IRAS Bright Galaxy Sample— if you follow the link, you'll see that Soifer is the first author on that article. A lot of follow-up work was done (and is still being done!) on the class of galaxies that was established in this sample. In particular, there are a class of galaxies that are particularly luminous at infrared wavelengths. There are things going on hidden behind dust that prevents them from always being particularly bright at optical wavelengths. Many of the galaxies, especially the brighter ones, in this sample show evidence of vigorous star formation. Additionally, many of these galaxies show evidence of being in the process of a massive collision, or being the recent result of a major merger.
The association between galaxy mergers and strong nuclear starbursts remains robust. A search at ADS on appropriate keywords can bring up a lot of recent papers, and perusing them shows that the observational evidence and theoretical reasoning that leads us to believe that major mergers of two big galaxies triggers a vigorous burst of star formation near the core of the merger.
Stars are formed from gas. A compact nuclear starburst associated with mergers means that gas is getting funnelled down to the centers of galaxies when they're merging. You'd also expect this theoretically, as gas clouds in the two galaxies run into each other and dissipate bulk orbital energy in the fluid interactions. What's more, we have known for more than a decade now that all, or at least most, big galaxies have a black hole at their core. Finally, we know that the phenomenon of Active Galactic Nuclei (AGN) is fuelled by accretion on to a black hole. If you put all of these things together, you would expect AGN activity to be correlated with galaxy mergers. To turn on an AGN, you have to feed the black hole at the center of a galaxy, which means getting all the gas down to the center. We know that mergers can send a lot of gas towards the center. Thus, you'd expect a correlation.
It is true that if you're talking about a "compact" nuclear starburst, it's not nearly as compact as the accretion disk around a black hole. The compact nuclear starburst will be of a size that's hundreds, or even a thousand or so, parsecs... whereas the size of the accretion disk is much, much less than a single parsec. So, just because you can get the gas down far enough to make a compact nuclear starburst, it may well be that you can't get the gas the rest of the way down to fuel an AGN as part of a merger, and that other processes are needed. It's so easy to say that now, with this new result (arXiv preprint of the paper) showing that AGN activity is not correlated with the morphological signatures of a major merger. If I am to be completely honest, before today I wouldn't have pointed out the scale difference between an AGN accretion disk and the (much larger) size of a compact nuclear starburst. Indeed, in popular talks I've given about interacting galaxies, I've repeated what the general consensus (or at least one popular model) was, that mergers can trigger AGN.
It will be interesting to see if this result holds. I do feel a small twinge of envy that I wasn't the one (or one of the ones) doing this statistical survey, as in a previous life it was something that I had on my list of research topics that would be interesting to try to confirm. But, clearly, it was a topic of interest to a lot of people, and it's great that we've got a result now... even if the result isn't the one I expected or hoped for. I look forward to the further unfolding of this story.