How do we explain the phase change between high and low star formation rates? During a recent talk I was at, someone, who models galaxy characteristics from dust amounts and spectral energy distributions (SEDs), quoted a fairly prompt change from 'high' to 'low' stellar formation rates.  
How do we explain this prompt phase change? (If at all!)
 A: My guess would be that this person was commenting on the distinction between galaxies undergoing "normal" star formation, as happens in the disk of the Milky Way, the majority of local galaxies, and, evidence increasingly suggests, as happens in the majority of galaxies at high redshift too. Recent work with the Herschel Space Observatory, published in a GOODS-Herschel: an infrared main sequence for star-forming galaxies finds evidence for a galactic star formation main sequence, somewhat analogous with the stellar main sequence. That is, there is a wide range of star formation rates observed, but star formation mostly looks the same, regardless of the rate of star formation. Instead of mass and luminosity (or color and magnitude) being fundamental, as for stars, the galactic main sequence is star formation rate and total stellar mass. 
Just as with the stellar main sequence, however, there are a fraction of galaxies that are in a different stage of evolution. In the case of galaxies, these outliers probably have recently experienced a merger. As a result of the merger, their molecular gas and dust were funneled in to a compact nuclear star formation region. These galaxies experience star formation rates that are significantly greater than would be predicted from looking at their total stellar mass. In the local universe, most ultra-luminous infrared galaxies (ULIRG) are experiencing extreme star formation events as a result of merger. At higher redshifts, however, the mean star formation rate in galaxies is higher, and many galaxies at high redshift form stars at very high rates, but do so in a manner that is more similar to a scaled up Milky Way than to a local ULIRG, such as Arp 220. A slightly older paper, Different Star Formation Laws for Disks Versus Starbursts at Low and High Redshifts, by some of the same authors, makes similar conclusions, on the basis of molecular gas observations.
A very recent paper in ApJ, A Universal, Local Star Formation Law in Galactic Clouds, nearby Galaxies, High-redshift Disks, and Starbursts, addresses some of the suggested mechanisms for producing these differences in star formation rate. They suggest that all star formation rates can be understood in terms of a single law: 

small solar neighborhood clouds with masses ~103 M sun to submillimeter galaxies with masses ~1011 M sun, fall on a single star formation law in which the star formation rate is simply ~1% of the molecular gas mass per local free-fall time

As the free fall time goes as the inverse of the root of density of gas, high rates of star formation relative to the stellar mass in a galaxy can be understood in terms of the high gas densities that result from mergers. This result suggests that there is no phase change, exactly, but rather enhancement in molecular gas density. 
