I was taught that it was primarily the effect of reduced downwash over the horizontal stabiliser, but that like verything in a helicopter there are multiple forces at play. Those included the moment arm on the CofG (see Shawn Coyle's post) and the possibility of an uneven redistribution of lift across the disk (not flapback) as per fingwing207.
I did find that flying Balck Hawks reinforced the lesson that it is primarily due to downwash reduction over the horizontal stabiliser. The Stabilator (big wing at the back) is driven by a computer that changes it's angle dependant upon a number of inputs, one of which is collective bias. This senses the rate and magnitude of collective movement and angles the stab to keep the nose level. When you lower the collective in the Black Hawk, the nose stays level, but the stab drives up to achieve it.
This explanation would suit all arguements, however the one that made lean toward the downwash theory was what happened when we lowered collective whilst fitted with the external support wings (ESSS). Those wings are situated forward of the rotor mast, and the stab computer doesn't know they are on. When we lower the collective, we get a pitch up. That would make sense with the downwash reduction over the ESSS not being countered by the stab computer, but it also backs up the CofG moment arm arguement. It tends to discount the effect of any redistribution of lift across the disk because that theory would dictate a nose down regardless of the ESSS.
I found the same thing on Hueys without a computerised (but moving) stabiliser. I did a maintenance test flight with an improperly rigged stabiliser, and during the max power check I contacted the forward cyclic stop trying to keep the rising nose under control. The only way out was to reduce collective and that restored forward cyclic authority. This suits the downwash arguement well. Lastly, when fitted with gunship "wings", the aircraft behaves like the Black Hawk with ESSS, ie it pitches nose up rather than down.