Abstract
The standard view in IE linguistics is that the PIE dorsal (tectal) stops evolved in two general directions. In the satem languages, the front dorsal series (i.e. "palatovelars") evolved towards sibilants (or even dental fricatives); in the centum languages, front dorsals shifted backwards to mid-velar point of articulation.
As corollary to this shift, back dorsals ("labiovelars") lost the labial feature in the satem group but it was retained, apparently, in the centum branch. Also, depending on one's stance towards the "neutral" dorsal, it would appear that either one or the other series merged with the putative mid-dorsals as a result of the centum-satem development.
In the following, I attempt to show that there might be reasons to reconsider the necessity to reconstruct back dorsals as phonologically labialized and to see the shifts of either series as in essence independent of the other and not related to the contrast balance of the whole dorsal area. I believe that it can either be the case that some of the features that we posit for the proto-language are in fact emergent as result of the dynamics of the dorsal subsystem or that at least the satem change could be described as a push-chain-shift.
The crucial mechanism in both cases is enhancement of secondary acoustic features of these elements compensating for the simultaneous neutralization of their primary features.