I hesitate on this explanation in this specific case just because the reason that we think head tilting is cute in dogs is because we also do it - and IMO we also do it when we're remembering something successfully - so I'd check for it in non-domesticated animals (wolves) and in other mammals.
Steven Pinker on the evolution of the shrug:
Coming up with a definition of intelligence that doesn't need an observer/understander is an interesting exercise; kinda demonstrates my point inasmuch as the word stops taking on a unified meaning and we find ourselves grasping around for the real Y-axis.
That said I think the head-tilting somehow increases cognition for a second, and the social signalling is an added benefit. Maybe it's a blood flow or cerebrospinal fluid thing. Big brains use a lot of blood oxygen and glucose while they're working. But that's just me BSing.
Regarding the cane corso-- fighting/guard dogs don't do the head tilt, but they're not gonna learn the words either...
Also, dogs evolved to communicate with dogs for ~(10E7) years, then learned to communicate with humans for ~(10E4) years. I think they mainly speak dog.. well, not dog, but puppy; the common path of mammal domestication being through juvenilization.
Certainly various breeds have been bred for different purposes, and are very good at them.
We think it is cute today. Did humans think it is cute thousands of years ago? Can you assume that humans that were domesticating wolves thought any animals were cute or attractive in some form?
These are border collies, a working dog designed to interact with a human handler. Take a look at something like a cane corso or great pyrenees. They don't head tilt nearly as much because such close communication is not what we have, through breeding, designed them to do.