A very interesting minireview to appear in Trends in Cognitive Sciences is asking if domesticated dogs' social skill in reading humans does not indicate a heritable trait selected for during domestication.
In specific tests of social cueing, dogs do far better than wolves-- and also even better than chimpanzees. The main example given is a test in which a new toy is hidden under one of identical buckets, and, as the dog enters the room, a human points to the correct bucket. Even litter-raised puppies, with little direct human contact, are able to understand that a human pointing a finger has an information content. Chimpanzees do not do well in this sort of test, but human infants as young as 14 months also perform it with ease.
Dogs will also prefer to "fetch" a ball back to an observer with open eyes over one with closed or masked eyes; and will carry the ball around to the front of an observer with back turned. Try to turn your back on a toddler!
What makes this interesting is that dogs are not exactly the top of the heap when it comes to problem-solving. It's rather social aspects, and the bond with humans, which seems to be where dogs shine-- even more than the quite social wolves. The authors believe that the genetic tendency to bond with humans, which was selected during domestication of dogs, led also to this ability to read social cues coming from people. They propose this when summarising a long-running (since 1959) experiment with domestication of foxes, which have been selected for (bred for) approaching humans without fear or aggression. The "domesticated" foxes were very good at the pointed-finger test, much better than the unselected control population, even though the breeding program had begun decades before that test was known.
Thus a specific social skill might be heritable, and selectable, and different from generic problem-solving intelligence. What makes this an interesting opportunity to a geneticist is the very big evolutionary distance between humans and dogs, and the availability of undomesticated wolves. It might be possible to tease out a genetic description of these tendencies.
It is obvious that many animal species are highly capable problem solvers, and that this reflects several flavors of intelligence. This is a tough field, rife with the danger of anthropomorphism. My own reaction on reading this story is needing to know more about the social skill intelligence of dogs and humans, its distinction from other cognitive skills, and how to develop a bigger range of tests around it. The second item would I think be to test feral dogs, who have not lived around the smell of humans. And lastly, I myself would want to re-read the chimpanzee data to see if this social intelligence is really completely absent from them.
UPDATE,August 9: Science Now (possibly subscription only) is also running a blurb on dog society. Dogs meeting at a free-run park use sterotyped signals to get one another's attention, like sniffing, galloping, play biting. The idea is they know if the other one is distracted or paying attention. (Note to self: could have applications in lab meeting.) From the blurb I'm again left wondering if this is unique to dogs, or part of the toolkit for a social animal. Cute pictures, though.