But, again, what reason do we have to believe that other humans have interiority if we ignore the indications of interior life that dogs share with humans? Humans produce more complex communications, but those communications are not necessarily better explained by interiority than by mere generative complexity.
This is not to say that I don’t believe humans feel pain, but instead, to say that if *you* believe that humans other than yourself feel pain then you should probably also believe that dogs feel pain, since there is roughly equal evidence; likewise, with intentionality, we can estimate it by looking at planning effectiveness. The closer a set of behaviors is to being the ideal path toward some goal, the more likely it is that those behaviors were planned by a goal-persuing system. Proving intentionality with a 100% success rate is not possible, just as proving interiority with a 100% success rate isn’t possible: a system with strong intentionality that is working off flawed axioms or flawed data or that is very limited in how many steps it can plan ahead will be almost indistiguishable from a system that performs behaviors semi-randomly based on simple rules without memory. But, communication is neither necessary nor sufficient to prove either interiority or intentionality: after all, limping and yelping are, effectively, communication insomuch as they are behaviors that provide information to humans and other animals about how likely the dog is to be in pain, and yet we can ignore those signals based on the assumption that all attribution of inner state to non-human animals is pure anthropomorphism.