Well, in the 1990's they discovered 'mirror neurons'. This book review goes over a couple of new books on this fascinating subject, here's a snippet:
The experience of artificial-intelligence researchers has taught us that what seems simple for humans (like recognizing faces) is difficult for computers, and what is difficult for humans (like playing chess) is easy for computers. Imagine that you are at a café with a friend. He reaches out toward a cup on the table between you and picks it up. What he is doing, and why, is obvious: He is drinking his coffee. But how does your brain figure all of this out in a split-second, without any need of conscious reasoning?When I see someone smile, I don't have to infer what it means, I know instinctively because the exact same neurons fire in my head when I smile. This has a lot of implications about empathy, language acquisition and autism, in that they argue there is something in these mirror neurons that is different in autistic people. Thus, even though there is no logical reason we must assume consciousness in others, it is a built in feature.
When you see a woman grinning broadly, how do you know she is happy? Apparently not through some abstract reasoning process that analyzes the visual information coming from her face and makes a logical inference that her expression is associated with happiness. Instead, you instantly activate parts of your brain that are active when you are happy. You may even make subtle, unconscious muscle movements to imitate her expression.
This explains why we naturally infer consciousness and emotions so much easier than AI--it's an assumption! I wonder how many other puzzles can be solved in the same way, that is, we wonder why people believe X, and the answer is, because it's hard wired.