Last June, Cecilia Heyes and Chris Frith published a nicely written paper on the cultural evolution of mind reading in Science. Note that when I'm talking about mind reading, I do not mean the science-fiction practice of telepathy. Like Heyes and Frith, with mind reading I mean the ability to understand the actions, intentions, feelings, and thoughts of another person (I prefer the term theory of mind or folks psychology). In the paper, Heyes and Frith defend the cultural evolutionary account of mind reading—this basically means that mind-reading is a skill like reading, writing, or bicycling, and that it is taught from one human to another. However, this does not mean that there are no genetics at work. You still need the right genetic tools that allow you to learn the skill (try to teach your cat to cycle a bike).
In their paper, the two authors distinguish between two types of mind reading: implicit mind reading and explicit mind reading. Explicit mind reading means that we are deliberately reasoning about and thinking of someone's beliefs and intentions. Implicit mind reading means that we are automatically thinking about someone's mental states (or at least, presumably think about someone's mental states). Whereas explicit mind reading evolved culturally, implicit mind reading did not (necessary) evolve that way. Indeed, there is a variety of evidence that show that the ability to explicitly reason about someone's mind is indeed learned by language and therefore this points to cultural evolution—if children are not exposed to a rich variety of mental-state vocabulary, they lag behind in explicit mind reading (or theory-of-mind reasoning). However, I can imagine that if you are not exposed to a variety of mental-state vocabulary, you would also have a hard time doing a test that requires you to use your mental-state vocabulary (even thought you could in principle do the test).
The notion of implicit mind reading comes from studies that show that infants exhibit gazing behaviour and prolonged looking time that corresponds to someone else's beliefs. Additional evidence comes from experimental research that shows that we are distracted by the perspective of another person (or avatar). Some researchers interpret this evidence in favour of a neural system that allows us to automatically reason about someone's mental states. However, as Heyes and Frith argue, most evidence in favour of this hypothesis could be interpreted in terms that do not require mental states (they call this submentalizing). This separation between implicit and explicit mind reading is necessary for Heyes and Frith their argument to hold. A fast automatic system that we already possess as infants and that allows us to reason about someone's mental states is unlikely to be the result of cultural evolution.
I like the line of reasoning that Heyes and Frith present in their paper. However, I do think that they are missing out on a large part of implicit mind reading. For example, the mirror neuron system (a system of neurons that activates not only when we perform an action but also when we observe the same action) that we supposedly use to interpret the actions and intentions (this is an ongoing debate) at some low level is not mentioned by the authors. One could argue that the mirror neuron system is an implicit system and provides building blocks for reasoning about the intentions of others. Furthermore, the ability to recognise emotions and feelings is also not discussed in the paper. One could argue that the recognition of emotions is an implicit mind reading skill that allows use to infer one's mental state without too much reasoning. These kind of implicit mind reading skills might act as input for more higher explicit mind reading skills. Although the hard distinction between implicit and explicit mind reading made by Heyes and Frith is appealing, it does not prove beyond any doubt that this distinction is real (in my opinion).