One symptom of autism is a lack of "imaginative" play... though what is meant by "imagination" is a little odd. Autistic children do have imagination, indeed Asperger writes of children telling astounding stories about spaceships, and other such implausible things. What seems to be lacking is a projection of animate qualities onto toys (think Toy Story). In autistic children, imagination happens somewhere else (though with an odd exception: I've seen several cases sited where autistic children pretend that they themselves are something else, such as an animal).
I can corroborate this from my own experience:
When I was younger myself and my next older brother had a fairly elaborate shared fantasy world. Also we would also sometimes pretend to be manatees (I kind of aquatic mammal).
One day I asked my dad to build me a rocket ship. He went down to his workshop and busily worked away... at a little wooden model of a spaceship with little wooden versions of us sitting in it. I was bitterly disappointed at the time.
Later, in high school, I convinced myself that a certain set of fantasy novels was real, and described a world it might be possible to escape to (I didn't like high school much). I also went through a phase of believing in certain invisible creatures that lived in my house, and of myself possesing some kind of spiritual aspect that could, if I weren't careful, or did not follow the correct procedures for certain actions, become detatched from my physical body. For example, I would have to open both the physical and non-physical aspect of a door before walking through it or risk leaving my spirit behind.
<digression> Invisible objects are a potentially useful mental construct. For example, I imagine the idea of an object that only person X could see would be a nice object-oriented basis for a "theory of mind" in which different people have different models of the state of the world. </digression>
I am wondering then if normal children make a similar kind of mistake in ascribing animate properties to toys -- is it normal to go through a phase of thinking they are really alive?
The difference I presume arises from the difficulty autistic children have with metaphors. A toy may be similar in so many properties to the real thing that a Cauchy classifier might mistake it for such. A Gaussian classifier would demand at least reasonable correspondence in all properties, and not make that mistake.
I wonder what it's like, being normal.