So.  Some stroke victims that have experienced paralysis on one side of their body experience something called "anosognosia".  They believe that the paralyzed limb attached to their body does not belong to them.

In fact, fibers in the motor cortex on the right side of her brain, which controls movement on her left side, have been irreparably damaged by the stroke, and she will never use her left arm again. But Mrs. M. is not a stubborn old woman refusing to admit a difficult truth. A few minutes later Mrs. M. looks at her left hand, resting inertly in her lap.

Doctor, she asks, whose hand is this?

Whose hand do you think it is?

Well, it certainly isn’t mine!

Then whose is it?

It is my son’s hand, Doctor.

They honestly, completely, consciously deny the fact that the paralyzed limb belongs to them. 

This article is fascinating if a bit long: The Brain That Misplaced Its Body

Should we design our AI to be weird too?

A tangential discussion can be raised around the attempt to create "hard" AI.  There a lot of people that think we are on the verge of creating AI.  I know, I know it's been talked about for decades but you must admit, there is compelling facts of computer speed and memory size that are providing some merit behind the discussion.

Well what if one of the key features of our intelligence is the plain weirdness of the architecture and structure of the human brain.  What if all the randomness and madness is a necessary "razors edge" that's needed to achieve consciousness and intelligence?

Can we architecture weirdness into a design?  Can logical computer scientists plan for all this fuzzy logic?

Can they make an AI doubt the existence of its limbs?