Our brains are great at knowing when they can disengage. Usually when we're doing something humdrum and run-of-the-mill. However, they spring into action when things are incoherent, or off in some way.

The Raw Ideas generated by ideate bot are designed to engage and target the storyteller^{* (learn more)} by introducing a bit of chaos and incoherence to otherwise semantically close concepts.

To accomplish this it presents small batches of raw ideas, 3 - 5 at a time to avoid overload (our brains can quickly become frustrated if overworked).

Raw ideas are purposefully obtuse because they need to prevent your brain from understanding the semantic intent of each idea instantly.

Semantics is the study of the logic and meaning in linguistics concerned with how the cognitive structure of words, their relationships and their meanings are codified in your brain. What's interesting about this is that there are semantic distances between these words and concepts.

Here's an example.

What is more closely related to a dog?

A) wolf

B) pig

C) submarine

You likely chose A because a wolf is the most closely related to a dog biologically. They are both canines.

What is the next most closely related, pig or submarine? You likely chose B, pig. It's not a canine but it is an animal.

So wolf is closer to dog than pig, but pig is closer to dog than submarine. The semantic distance between dog and submarine is large, while the distance between dog and wolf is small.

Trillions of these semantic relationships are sitting right there in your brain, just laying around.

Time to put 'em to work.

Back in the 2000s a neuroscientist called Michael Gazzaniga, working at Roger Sperry's lab at Caltech, was studying the effects of having the corpus collosum severed.

The corpus collosum is a dense bundle of neurons that connect the two halves of our brains. If this is severed it means the two hemispheres can no longer communicate with each other, effectively isolating the functionality of each half. The condition is called split-brain.

Sharing his experiences in an article for Discover Magazine, Dr. Gazzaniga relates this interesting story.

We showed a split-brain patient two pictures: To his right visual field, a chicken claw, so the left hemisphere saw only the claw picture, and to the left visual field, a snow scene, so the right hemisphere saw only that. He was then asked to choose a picture from an array placed in full view in front of him, which both hemispheres could see. His left hand pointed to a shovel (which was the most appropriate answer for the snow scene) and his right hand pointed to a chicken (the most appropriate answer for the chicken claw).

We asked why he chose those items. His left-hemisphere speech center replied, "Oh, that’s simple. The chicken claw goes with the chicken," easily explaining what it knew. It had seen the chicken claw. Then, looking down at his left hand pointing to the shovel, without missing a beat, he said, "And you need a shovel to clean out the chicken shed." Immediately, the left brain, observing the left hand's response without the knowledge of why it had picked that item, put it into a context that would explain it. It knew nothing about the snow scene, but it had to explain the shovel in front of his left hand. Well, chickens do make a mess, and you have to clean it up. Ah, that's it! Makes sense.

What was interesting was that the left hemisphere did not say, "I don't know," which was the correct answer. It made up a post hoc answer that fit the situation. It confabulated, taking cues from what it knew and putting them together in an answer that made sense.

In short, we come with specialized mental machinery designed to confabulate. To take seemingly random facts or concepts and combine them into a coherent (more or less) story.

That's interesting for a neuroscientist. It's pure gold when you're a plucky little ideation bot wanting to solve the problem of creativity on-demand.