Home > Uncategorized > Tabletop – an AI model of how we make analogies

Tabletop – an AI model of how we make analogies

Try this one night after dinner, before you clear the table. Grab some of the objects near your side of the table, say some pieces of fruit, and do something with them — make a pile, throw them around, create a shape, do a puppet show. Whatever comes to mind. Then ask the person sitting across for you to do the same thing.

Chances are your dinner partner doesn’t have exactly the same objects at her disposal, so what will she do? Well, that depends on what she perceived to be the most salient feature of what you did.

Let’s say you balanced a banana on top of an apple that you plopped on the remains of a strawberry tart. What’s the most important feature of that act? Is it the putting-things-on-top-of-other-things aspect of it, the gathering-fruit aspect, the yellow-green-red pattern, the balancing act, or what? Of course there’s no right answer, but it’s fascinating to see what people actually focus on, and how they have to stretch their analogies based on the objects available to them.

This ability to perceive analogies and adjust them to the current situation is at the heart of the IA research being done by the Fluid Analogies Research Group (FARG), led by Douglas Hofstadter of Gödel, Escher, Bach fame. FARG thinks that analogy-making and “cognitive fluidity” are fundamental to the way humans solve problems, and they built an AI software architecture that simulates how they think this actually works in the human brain. The result is the Copycat framework, which consists of three pieces:

• The Slipnet, a model of long-term memory with concepts connected to each other with links of varying strengths

• The Workspace, a model of short-term memory where analogy “fragments” are assembled and tested

• The Coderack, a collection of executable code that can test new hypotheses in the Workspace

[ok, now i’m just going to copy from Wikipedia:] “The resulting software displays emergent properties. […] It shows mental fluidity in that concepts may slip into similar ones. It emulates human behavior in tending to find the most obvious solutions most of the time but being more satisfied (as witnessed by low temperature) by more clever and deep answers that it finds more rarely.”

Robert M. French, one of Douglas Hofstadter’s students at FARG, used this architecture to create a program called Tabletop, which plays the game I described above.

Advertisements
  1. No comments yet.
  1. No trackbacks yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: