Reading guide: ACT: A Simple Theory of Complex Cognition
Background
This paper describes one of the major symbolic cognitive modeling
architectures. What's a cognitive modeling architecture? Think of it
as a programming language, and approach to programming that are meant
to reflect human cognition.
Things to watch out for
- What's recursion? It's not important that you know how to do
it, but that you get a bit of a sense of what it is and why it's
tricky. Really think about about the example rule on the first
page, and think about what it means that recursive programming can
be modeled with 500 such rules.
- The main claim is on the second page. Fig 1 simply shows that
the ACT programming model behaves like humans in terms of how the
error rate changes with practice.
- The guts of the architecture start in the Representational
Assumptions section. In Fig 2, the shape of the thing doesn't
matter, but the collection of related bits does. What's the
relation with the first bit on page 357? Really dig in here and in
Fig 3, trying to understand how it works.
- The Knowledge Acquisition section talks about learning. This is
important stuff. How do these neat little rules get in the model
in the first place?
- Next section: How do we use the right knowledge at the right
time? Also important. Try to get the gist of Fig 4. Don't worry
about memorizing the formulas, but do try to understand what they
are saying about the factors that affect knowledge deployment. The
rest of the figures here reference various ACT theses which support
the cognitive plausibility of ACT. Try to get the overall idea of
as many as you can.
- The last section has a nice analogy for cognition. Do you buy it?
Last modified: Mon Sep 22 18:16:12 CDT 2008