ACT-R

From emergent
Jump to: navigation, search

The ACT-R cognitive modeling framework is available as a native simulation environment within emergent.

Current Status

This is a work in progress -- currently (Aug, 2013), the parts that remain to be implemented include:

  • aural, speech modules
  • blending in declarative memory

Things that probably won't be implemented:

  • production compilation

Overview

ACT-R can be very useful within emergent both to develop higher-level cognitive models, and also as a powerful high-level programming language to control the behavior of emergent -- ACT-R models in emergent can call Programs and set variables, and productions can match against all relevant aspects of emergent state.

Compared to a normal procedural programming language, ACT-R is more flexible in using IF-THEN production system logic, that can detect various conditions and execute appropriate actions. It is relatively easy to add new flexible cases and situations. It also can flexibly retrieve memory information in a content-addressable manner, and use that to guide ongoing processing.

Demo Chapters

The best way to learn the system is to explore it through these demos. In the emergent demo directory, there is a collection of demos in the actr subdirectory, corresponding to the units in the ACT-R tutorial, with specific info about how the emergent impl differs.

  • Unit1 -- ACT-R/demo/count -- first introductory demo model -- start here for big picture overview and core modules (procedural and declarative).
  • Unit2 -- ACT-R/demo/demo2_vismotor -- visual, imaginal, and motor modules.