ACT-R is a cognitive architecture -- a theory of thought and behavior embedded into a computational framework for representing human cognition. ACT-R and related cognitive architectures provide a way of simulating cognition: a user can specify a cognitive model of a particular task that interacts with a task environment and simulates the cognitive, perceptual, and motor processes necessary to perform the task. The architecture and associated models strive to perform tasks as much like humans as possible, and researchers typically compare model behavior to human behavior in order to demonstrate the psychological validity of their models and theories. Originally developed by John R. Anderson at Carnegie Mellon University, ACT-R is now actively being used by an international community of researchers and practitioners to study a variety of tasks.
This simulation and development environment provides an implementation of ACT-R in the Java programming language. The system can be used at three levels:
- Beginners. For users learning
ACT-R or writing simple models, the system provides
an easy-to-use interface that wraps the core ACT-R
implementation in an integrated editor and
simulation environment. The user can write ACT-R
models and immediately run them and inspect the
resulting simulation traces. When used in this way,
the system requires no Java programming
- Intermediates. For users wishing to code their own custom tasks, the distribution contains a Java archive file that can be included as an external library into a new code project. The user codes a task environment by extending the built-in classes for the task interface and associated components.
- Advanced Users. For users wishing to modify the
ACT-R architecture itself and/or the standard
development environment, the full source code is
available under an open-source license.
The current release of the ACT-R system is now available as a
downloadable application and as a runnable