ACT-R

Overview

Adaptive Control of Thought—Rational (ACT-R) is a cognitive architecture that primarily focuses on modeling human cognition. It is a theory about how human cognition works and is also a computational system that instantiates this theory. ACT-R is grounded in the belief that cognition can be understood in terms of the coordinated interaction of a set of modules, each of which is responsible for a different aspect of cognition.

History

ACT-R is the latest in a series of cognitive architectures developed by John R. Anderson and his colleagues at Carnegie Mellon University. The first version of ACT was proposed in the late 1970s, with subsequent versions (ACT*, ACT-R) evolving over the next several decades. The "R" in ACT-R stands for "Rational", reflecting the influence of Herbert A. Simon's theories of bounded rationality on Anderson's thinking.

Photograph of a timeline showing the evolution of ACT-R from its inception to the present day.

Theoretical Foundations

ACT-R is based on a number of key theoretical ideas, including:

- Modularity: ACT-R posits that human cognition is organized into a set of modules, each of which is responsible for a different aspect of cognition. These modules include a goal module, a declarative memory module, a procedural memory module, a visual module, and a motor module, among others.

- Production Rules: ACT-R uses production rules to represent procedural knowledge. A production rule is a condition-action pair: if the condition is met, then the action is executed.

- Declarative Knowledge: Declarative knowledge in ACT-R is represented as chunks, which are pieces of information that can be stored in and retrieved from memory.

- Subsymbolic Processing: In addition to its symbolic level of processing, ACT-R also includes a subsymbolic level of processing that captures the probabilistic and timing aspects of cognition.

Architecture

ACT-R's architecture is composed of a set of modules and buffers. Each module is specialized for a different type of cognitive processing, such as visual processing or motor processing. Buffers are used to hold the current contents of each module, and only one chunk of information can be in a buffer at a time.

The central module in ACT-R is the goal module, which guides the system's behavior. The goal module interacts with the other modules through the buffers. For example, the visual module might place a chunk representing a visual object into the visual buffer, and the goal module could then use this information to guide the system's behavior.

Applications

ACT-R has been used to model a wide range of cognitive phenomena, including memory, attention, learning, problem solving, decision making, and language processing. It has also been used to create intelligent tutoring systems, to design user interfaces, and to predict human performance in complex tasks.

Photograph of a computer screen showing an ACT-R model in action.

Criticisms and Controversies

While ACT-R has been influential in cognitive science and related fields, it has also been the subject of criticism. Some critics argue that ACT-R's modular architecture is too rigid and does not adequately capture the flexibility and adaptability of human cognition. Others argue that ACT-R's reliance on production rules is outdated and that more modern computational techniques should be used instead.