In 2016, Prof. Ron Sun led the CLARION (Connectionist Learning with Adaptive Rule Induction Online) project. CLARION is an integrative cognitive architecture used to explain and simulate cognitive-psychological phenomena. It could potentially lead to a unified explanation of psychological phenomena.

CLARION distinguishes between implicit and explicit processes and focuses on capturing the interaction between these two types of processes as a critical feature. Ron Sun's research team developed the system.

Clarion Project

Clarion is a project that looks into the basic structures and workings of the mind. In particular, it looks at how implicit and explicit thinking works together, focusing on bottom-up processes (i.e., from implicit to explicit methods). It also looks at how motivation, thinking, and thinking about thinking work together. The project combines many essential ideas into a single (theoretical and computational) model. The goal is to develop a generic cognitive architecture that describes a wide range of psychological processes in a unified, coherent way. This way, we can explain a wide range of mental phenomena suitably.

The current goal of this project is to do two things: 

(a) create artificial agents that can do cognitive tasks in different areas, and 

(b) figure out how humans think in these same areas. 

Layers of CLARION

The core theory of mind is the first of the CLARION theory's three layers. The fundamental CLARION structures are into several distinct subsystems, each having a dual representational system (implicit versus explicit representations).

Some of its subsystems are:

• The action-centred subsystem,
• the non-action-centred subsystem,
• the motivational subsystem, and
• the meta-cognitive subsystem

The basic theory is in the second layer by computational models, which are more specific than the first level theory but are nonetheless general. The models and simulations of psychological processes and phenomena comprise the third layer. The fundamental idea and broad computational models serve as the foundation for the models of this layer.

CLARION and other kinds of brain structures

• ACT-R separates procedural memory from declarative memory, similarly to how CLARION separates its Action-Centered Subsystem from its Non-Action-Centered Subsystem. But ACT-R doesn't make a clear distinction (based on process or representation) between implicit and explicit processes, which is a crucial assumption of the CLARION theory.
• Soar doesn't distinguish between implicit and explicit thinking or procedural and declarative memory based on how they are represented or used. Instead, it is on the ideas of problem spaces, states, and operators. Different productions suggest different operators and operator preferences to do the job when there is a goal on the stack.
• EPIC uses a way of making things that is similar to ACT- R's. But it doesn't have the difference between implicit and explicit processes, which is an integral part of CLARION.

Conclusion

The following psychological tasks are using CLARION: 

• the serial reaction time task, 
• the artificial grammar learning task, 
• the process control task, 
• a categorical inference task, 
• an alphabetical arithmetic task, and 
• the Tower of Hanoi task 

In contrast to the Tower of Hanoi and alphabetic arithmetic tasks, which are high-level cognitive skill acquisition tasks, the serial reaction time and process control tasks are typical implicit learning tasks (mainly involving implicit reactive routines) (with a significant presence of explicit processes). A complex minefield navigation task requiring complex sequential decision-making has also received much attention. Additionally, work has begun on metacognitive tasks, social simulation tasks, and organizational decision tasks. Furthermore, other applications of cognitive architecture include creativity simulation and investigating the computational basis of consciousness (or artificial consciousness).