A cognitive architecture is basically a theory to approximate the process of cognition. That is,
if you sort information using these processes then you sufficiently mimic cognitive process.
Clarion does so with the basic assumptions that
a) cognition is always means of fulfilling motivation
b) motivation is either the effect of a drive or a goal structure
c) action and knowledge-bank both operate two separate processes, implicit and explicit (which I think are better called intuitive and declarative)
It should be directly stated that this is not literally true. The following subsystems are not associated with specific regions of the brain whose job it is to perform them. These are complex and loosely penned cognitive processes that are not comprehensive but are sufficient to approach cognition. If we better understand our cognitive process then we are more readily able to recognize hang-ups and solutions especially in regard to how we structure our learning.
The illustration looks much more complicated than it actually is. Let’s look at each subsystem.
- The kickstarter of cognition is the Motivational Subsystem. You start with a drive. Cognition itself is specifically designed as an instrument to fulfill this drive. It can range from food, water, or avoiding danger, to autonomy, dominance, recognition, or belonging, etc. We are pretty familiar with this kind of psychological assumption. You may also be motivated by a process created by a goal structure that in turn services a drive. That's fortuitous for us, since we want to use our natural tendencies to procedural ends and like goal-setting anyway.
- All explicit knowledge constantly cycles through Meta-Cognitive Subsystem. Meta-cognition is simply monitoring and direction. At this stage, your brain decides what is appropriate to do. The MCS filters, selects, and regulates information to be acted upon. It reviews progress, reinforces positive processes, and sets goals to solve for unsatisfactory processes.
- The Action Subsystem controls thought/action. It has an explicit and an implicit level that operate simultaneously.
- Non-Action Subsystem maintains your knowledge-bank. It has explicit and implicit resources.
- Explicit: declarative, specific, more accessible.
Explicit knowledge is representable. It can be communicated easily through words, symbols, or named concepts. In a melee context, all knowledge is procedural, a rule for action. I like the term Declarative because it can be and very often is declared.
- Implicit: intuitive, holistic, fundamental, less accessible.
Implicit knowledge is not representable. It is automatic and outside of our verbal grasp. Colloquially we sometimes hear it referred to as Unconscious but in this context that’s extremely confusing. I prefer Intuitive.
We can now discuss the two learning processes suggested by the model and its researchers.
Learning principles to apply to action. The design is placed before the experience. This knowledge is either heard and then internalized through practice in its application OR it is intentionally sought out in experience, then sorted. Anything that is or can be “worked out.” This is the method that is most familiar to us as an educational model. It is on the whole very efficient. A Top-Down approach is weak in two meaningful respects. First, the knowledge is only as good as how well it is understood. A limitation in understanding is an immediate limitation on knowledge and capacity. Second, this knowledge is limited by the capacity of its delivery. If the words fail to sufficiently address the situation then the knowledge fails to sufficiently address the situation.
Trail and error to arrive at unconscious mastery. The experience is placed before the design. This knowledge is necessarily out of reach of our conscious meta-cognition and thus poorly understood if understood at all. In this way it is almost entirely built from experience. The best example of this is grammar. Although we can learn grammar using declarative rules, most of us learn a language at too young an age to use them. We internalize the correct rules of grammar through a huge amount of trial and error. We have an acute intuitive sense for these rules but it is extremely difficult for us to formulate them on demand. Intuitive knowledge is incredibly profound, capable of magnificent complexity, and effortless to put into use—we rarely even notice. However, it is obscenely inefficient. Running yourself through the trails and errors required to assuredly arrive at correct intuitive knowledge is a monstrous task. It is also highly susceptible to misdirection by biases in interest/activity/wrong opinion/etc.
But the beauty of the dichotomy lies in the overlap. These models are not mutually-exclusive. Because they operate simultaneously, a conscientious learner can actively take advantage of both. He may gently guide the progress of his bottom-up learning with focus and mindfulness or later conceptualize the solutions that he intuitively arrives to as if by accident. This is the optimized form of Bottom-Up. He may intentionally internalize his top-down assumptions with practice, allowing a methodology to sink into an unconscious procedure that is more sensitive to exceptions. This is the optimized form of Top-Down.