This article was last modified on December 5, 2003.

The Tri-Level Hypothesis: Cognitive Science and the Leap of Faith

The field of cognitive science is growing daily, with programs appearing on college campuses worldwide. Various approaches exist, some better than others, and progress is slowly but surely being made. One of these approaches, the tri-level hypothesis, is the personal choice of connectionist Michael R. W. Dawson. This approach breaks the study of cognition down into three distinct levels for easier understanding of each level. The levels are (from simple to complex) the implementational, algorithmic, and computational. I shall discuss each on their own, and follow up with two overall objections to the tri-level hypothesis and cognitive science in general.

Implementational Level

The implementational (or physical) level studies the actual physical makeup of the information processing system. In the case of humans, this would be the brain — whereas in computers, it would be the hardware. For the sake of simplicity, I will only be focusing on the human brain. Neurologists, psychologists, and the like study the brain and its various components. Dawson focuses on aphasia, and how with various experiments scientists were able to locate specific areas of the brain that are responsible for language. Those that focus on the brain and its individual areas aid cognitive science in isolating the sources of higher level functions. In theory, with each area isolated to match its function, it will be easily to replicate in future models and potential attempts at creating artificial intelligence.

There is a lot to be studied yet in this area. While man has been studying the brain for centuries, mysteries still remain that need to be figured out. Of interest to both philosophers and occultists is the pineal gland, for example. Descartes believed this gland to be the gateway to the soul, and many occultists (specifically followers of Crowley and Blavatsky) still contend this gland is a form of “third eye” that can be activated by some means, perhaps trepanation. While I doubt this gland houses the soul or can give us greater esoteric knowledge, the fact is that cognitive scientists will need to fully dissect and analyze this gland in order to understand the brain (and therefore the mind).

Algorithmic Level

The algorithmic level, as the name implies, is the level at which the algorithms and functions of the brain are studied. This level is of particular interest to logicians and mathematicians due to its focus on logic and complex chemical functions. While the implementational level left off at the physical properties of the brain, this level wishes to know how they work. Each gland has a specific function it performs and an algorithm that it runs. Each gland has its limits of what it can do — and the limits are placed there by the functions. Humans are not programmed in the same sense that computers are, but we can think of the brain in this regard.

If we have a personal computer running Microsoft Windows, there are certain things that can and cannot be done with this program. More to the point, the physical components of the computer (the hardware) will only support certain operating systems. The standard PC has no problem carrying out the functions and algorithms of Windows, but an Apple computer could not do this. Humans are fully capable of processing the five senses, but it would be impossible to process a sixth sense without different hardware or a different function. When we learn what functions the brain has, we can (in theory) replicate this mathematically in robots.

A specific example of a function is human grammar. Our brains are designed to understand this, where other brains or information processors are not. When we are asked “Where do you live?” we can respond “I live in Cleveland” without much difficulty, because the grammar necessary is built in to our brains through the glands and functions we were born with. There are limits to this function, though. If we are asked “Live where you do?” we will either be at a loss or we will have to process this through a preliminary function to determine the true question before it is run through the standard function that processed the correct query mentioned before. Some people would place this “sorting” feature on the third level, but this remains in contention and in my mind seems far easier to comprehend as an extra program than as a part of the third level. This issue will be addressed more in my objections.

Semantic Level

The computational level (which I prefer to call the semantic level) is where consciousness takes place, meaning is imparted, and the qualities we tend to term “human” are found. Values and morals are also found on this level. The ideas of meaning and humanity will be briefly discussed.

To explain the idea of meaning, I will call upon the oft-cited Searlean Chinese Room. John Searle speaks of a room where Chinese questions are given to a man who does not know Chinese. He uses a guide to match the question with a pre-written answer and passes on this answer. While this is information processing (input/output), it is utterly devoid of meaning. This is what separates humans from machines — our ability to give and receive meaning. Whereas computers (or the man in the Chinese Room) do not know what they are actually processing, a Chinese man in this scenario would. At any given time, we can stop and ask ourselves why we are doing what we are doing.

The semantic level is also where all abstract thinking takes place, including the idea of beliefs. If we believe something, it is not simply due to an instinct or automated response from a program. There is some deeper thought involved and an abstract conviction that can only be achieved through consciousness. On a related note, much of religion is largely symbolic — actions and signs with meanings that are not superficially available to the onlooker. How can we expect to truly “know” or “believe” that the body of Christ is transubstantiated from simple bread when it clearly appears to remain bread? Yet, if we are Roman Catholic, we believe this.

The Reductionist / Materialist / Etc. Argument

Various groups of cognitive scientists and philosophers object to the tri-level hypothesis. Some of these groups are the behaviorists, materialists, physicalists, functionalists, and the reductionists, to name but a few. While their arguments vary depending on their various angles and agendas, the general consensus of these groups is the same: the third level (cognition) can be explained in terms of the second level (functions and algorithms).

While we see ourselves as thinking beings, it is possible to conceive of the human race as merely advanced automatons. Meaning can be given through the aid of a program. It would cease to be real meaning, but we would not know this. This belief is supported on at least two grounds (again, depending on your specific agenda). First, that everything is “cause and effect” or determined. Behaviorists have shown that a desired response can be achieved from known stimuli in lesser creatures, and in some cases man. We are also subject to the causal laws of nature, which affect our brain as much as any other physical object. In this case, the “meaning” would just be a response to past stimuli — not unlike fearing heights in your adult life because of a nasty spill from a bunk bed as a child.

Second, drawing from the conclusions of the first point, is the reliance on memory in making future decisions. While we may be able to consciously make choices and adhere to beliefs, much of what we do is clearly a response to past events or a pattern we have been a part of. Some stingy behaviorists will even use the word “because” to catch you in their trap. If you answer a “why” question with “because”, you are admitting there was a cause for your action. You may even have chosen to act contrary to your routine BECAUSE it is expected of you to act a certain way. But you would still be acting within the limits of your program, which tells you to be contrary on this occasion.

The objection of the third level is becoming increasingly valid as more of the brain is studied and the functions become cataloged. A specific function related to what we know as consciousness has not been found, however.

The Leap of Faith (Emergence Theory)

Any theory of cognition, regardless of whether or not they utilize the tri-level hypothesis, has as one of its components some version of emergence theory, or the belief that the mind “arises” through the brain by its sheer complexity. Some definitions of cognition are coherent in that they see the brain as a program running on the brain and entirely explainable in physical terms. Others, specifically those generally believed by the average person, see cognition as a higher state entirely — coming either from God or through some form of magic. This view is especially prevalent in those who equate the mind with the soul.

While I can not say with any authority that this belief is wrong, I can say for certain there is currently no logical or scientific knowledge to support it. Research shows the mind to be directly connected to the brain (for example, we are fairly certain it ceases to exist when the brain dies). To believe otherwise is a strictly religious and non-scientific approach.

In the nineteenth century, the great Dane Soren Kierkegaard spoke of the “leap of faith” in regards to Christianity. He was aware that the idea of Christ was absurd and in complete contradiction to reason [in my opinion, Christ can be found with reason, but this is another issue entirely] but still remained true. The leap of faith was the abandonment of reason in favor of subjective truth, and therefore a personal Christ.

This leap of faith is today the only avenue or defense for the “mind as soul” viewpoint. As much as we continue to confront the American public with science, they refuse to give up on their convictions. This is much to their credit. However, in the terms of a cognitive science, this view must be thrown out and a more serious approach must be taken — eliminating the need for a third level entirely.


The tri-level hypothesis is not without its flaws and objections, but it remains a valuable tool to aid in the study of cognitive science (and cognition in general). Separating the three levels allows those interested in cognition to focus on the level they are most interested in, while simultaneously allowing those not specially interested in cognition (like computer scientists) to provide valuable insight from their specific point of view.

Dawson has his critics, and he certainly has his biases, but his approach is quite possibly the best one out there today for the budding cognitive scientist. I predict within our lifetimes we will see just how right (or wrong) he was.

Also try another article under Philosophical
or another one of the writings of Gavin.

One Response to “The Tri-Level Hypothesis: Cognitive Science and the Leap of Faith”

  1. The Framing Business » Where Is My Mind? Says:

    […] The following response was written May 2, 2007 (which some changes made for clarification purposes). For more information on the Tri-Level Hypothesis (from students who probably didn’t fully understand it), check out James Skemp’s article on it. Or perhaps read my own version, which is quite similar to this essay. (Really, if you just read that one, you may not need to read this one at all.) […]

Leave a Reply