An Alternative Means To Intelligence Essay, Research Paper

Part 1: The Issue

Through cognitive science, computer science, and psychology there has been an underlying question as to what qualifies for intelligent action. Allen Newell and Herbert A. Simon have proposed that a physical symbol system has the necessary and sufficient means for intelligent action. This is a view shared by many other notable figures from a variety of disciplines.

What I would like to do in this essay is present an alternative means to attribute intelligent action. I will try to show that there are limitations to the physical symbol system, and that something is missing in the theory.

Part 2: Method and Presuppositions

In order to show that the physical symbol is not the only means for intelligent action, I am going to attempt to give examples of alternative methods. I will also point out where I feel that Newell and Simon’s theory is missing a piece of the puzzle. First I will state the theory of the physical symbol system. I will then give what I feel are appropriate criticisms of the theory. Finally I will show that there are alternative means for ascribing intelligent actions. I presuppose what is meant by intelligent action. This is the underlying question and if this is not already understood then I do not believe that we should be discussing a means for describing it. I will also presuppose what qualitative laws are and how they are used in science.

Part 3: The Text’s Argument

Newell and Simon believe that symbols and physical symbol systems are fundamental in explaining intelligent action. In order to understand what a physical symbol system is one must first understand what symbols are. According to Newell and Simon symbols lie at the root of intelligent action. (Newell & Simon, 83). They believe, as do most people, that computers process information in the service of ends. The intelligence of a computer system is measured by it’s ability to reach the ends. To reach these ends a computer system must rely upon a key requirement, the ability to store and manipulate symbols.

Newell and Simon now propose a complete definition of a physical symbol system. It is first explained that the system obeys the laws of physics and are made of engineered components. It is also noted that the system is not restricted to human symbol systems. Newell and Simon state the a physical symbol system is comprised of a set a entities, which are called symbols. These symbols are physical patterns which make up other entities called expressions.(Newell & Simon, 85). A system also contains what is called, processes or rules. These processes act on the expressions to produce other expressions. The processes can be broken down to four categories; the process of creation, modification, reproduction, and destruction. To pull this together, a physical symbol system contains two parts, symbols and expressions, and processes that act upon the symbols and expressions.

Newell and Simon give five rules that designate a physical symbol system. First is that a symbol can represent any expression whatsoever. Second is the designation of expressions to cover all processes that a machine can perform. Third, there are processes for creating and modifying any expression. Forth, once the expressions are created they will continue to exits until they are modified or removed. There final rule is that the number of expressions contained in a system is limitless. (Newell & Simon, 87).

It is from this that Newell and Simon are able to formulate their hypothesis which states, “A physical symbol system has the necessary and sufficient means for general intelligent action.”(Newell & Simon, 87). They are quick to point out that the hypothesis is a law a qualitative structure. It is a means to produce a class of systems capable ascribing intelligent action.

Part 4: Analysis of the Text’s Argument

I would like to first propose some criticisms to the physical symbol system. My first criticism would be of Newell and Simon’s hypothesis. Their use the term “necessary” I feel is wrong. They take it to mean that any system that exhibits general intelligence will turn out to be a physical symbol system. I believe that it is unfair to make this claim. There are two plausible theories that I am aware of that do not use symbols, governed in accordance with rules or processes, that rely on descreate-state machines, to ascribe general intelligence.

The first of these theories is the theory of dynamical systems. Dynamical systems are not symbol manipulators. They do not use programs to arrive at ends but instead use deferential equations. They do not operate by sequential steps but instead operate in “real” time. I believe that dynamical systems (although they not without their own problems) are at least an equally plausible theory for describing intelligent action. What I would like to make clear here is that, I do not believe that dynamical systems are necessarily better than physical symbol systems, but that Newell and Simon might not be correct in saying that all intelligent action is explained by their theory alone.

The second theory I wish to propose in opposition to Newell and Simon is the connectionist theory. This theory I feel much more strongly about and is very different then a dynamical system. If we are going to attempt to describe human like intelligent action or just intelligent action in general, I believe that it is much more realistic and logical to create a system that mimics brains. A connectionist theory is based on how the brain works.

Newell and Simon have used a digital frame work, such as a digital computer, to rest their theory on. The connectionist theory does not do this. It uses the architecture of the brain to manipulate procedures and algorithms. Where Newell and Simon use a digital method a connectionist system uses a quantitative one. This is the part of the theory that I believe Newell and Simon are missing. They claim that the architecture does not matter, I believe that it does. If you are going to create human like intelligence, it seems only logical to me that it should be created with human like architecture and not a digital computer. I argue this partly from the fact that a connectionist network is capable of much more than a digital computer is, learning for example.

There are objections that can be raised against what I have just stated. Many people believe that it is totally irrelevant as to what type of architecture computation is performed on. They believe that computation and intelligence can be arrived at by any means as long as an end is met. I feel that this is not the case. Yes, it is true that computers arrive at their desired ends, but this is not human like intelligence. Connectionists networks also arrive at their desired ends but in a much more human like way. They follow the architecture of the brain. This pattern has lead to concepts that the digital framework cannot achieve, such as human like learning and a strong reliance upon it’s environment.

Part 5: Conclusion

I do not believe that the argument I have given for the accuracy against the physical symbol system is fully complete. What I do claim, however, is that I have shown that there are weakness in the theory of physical symbol systems. Overall, I believe that, to say anything that displays intelligent action must be a physical symbol system, such as one described by Newell and Simon, is not fully justified. This being because of the examples stated above.421

Part 6: References

Newell, Allen & Simon, Herbert. “Computer Science as Empirical Inquiry: Symbols and Search.” In J. Hougeland (Ed.), Mind Design II (pp. 81-95). Cambridge, Massachusetts: The MIT Press, 1997.