Computer models are very good at producing answers to questions based on logic and mathematics. A computer can easily beat even the most experienced of chess players or mathematicians, but whether or not this gives them human intelligence is debateable. Early cognitive scientists argue that computers have human intelligence if they can produce the same answers as humans, and how they do it is irrelevant. Churchland and Churchland, 1950 states that ‘there is a finite set of operations that can be applied to a given input, and then applied again and again to the successive results of such applications, to yield the functions output in finite time’. Many tasks that require human intelligence can be broken down in this way and completed by a computer. The end result will be the same, but the steps taken will be very different to human cognition.
Another example of computer modelling representing human cognition is the Turing test, proposed by Turing in 1950 (as cited in Fancher). He quotes “ What will happen when a machine takes the part of (the man or the woman) in this game? Will the interrogator decide wrongly as often when they game is played like this as he does when the game is played between a man and a woman” (as cited in Fancher p 463) Turing developed a program that could respond to human conversation. If a human was fooled into thinking they were actually conversing with another human being rather than a computer, then the computer was said to have human intelligence. This was proved to be successful, because many participants really did believe they were talking to a human, so on these grounds the computer does possess human intelligence. Computers can definitely work out complex tasks that would require intelligence if carried out by a human, but the processes involved are very different.
Dreyfus (1972) states that how a computer processes information is just as important in deciding if they have intelligence as the result the computer produces. The processes involved in reaching an answer to a problem are very different, so on these grounds computer modelling does not represent human cognition. Dreyfus believed that computers couldn’t be seen as having human intelligence until they can process things in the same way, and therefore must also possess background knowledge. Searle, (as cited in scientific American, 1990) believed that computers can possess intelligence capable of complex tasks such as playing chess, but did not believe that a computer will ever be able to possess understanding and empathy. He demonstrated this with his famous Chinese room experiment. The idea is that if you put a man in a room with a book of rules explaining Chinese grammar, the man will be able to write responses to a Chinese sentence. His point was that although the man will be able to write a correct response, he will have no idea about what his response meant. He believes that computer modelling is similar to this, and that it has syntax but no semantics, but therefore can never represent human cognition. Computer hardware can never represent a human brain due to the biological make up of the brain, and therefore can never possess intentionality. The most notable response to this claim comes from modern day cognitive scientists, who state that the lack of intentionality is due to the fact that computers are not yet developed sufficiently. They believe that if a computer can be developed that can process in parallel rather than serial, intentionality will be possessed. Computers currently process information serially in a step-by-step fashion. A computation is a result of the last computation, which combatively builds towards the end result. The brain is a parallel processing system, which means everything is interconnected and works together. This makes processing quicker, more tolerant to faults, and generally more effective. Cognitive scientists believe that if they develop a parallel processing system it will also possess intentionality and semantics, and will therefore have human intelligence. Dreyfus’ claim that the way a computer processes information obviously had an impact on the cognitive scientists because they previously believed that it was only the end result that mattered, and they way a computer processes is irrelevant. They believe it is only a matter of time until a parallel processing computer is developed, but even if a computer can possess intentionality I am dubious that they will ever be able to be phemonological.
At present I do not believe that computer modelling is an accurate representation of human intelligence. I do agree that they are very effective when it comes to solving complex problems, many of which even the most intelligent human could not solve. Computers play a huge part in our everyday lives, and without them many things would be very different. However, computer modelling and human cognition are two very different forms of intelligence, which suggests that computer modelling is not very useful in understanding human cognition. At present the only real similarities are the results they produce, but I believe human intelligence is more than just number crunching. Until a computer can process information as a human, I don’t believe it can be used as an explanation for human cognition. If a computer is developed that can process in parallel more of a comparison can be made, but even then I don’t believe it will be an entirely accurate representation of human cognition. Even if a computer is developed that can possess intentionality and semantics, I don’t believe they will ever be able to show compassion or phenomenology, and I believe these are vital characteristics about being human. Experiments on human and computer intelligence demonstrate just how complex human intelligence and processing really is, and will require more than a computer program to understand. Cognitive science has been based upon computer modelling, and although I believe humans and computers are very different, I don’t think the whole of computer science should be discredited. I disagree that computer models are an accurate representation of human cognition, but I still believe there are many valid arguments made by cognitive science. At present there is no alternative explanation for cognitive science, and computer modelling is as close as we can get so far, so in this respect I think it is useful. Cognitive science still remains the most plausible explanation for reasoning and behaviour in both humans and animals, and with technological advancements I believe our understanding of human cognition will be improved greatly.
Bibliography
Dreyfus, H.L. (1972) What computers can’t do: A SOMETHING of artificial reason. Harper and Row, New York.
Charnaik, E. and McDermott, D. (1985) Introduction to artificial intelligence. Addison-Wesly publishing company inc.
Kirsh, D (1992) Foundations of artificial intelligence. MIT press.
Fancher, Raymond E. (1996) Pioneers of Psychology, 3rd Ed. W.W Norton and Company Inc, NY.
Searle, J.R. Is the brains mind a computer program? Scientific American, January 1990.
Churchland, P.M and Churchland, P.S. Could a machine think? Scientific American, January 1990.