送交者: insight 于 2005-9-24, 06:50:45:
1) Computation is a mechanical process based on certain algorithms; Strong AI supporters believe that one can simulate the human mind using algorithmic calculations, because they reason that since the human brain works according to all the known physical laws and the mind is just a state of the brain which is computable;
2) However, it is noted that we do not know if the universe is super deterministic, and many physical processes are not amenable to deterministic computer simulation, even though some of them may be considered possible in principle. For example, quantum events and human behaviors can not be deterministically predicted using computer algorithms.
3) The mind, being a mental state of the brain, consists of subconscious and conscious activities:
a) Perception & Emotion (subconscious & instinctual)
b) Imagination & Feelings (consciousness directed)
c) Intuition, abstract mind models, insight (both subconscious and conscious)
d) Deduction & rational and analytical processes based on abstract mind models (consciousness-directed)
4) It may be possible to use algorithms to simulate the subconscious and conscious activities in a), b) and d) with the computer blindly performing mechanical, although adaptively, computation, without any understanding of all the processes. To have understanding, it is necessary for our computers to take intuitive leap to form abstract concepts or mind models of the activities in order to gain insight into them. The key to simulating consciousness is thus to simulate understanding, and this itself requires the simulation of intuition and insight. But can intuition and insight be simulated by computation? So far there is no evidence that supports this claim. Godel’s incomplete theorems suggest otherwise: our mind must take intuitive leap to generate new rules and axioms, otherwise there bound to be propositions or problems we can not prove or solve using algorithmic procedures. In summary, Godel's incompleteness theorems show that it is impossible to use algorithms to simulate intuition and insight, thus understanding. This leads to the the conclusion that mechanical computation alone can not simulate consciousness.
5) The autistic mind is an extreme example of a mechanical mind which can not easily obtain the gist of details, i.e., gaining insight by intuition.
a) Autistic persons are visual thinkers, which explains why they can easily systemize things. However, autistics find it difficult to read other people's minds, meaning that they lack the ability to take the inductive leap to form mind models and concepts which confer intuition. As a result, they can not figure out what is behind the face of a person and have poor sociability . Therefore they have to try very hard to use logic within a closed system or fixed rules to figure out the gist of many details, which normal people with good intuition can easily obtain without paying much attention to the details. To put it another way, autistics can easily see the trees but find it difficult to see the woods, while normal people are the opposite.
b) Verbal thinkers belong to the other side of the spectrum, who may have a good command of languanges and abstract numbers and names, but do not have a good sense of space and patterns, thus not good at remembering visual details and Maths. Verbal thinkers can take the intuitive leap to form mind models and associate seemingly disparate systems easily with the help of languages. Unfortunately, during the process of doing these, they may lose the details that the autistics can not forget.
6) Another example, which shows that understanding can not be mechanical, can be found in language learning and its relationship with consciousness:
a) A child's learning starts from the very primitive brain functions: perception and emotion, which are a sort of action and reaction without much self-consciousness. At this stage, perception involves mainly audio and visual images which can be considered manily mechanical without much understanding. This lays the foundaction for phonology.
b) Gradually the child's self-conciousness develops and starts to take control of his/her mental activities, which means that he can manipulate the audio/visual images at will to generate mental models as imaginations and feelings. As a result, the meanings of things emerge. Thus the child can express himself with sounds meaningfully while running his imagination in his brain. And he begins to aquire the grammar and semantics of a language.
c) At later stages, consciousness further develops, and the child starts to use induction to form abstract mental models and concepts and learn abstract symbols, such as alphabets and words. And these abstract symbols can then be linked to imaginations developed earlier, which are run like virtual reality in computer terminology.
d) Many adults, on the other hand, learn a foreign language in a reverse order, simply because he has a fully developed capabilty to acqiure abstract symbols due to his strong and dominant self-conciousness.