It is well known to us that memory is defined to record and recall information. Ostensibly, the computer is able to record the information, which is given from outside, in its memory chips and disks, and this information can be used, transferred and read out. Man can store information in the brain and can recall it. Both memories seem almost the same by simple comparison, but, in fact, the performances of both memories are vastly different.
Generally, memory comprises two portions: One records the information received or sensed mainly from outside. The other portion records the process of the activities of the subject itself automatically. The computer has the former only while human brain can do both. The memorizing process of the computer is passively to execute programs. The things that human brain records, with a translation, are not the received information only, but also, the most important, the course of processing information or the content of the conscious workings of the brain itself. The second portion of memory goes concomitant and uncontrollable, however the first portion can be controlled.
To solve new problems with experiences needs induction and inference. This capacity of human being is based on the memory structure of human brain. The main process of memory during early stage of development of human brain is the establishment and accumulation of fundamental materials or source materials and forms basic blocks and piles of experiences, so-called mechanical memory. And the main process of memory in maturity of human brain is the correlation and reorganization of the experience blocks and piles, namely correlational memory. The brain activities performed by correlational memory result in a mode of human thoughts that has naturally a capacity of induction and inference. The reorganization of experiences forms mans new experiences, so man makes progress. The advantage of man's memory structure is capable of fuzzy discrimination and memory patching, and the disadvantage is oblivion because the old related compositions are dispersed and fade away.
It is not so difficult for man to remember a human face as to remember a foreign word, but it is contrary to a computer. It would rather memorize a telephone directory book of a city. The computer memory is almost completely mechanical and accurate, and the computer can not use its memory to induct and infer things. Although the computer can memorize, it can never possess experiences or it is unable to utilize its experiences, which is the necessity to actuate intelligence.
The computer records all input materials when memorizing. To be more exact, the disks in the computer do not belong to its "brain" totally. The data area in the disks is like a notebook or a bookshelf to man. The data in disks are outside its "brain".
When memorizing, human brain needs just recording the new relations among the fundamental materials and experiences that have already resided in human brain. The fundamental materials are referred to as the mechanical memory of basic physical sensing of human being.
The fundamental materials, in fact, are not so many that an infant can gain them in a few years after he is born enough for his memory in his whole life. From then on, his memory just correlates the materials into blocks, correlates the blocks into piles, re-relates them into new blocks and piles and so on. That makes the complex memory in human brain. The anatomy and analysis on human brain support this explanation. Furthermore, this explains why a man's head does not grow bigger and bigger when gaining more and more memory and experiences, and why human brain consumes so little physical power energy. Correlational memory does not let the size of our brains be proportional to how much we have memorized. If we remembered all we sensed and experienced like an audio and video recorder, either our heads would inflate to explode or each of our heads would grow out as large as a tall building.
Here we get two important points: The first, if we endow a computer with these fundamental materials, the rest questions will be just how to have the computer correlate them. The second, in the process of correlation, it needs neither recording any of the fundamental materials nor re-memorizing the experiences already in the existing blocks and piles. All the blocks and piles can be shared in the memory. In this way, it just memorizes the new "lines" among the blocks and piles and therefore saves a lot of hardware or media of memory. Any correlation can be either strong or weak, however new lines can be regarded as a new block or pile. This is not only significant in intelligentizing the computer, but also revelatory to the education and training of children, melioration of human memory and enhancement of human intelligence.
Animals also have correlational memory, why is their intelligence so low? This is because their memory can only actuate the simplest correlation and infer things at the lowest level. The behaviors of animals are governed by their instinct, therefore they seldom or never pay attention to the correlation among things irrelative to their instinct.
The mechanical memory of the computer is by no means lousy. The capacity of the computer to remember phone books and process database will never surpassed by that of human brain. Moreover, the memory content of the computer can be read out, overwritten and copied. This is the advantage of computer memory over that of human brain, since the memory of human brain can not be read, written and copied.
Human emotion and intelligence are closely correlative to the memory characteristics of our brains, and the conscious workings of our brains represent the memory activities to a remarkable degree. To explore and understand the memory performance of human brain is an important step to actuate AI or AL. Trying to evade this obstacle is bound to fail.
One of the key reasons why the flying advance of computer science and technology brings about so little progress in computer intelligentization is that the way of memorizing process of the computer remains still at its embryonic stage.