Process of Attaining Knowledge

Attaining Knowledge: Why throwing a baseball would not be possible for a neural network.

Learning how to throw a baseball occurs in stages. First, someone, usually a parent instructs a young child on the basic process of throwing the ball, before the child is scheduled to participate in a game. The child is first shown the basic motion as to how to curve his or her body when tossing the baseball, to get the ball across the field with some measure of force. Then, when the child gains a sense of physical confidence in the simple act of throwing, the parent shows the child how to hold the ball in such a fashion that the ball travels farther, with a sharper degree of accuracy.

The physical motion is only the most rudimentary aspect of throwing a baseball. Gradually, over time, the child enters into a real game with similarly skilled players, usually in elementary school or a Little League game. Through experience and more skilled coaching, the child learns that throwing a ball from first base to the pitcher's mound, or actually pitching the ball to a batter requires a slightly different level of speed, skill, and aim. If one is a pitcher, one learns different kinds of pitches. As one grows even older and becomes personally familiar with different individuals in a particular league, one grows to know different player's strengths and weaknesses regarding certain pitches. A young player watches the games of other young people, as well as professional athletes, and learns from their unique playing styles and various other individual idiosyncrasies of their methods of playing the same game.

Learning to throw a baseball does not end with mere kinesthetic participation or observation of others. One's own body develops simultaneously as one learns. One develops stronger muscles and bones, and a higher stature. This usually aids in throwing a ball with greater force and dexterity. Also, a developing baseball player gets to know his or her own body and have more physical confidence, the player's skill level similarly increases.

Neural networks in computing are based on the processing and memory abstraction of human information processing, paralleling the knowledge acquisition of the human brain. (Smith, 1996) In contrast to an expert system that merely uses preset notions of rules and data to produce a decision or recommendation, neural networks, on the other hand, attempt to simulate the human brain by collecting and processing data for the purpose of remembering or learning past actions. For instance, when one realizes through experience alone that a ball thrown over rather than underhand as a child has greater force, one engages in a dialogue with one's body and environment, judging the outer stimulus and modifying one's process in a way similar to a neural network, as opposed to an expert system which always follows the rules of the game without deviation, much like following a batting line up's sequence, as opposed to actually swinging a bat.

Yet although neural networks may engage in such observed discrimination similar to that of a human brain engaged in a lesson about throwing a ball, and the network's learning process may engage in observation similar to the human brain's process of gleaning information from watching or playing a game, real brains have a greater complexity not simply in terms of the functions they can perform, but also because they are connected to physical operating systems in the form of bodies.