The Nature of Reason

Reason involves at least three abilities: correspondence, logic, and retrieval and presentation. Each of these abilities has been mechanically reproduced by computers today, and is thus neither mysterious nor unnatural.

The power of ‘correspondence’ is the ability to recognize that a particular pattern matches many others (and is thus a repeatable pattern). Computers are doing this, for example, when they perform Optical Character Recognition (OCR) on a written page, recognizing a certain specific pattern of ink as a certain universal letter. This means assigning a name to a recurring pattern, and detecting when that pattern arises in our perception.

The power to carry out logical operations is the ability to analyze data in various ways, such as adding or subtracting, or identifying patterns within a larger pattern. Every form of logical operation humans can imagine has been performed by a computer, from Boolean Algebra to Fuzzy Logic—even creative scientific induction—and, like OCR and other recognition routines, we know the mechanical-causal steps that can carry out such functions. They are basically just practical, truth-finding ‘rules’ for extracting information from complex patterns of data, for managing a linguistic train of thought, and for arriving at probable conclusions from an enormous pool of information.

Finally, the power of ‘retrieval and presentation’ is the ability to remember and call up relevant data for analysis from among a vast store of information, such as comparing one thing to another, combining two things to achieve a third predetermined goal, and so on. This is also an ability we have mechanically programmed into computers, running routines called “Creative AI” (Artificial Intelligence), where computers can store, even learn, huge amounts of disconnected data and creatively “guess” what particular item might be relevant to solving a new problem never encountered before, then retrieve it and apply it. Computers have, for instance, learned all on their own how to regulate a power-plant’s cooling system through trial and error and applied logic. A computer has even been programmed to invent its own scientific hypotheses and test them, or use common sense to plan someone’s trip—for instance, taking into account their claustrophobia and their passion for nursing, without even being told that claustrophobics don’t like going through long train tunnels, or that nurses would be interested in a famous medical exhibit.

However, computers have yet to accomplish three other things essential to human reason: a vast processing capacity (the smartest computer that exists today has the equivalent of a mouse, a far cry from what humans have and need), an advanced integrated perception complex (though computers have been trained to “see,” for example, their powers of visual perception are still at the stage of lower animals, and they have not yet reached the complex level of integrating several different sensory systems into the same process of continually constructing a model of their environment), and self-awareness (constructing a model of oneself: see http://pastebin.com/GxHsN9ap). Otherwise, contrary to popular belief, computers can abstract, recognize abstractions in existing things, and talk about them.

For example, a computer can learn how to recognize “circleness” or “letter-k-ness” all on its own. It can then use that knowledge to see where circles or the letter “k” exist in things, distinguish their attributes from others in the same object, then analyze and output the features of circles or the letter “k” that are universal (i.e. repeatable), and reason from those features to other conclusions (such as that a circular object might roll down an incline, or that a “k” might be associated with other letters before and after it to form an actual word rather than a random string). But consciousness of abstract objects requires consciousness of a self that exists in relation to them, and that requires a computer more complex than any yet within our technological grasp.