We all aspire to possess knowledge, but really, what is knowledge? If you are reading this, it’s safe to assume that you’re not satisfied with the dictionary definition of “facts acquired by a person through experience.” How exactly does a person “acquire” facts? Facts are not physical objects, so they are not acquired in the same way that, say, a book is acquired. Furthermore, computers are capable of acquiring vast amounts of information, yet no one is ready to claim this information as “knowledge.” Surely something is missing in this definition.
What Is Knowledge?
As I see it, knowledge manifests from relational maps in the nervous system, mainly within the brain. Neural networks serve as the physical substrate for these relational maps, and physical stimuli shape their configuration. The informational relationships between brain regions and stimuli give rise to experience, of which knowledge is a particular state. This interpretation is adapted from a scholarly article on the mathematics of qualia.
Ideally, one’s relational map should accurately reflect reality. The amount of reality accurately reflected in one’s relational map is a measure of their knowledge. Stimuli shape the configuration of the map, but without some kind of validation process, the information contained in stimuli may be miscategorized or may even fail to be categorized at all, resulting in a map that deviates from reality either by falsehood or omission. Such information would not constitute knowledge according to our definition.
Let’s consider two stages in the validation of stimuli (and thus in the acquisition of knowledge).
Method #1: Logos, Pathos, and Ethos
Logic/reason, emotion/intuition, and authority/faith are the three conventional tools of persuasion/validation. One should always employ these faculties while assessing the validity of information.
Method #2: Cross Referencing
Every person, young and old, has an existing body of knowledge with which to cross reference any new stimuli. This stage of the validation process is crucial for building one’s relational map. A relational map requires every piece of information to be related to at least one other. If new information is not cross referenced with existing information, it may not be properly categorized and may introduce inconsistencies. The result would be an inaccurate understanding of reality. The more stimuli are cross referenced with an existing body of knowledge, the more accurately the relational map will reflect reality (unless the existing body of knowledge already deviates from reality).
It was previously claimed that “the amount of reality accurately reflected in one’s relational map is a measure of their knowledge.” Let’s broaden the scope now to encompass knowledge of things other than reality. After all, one can have knowledge of, e.g., a fictional world invented for a TV series.
To capture the possibility of possessing knowledge of a particular thing, whether real or imaginary, consider the following definition of knowledge.
The property of a relational map to accurately reflect information associated with a particular fact or skill.
The definition above sufficiently answers the question of what it means to possess knowledge of something. It also more broadly answers the question of what knowledge is possessed by, attributing it to a relational map rather than a person. After all, animals surely possess knowledge, too.
However, something still is missing because we do not yet know how to examine the question of whether a computer possesses knowledge. Does the information on a computer hard drive amount to a relational map? Well, that information is generally stored in a hierarchical fashion (with folders and subfolders), so it is at least organized. But the relationship between individual files is not known by the machine, and it has no way to relate the information within individual files to anything existing outside of that file. When software is run to open those files, the computer may experience the display of an image or the execution of a script, but those experiences have no meaning to it because it lacks a broader relational network to interpret that meaning within the context of some other experience. So it seems that computers do not possess the relational mapping necessary to possess knowledge.
This discussion of computers, however, gives insight on how to broaden our definition beyond its reliance on the term “relational map.” Now we can adjust as follows.
The property of a system to accurately reflect information associated with a particular fact or skill and to interpret that information within some context that gives it meaning.