The Question and Possible Answer
What exactly is an abstraction? Consider that the more we study it, the more the brain appears to be a vast neural net of nodes and connections. Estimates hover around 5 billion nodes, each with thousands of connections. Less understood is how the "mind" uses and organizes these bazillions of neurons and their connections.
The
diagram on the right is a simple example of a Concept Map, with 5 nodes and
4 connections. The animal, carnivore and vegetarian nodes are abstractions,
while the wolf and rabbit nodes are facts. (Actually a fact is just a widely
known and accepted abstraction, since for example the wolf node is not the
wolf itself.) The diagram is from a very educational dissertation
by Rob Kramer. Click the dissertation
link and scroll down a bit to see the original diagram in full context.
Recently my research has dealt with Concept Maps and knowledge structures. While viewing a simple Concept Map example (the one above) it suddenly occurred to me that while most Concept Map nodes are facts, some are purely organizational and quite arbitrary, and are abstractions rather than facts. In fact (pun intended) there are so many abstractions they themselves need organization, and so the mind has major and minor abstractions.
Learning can be defined as "the integration of new knowledge with old". Some learning is the accumulation of more facts into existing knowledge structures, such as learning the new colors mauve and chartreuse, or a new route to your favorite dance hall. Perhaps most learning is this sort - one just adds the new knowledge to existing knowledge structures, making integration easy.
But what if there's nothing to integrate new knowledge with? Ahhhhhh, you have to spot this in the first place. For example, suppose a new driving short cut involves turning right at a traffic light rather than turning right just before the light by taking a dedicated right turn lane that curves right. Hardly anyone does the first - instead they nearly all do the latter, just as the traffic engineers planned. But the former is perfectly legal. It may even save time if the dedicated lane is backed up.
Back to the point of this discussion. If your new short cut has a "sharp right" at an intersection that has a dedicated right turn lane, you could just memorize the short cut, perhaps mentally remark about the novelty, and let it go at that. Or you could recognize the new abstraction of a "sharp right at a dedicated right turn lane intersection", create a new Concept Map node in your brain's storage for this abstraction, connect it to the new route, and then use the new abstraction to your advantage. For example, you might notice one of your wife's routes could be shortened by applying the new abstraction, or when driving in a new town and you miss a dedicated right lane turn, you could save the day by taking a sharp right.
Thus the nature of abstraction is the use of "abstract nodes" when appropriate. Those who are good at abstraction are good at constantly reorganizing or creating new abstract nodes deep in their brain as necessary, plus the associated connections. This is not easy. It probably needs to be cultivated at a young age, such as with music, puzzle solving, penetrating articulate conversations, extensive reading and humor.
Explicit Knowledge Structures
We have seen that the mind contains huge knowledge structures whose nodes are abstract or factual. When one is learning something, they are really integrating a new knowledge structure with their own existing knowledge structures. Wouldn't it be far more efficient if the new knowledge structure was explicit, rather than implied?
Most knowledge is transfered by speech or text. The rest is by experience and observation. After childhood, most "deep" knowledge is transferred by text. In the case where transfer is not easy, if the author took the time to put the meat of the matter into a visual knowledge structure such as a Concept Map, learning would go faster and more accurately. This is because the learner would be exposed directly to the conceptual structure the author felt was most appropriate. There would be far less lost in translation.
Thus what we need is more Explicit Knowledge Structures (EKS) for learning.
EKS need not completely replace text. Rather it would be the clothing that complex text needs to be more easily introduced, the spice that makes the incomprehensible more digestible. EKS would be the form that crucial knowledge was expressed in, just as science uses math and formulas, and musicians use sheet music. EKS is the backbone of complex knowledge, and text (with images) the ribs.
EKS is already in use. Every time a new concept is worked out on a white board with quick diagrams, it's probably a simple Concept Map, especially if connections are used. The Dewey Decimal System is a widely accepted EKS. Everytime a database ER diagram is drawn, it's another type of Concept Map. Even class diagrams and UML fall into this rubric. Then there's the internet, probably the world's largest EKS. It offers a good example of a knowledge structure where connections (hyperlinks) greatly outnumber nodes (web pages), and shows how connections can make all the difference to the effectiveness of a knowledge structure.
Returning to our original question, "What is the nature of abstraction?", we can now embellish the answer a tad. Here goes: Abstraction is the use of implicit or explicit knowledge structures with the liberal presence of appropriate abstract nodes. We must include "implicit", because currently most are. Here implicit means implied, rather than expressly stated.
There is no doubt that the abstractional portion of knowledge is by far the most important part.
Imagine that. For millennia we've been merely implying the essence of knowledge....