Learning

Learning occurs when a person can relate new information to their current knowledge. That is the whole of it. New information on one side, current knowledge on the other, and a connection made between them. When the connection is made, the learner's current knowledge changes. That change is the definition of learning.

The change can take several shapes. Knowledge can grow: a new concept gets added to the network. Knowledge can be corrected: something the learner held wrongly is revised. Knowledge can become more connected: two concepts the learner already held separately are now linked. Knowledge can be reorganized: the learner sees a better structure for what they already knew. Any of these counts. If we call the state of a learner's knowledge before learning K1, and the state after learning K2, then learning means K2 > K1 along at least one dimension.

Notice what this definition rules out. Hearing a fact without being able to attach it to anything already in your head is not learning. It is noise. The words may be repeated back on a test, but the knowledge has not actually changed, because no connection was made. This is why teaching that ignores what the learner already knows fails so reliably, even when the teacher is saying correct things. The information has nowhere to land.

Q. If knowledge is a network of related concepts, what kinds of moves count as learning? Adding a node. Making a new connection. Correcting a node that was wrong. Reorganizing a cluster of nodes into a cleaner shape. All four are forms of K2 > K1. None of them is more "real" than the others. A learner whose knowledge becomes better connected has learned, even if no new fact was added. A learner whose old mistake has been corrected has learned, even if the total count of things they know has stayed the same. The vector matters, not the single-number sum.