Concept Association

Let us begin by supposing that part of our mind contains a huge catalog of concepts. Almost anything can be considered a "concept" for these purposes, such as an image that we're seeing, a name, or a representation of a place. Each of these concepts is connected to some set of other concepts. Once you're thinking about one concept, you're more likely to think of the concepts that it's connected to next. This should be extremely familiar to everyone, because it is fundamental operation of memory. This is why hearing certain music or smelling certain flavors brings back memories which are associated with these things.

Now, this linkage from one concept to another can vary in strength - some concepts can be quite strongly linked, such that thinking of one is almost always immediately followed by thinking of the other, or they can be linked very tenuously such that the connection is barely apparent. Furthermore, at any given time, you have several concepts "in mind." Concepts can come from sensory perception. They can also come from the "short-term memory" or "conscious mind", which can be seen as a list of the last few concepts which have been used.

Here's the key part: the decision about which concept will be fetched next usually doesn't depend on the connections from any single other concept. It depends on the combined connection strength from all the concepts which are in your mind or being presented to you by your senses. Thus, while there might be no single thing which would cause you to remember something, there might be three things which would, if present in your mind all at the same time, add up to enough strength to trigger the memory.

and that's it. That mechanism is enough to explain enormous amounts of extremely complex behavior.

There are some things which are glossed over in the model. The idea of linkage strength is very qualitative; making a numerical model with it would require significant refinements. The model says nothing about which concept is removed when a new one is brought in. Cognitive research implies that it may also be possible for link strength to be negative, but no use of that is made here. Nonetheless, I have found this model to be extremely useful for actually doing the things I want to do, and perhaps you will as well.