Consider a memory, it forms as a picture in the mind, perhaps a moving picture. Now that memory does not float around as a disembodied spirit, it is the result of the workings of the body as they manifest in what we call "the theatre of the mind" or consciousness. So a picture implies a process that makes that picture. The implied process does not have to be the real one(s).
One aspect of cognition then is to go from the process of the brain to the picture in the mind. Or: Process -> Picture.
Now consider walking into a room, items are scattered about that were in order the last time you walked into the room. They did not leap up and throw themselves around, so the picture, or more specifically the difference between two pictures, implies a process that connect the two. Or: Picture -> Process -> Picture.
But how do we connect the two pictures?
One way is by the evolved mechanisms of the mind, as they have been developed in a particular person by that person's experiences. There are more specific, more accurate, and therefore more opaque ways of making this point more exactly, but that's not necessary. All that is necessary is to show that some things that the mind can do, while they are conditioned by social interaction, are not defined by it. So we have a sense of balance, and both its strengths and flaws are shaped by the hard lumps of experience.
Another way is by far more socially shaped mechanisms, which, while they have innate senses as their parts, are far more determined by social reality in experience. Language is an example. Expose a child to language, and the child will form an idiolect of that language. But the language acquired is so woven with the feedback from society that it is not based on some physical externality, but on an externality which is social.
There is no clean division: our experience with that which is physically external, and personally genetic is woven in with social reality so tightly that other than in clear cases, it is almost arbitrary to divide the two. We feel we are falling, but it is our social context that tells us how good or bad that is, and the social context in its turn, rests on how a group of people have mediated externalities beyond their control.
However, while we see or feel flashes of a picture of an event, of an idea, of a person, we experience the world through the arrow of time: as a sequence of moments, and this is where the process -> picture model of the world becomes woven in the mind. The time honored example of this is that of a movie: what starts out as a set of discrete images, either on film or as digital patterns, when run one after the other fast enough, creates the illusion of movement on the screen. Nothing on the movie screen is, in fact, moving. We can even be aware of both process to picture truths at once, reminded that this is a projection, even if we still see the illusion of movement.
So one one hand, there is a real evolutionary win in seeing pictures, not just as they happen, but remembering, but that win is created by the ability to correlate one picture to the next one, or to the last one. See a track, know an animal made it, and thus have some sense of how long ago it was made, and how far away it might be and in which direction.
But to string that together requires not just a string of exact visual memories, but the ability to relate a perceived image, to a mental process. A spoken word is not compared to an exact recording in the mind, and rejected if it isn't an exact match, but to a contour of the word. To interpret a picture requires imputed process at work on the picture.
So picture implies process, and process implies picture, and the brain shuttles back and forth between pictures of what is, is sensed, or what is expected to be sensed, and different ways of connecting picture to process: a model.
That preface out of the way, that picture <-> process is a model, it's time to get to something a bit more original, since of course what I just outlined is a gloss on Lakoff-Johnson Doctrine of incorporation, Chomskian generative grammars, Kuhnian Paradigmatic structure, and basic evolutionary ontology - it's hardly new, and hardly original in any way to me.
However within the understanding of how picture <-> process works, there is a great deal of chaos and confusion, in part, because the people who think about thinking are remorselessly adverse to pinning their models on evolutionary ideas about cognition, because, in no small part, the most popular versions were thinly veiled anglo-capitalist propaganda masquerading as science. That would be Richard Dawkins et als. So while the philosophy of mind remaind relentlessly locked in myth and medievalism, it was, in no small part, linked to a reaction against appeal to the authority of grants and useful idiocy. On this more in some other place at some other time, because debunking useful idiots is not really important, the graveyard will make all the converts it needs in time.
Back to connecting picture and process. How do we do it?
The technical answer would track how different parts of the brain chew on problems, message other parts, receive messages, and create an attractor of thought. However, it would be virtually incomprehensible to most readers, and also inaccurate, since our state of knowledge is far from complete. Something can be more exact, and wrong at the same time.
It would also be useless even if accurate, because of the very problem of ontology, existence, mentioned above: it would take too long to figure out what anything meant using it. So part of what we have to do is not merely have a model of how process relates to picture, but we are interested in forms of process <-> picture that have short cuts, ways of getting to a close enough to useful answer faster than reality gets to the exact one. 3.14 is a good enough estimation of pi for most uses, and pi, is, in fact, a process as well.
Don't believe that one? Take a piece of paper, measure out lines spaced the width of a topic apart, stand up, and drop spinning toothpicks on the paper. Over time, the ratio of toothpicks touching a line, will be pi. Pi is not just a number, it is a process.
You can see from this that the "model" that connects picture and process does not have to be closely related to the actual process, or the actual external thing that is sensed. Now can often beat right. It is better to jump at three shadows, than to be eaten by one tiger.
So the summary of this first part.
We remember pictures.
A picture implies at least two processes.
One is what the picture is a picture of, and the other is how that picture was produced. This means a picture is a process.
The way to connect a picture to a process is a model.
Because of the ontological truth that the universe is bigger and faster than we are, but our concerns are much smaller, people gravitate not to the most accurate models, but to the ones that get the right result fast enough, often enough.
This means that models do not have to be "best."
But what kinds of models do we use?