Neurobiology: Like LLMs, the Brain Isn't Predicting

There is a recent article in Tufts Now, AI Chatbots Work by Predicting the Next Word. So Do Our Brains. Is There a Connection?, where a professor stated that "A large language model is trained with very, very large bodies of text.

It’s trained to predict upcoming words. In doing so, it picks up the statistical contingencies, the very complex higher-order relationships between words.

Our brains are constantly predicting the next word, too, and our brains are very good at picking up statistical contingencies and very complex higher order relationships between words.

So, in one sense, you could say the reason why these large language models are so successful is because they capitalize on something that our brains do all the time.

But just because these chatbots can predict the next word and perform so well doesn’t necessarily mean that the precise mechanisms by which they do that are the same as those used by the human brain."

So, what might those mechanisms be, or where are words in the brain emanating statistically? If LLMs are already predicting like the brain, should LLMs not be sentient, or have a form of it already?

When it is said that the brain generates predictions, how does that happen? Is it the neurobiology of the brain, with tissue grooves, elevations, or blood vessels?

Because if the brain is predicting, is it reduced to certain parts or all?

It may be said that the brain is used in place of the mind, then how does the mind generate predictions? Are the components of the mind the same as the components of the brain?

There is a recent file by NOVA PBS, Your Brain: Perception Deception, where it was tad mentioned that the brain has simulations of the world or creates illusions of it and makes guesses.

So, the brain creates an illusion of an external event, then that event causes depression or anxiety. Are those real or illusions as well?

Also, the simulation of the world that the brain has, how come it is fairly accurate for survival if it wasn't actually processed towards near accuracy?

There are agreements about several missing parts in understanding the brain. However, there could also be descriptions that paint a partial picture.

Whatever is generally described as sensations and perceptions about the brain are more directly about the mind.

It is the mind and its components that help to experience the world. The mind has a structure; it has its functions and components. The mind is where all memory, emotions, feelings, and reactions are based.

There are observations of the mind with labels that have almost become the default for how the mind works, ignoring that the mind has its mechanisms, which exceed those labels.

The mind is defined by its mechanisms, not the observed labels. There is no difference between how the mind sets what is labeled as working memory and predictions.

Working memory to hold things in mind—temporarily, means its relay and retrieval across mind locations, which is the same way the transport for prediction—then its confirmation or correction—goes.

This means what should be sought is what transports across mind locations and what those locations are, not just what they are observed to do. Conceptually, the mind has quantities and properties.

It is the relay of quantities to acquire properties, across mind locations that determine what is experienced, internally or externally, in any instance.

Quantities and properties have features that they use to decide for all experiences.

Feature image: fMRI Image of Preteen Brain. Credit: NIH Image Gallery