Conversation net
Includes responding thoughts and their responses.
Hypothesis
High-level cognitive functions are linguistical in our brains.
High-level concepts and relationships between them exist linguistically in our brains, and cognitive functions based on these concepts and relationships are also encoded in sentences-like linguistic memories.
Our brains can
a) store models of the world in the sentences like linguistic memory. E.g. "Deers come to this spot when there is a drought." and
b) Construct new knowledge/predictions by constructing new sentences following syntax rules E.g. "there is a drought now, if we go to this spot we might find deers."
High-level human cognitive functions are the enterprise of our braining employing these two faculties. We don't have dedicated circuitries for each model expressed in linguistic memory, we just need the basic circuitries for language processing.
Note that this hypothesis is different from linguistic determinism.
Responds to
Learning and models in our brains are hierarchical.
It's hard to imagine higher-level cognitive faculty without some form of hierarchical information processing. ATB proposed that such a hierarchy corresponds to the hierarchy of objects in the real world. This might be a bit too speculative. Columns learn from prediction errors. They can predict raw sensory inputs; they can also predict signals by other columns produced from sensory inputs. Thus, learning can happen when there are raw sensory input prediction errors as well as when there are other column signal prediction errors. Learning in columns can easily be hierarchical - naturally, models (or knowledge) learned from them are hierarchical. That being said, there is no reason to believe that the hierarchy is a neat pyramid with clear-cut division between layers. Any column can learn from any other columns as long as their signals are useful. It's just that learning, and thus models, can happen orders away from raw input signals.