This sounds like this post I made a little bit ago. Am I misreading that?
Something I’ve been thinking about with the way learning “happens so easily” is that we start learning the features and behaviors of things as we are learning about object shapes as infants. We don’t remember acquiring any of that, but it is what everything else was built on. As adults it is easy to forget the brain started from scratch. No objects, no behaviors, features, etc. learning everything about the world in parallel and making associations between very different models for different senses.
I’m assuming that all columns get input from more than one sense to create the complete model of the world it contains. Those senses could form parallel structures across columns that are associated together in the columns, allowing each column access to the complete model in all the modalities it’s working with through internal connections. I’m guessing that would be happening in ‘mini-columns’ inside the cortical columns. Synesthesia would be caused by abnormalities in that structure.
This is a common assumption, but it isn’t quite right. Each cortical column receives direct sensory input from only one patch of one sensory organ. There is no column that gets direct sensory input, for example, from both vision and touch. I think it’s also worth stating that the models inside columns are never directly transferred between columns. There is no data transfer of the internal models between columns in the way software modules pass structured data or objects between functions.
That said, there is a lot of information being communicated between columns, some of which we’re very confident about, and some of which is still at the edge of what Thousand Brains Theory can currently explain with high confidence. Examples include:
voting
associative connections being formed between, for example, a model of a chair in a touch column and a model of a chair in a visual column
hierarchical connections building compositional objects
long-range broadcast signals for learning causality
It’s easy to fall into the trap of thinking that the holistic experience you have of the world means there must be a single unified model that fuses everything you know about the chair across all senses. But we strongly believe that this is not the case. We think your perception of stable reality, where your senses seem fused together appropriately, is your awareness of the consensus reached through voting and associative linking.
I wanted to make sure that assumption was off because it was a new one and it felt off. I had previously thought it was as you described for the most part because of the modularity. I got off course. In thinking about it since then, behaviors seem like they have to be “happening higher up” in perception, rather than in sensory processing.
Being wrong in the voting would probably be a learning experience for a column as well.
