According to the Functions Operating on Mental Models (FOMM) theory of human intelligence articulated in this blog, language is an example of a Procedural Behavioral Model (PMB). A PBM is a hierarchical, nested, recursive, model that is executed in a sequential manner on a conceptual level and a parallel manner on lower levels (e.g., neural). In recent years, research was published that can be interpreted as supporting the idea of language falling into a class of cognition rather than being a unique process.
The traditional view of language was that it is a complex skill with its own circuits dedicated to language processing. More recently, research demonstrated that areas related to language processing are also involved in fine motor skills. Previous research showed a correlation between skill at using tools and having good abilities with language syntax.
The most recent study in this area demonstrated that similar brain areas were activated during the use of tools to perform fine motor tasks and language syntax exercises. Furthermore, they demonstrated that practice with fine motor tasks improved performance on the syntax tasks and vice-versa.
This would be expected from my formulation of language as a PBM. Using a tool to perform a fine motor task would be a PBM. According to my conceptualization of PBM’s, this relationship would not be limited to fine motor skills. Indeed, any complex behavioral pattern is a PBM (such as dancing, jumping rope, exercising). Humans are capable performing a wide array of complex motor behaviors and language is just one of them. Language does have some special features that other PBM’s do not. Specifically, it is a universal cognitive tool that can be used to compress, store, transmit to others, examine, and alter other PBM’s.
I would predict that practicing any complex motor behaviors would improve language syntax ability and vice versa. In other words, tool use and language tasks are just part of a larger class of mental-behavioral models, which I have termed Procedural Behavioral Models. Additionally, I would predict that performing complex motor tasks and simultaneously describing the task verbally would eventually result in the highest gains in both the motor tasks and performance on language syntax tasks.
AI research will need to progress in the development of PBM’s in order to advance further towards AGI. Competitive queuing is one place to start when considering how this could be implemented. CQ is far too simplistic as currently conceived, but it may be a good starting point. Intuitively, I suspect that the human brain queues up multiple PBM’s outside of awareness and may bring in components of other cued PBM’s into the active PBM in real time. It is also obvious that humans, and non-humans, have the ability to execute multiple PBM’s simultaneously (e.g., walking and chewing gum; not to mention all of the non-conscious PBM constantly running in the background like eye saccades, respiration and heart rate). The research showing the association between brain areas used for tool use and language are a serious challenge to theories positing that only the the cortex is important for intelligence (e.g., Jeff Hawkin’s 1000 Brains Theory). This research showed that the basal ganglia, a subcortical region, is used in both tasks.