Musing on Intelligence — Part 4: Brain Architecture


Human brain is a marvelous machine capable to learning, creating, planning, and feeling–in a nutshell, defining a person.  In Part 1, 2 and 3 of this series, I argued that AI development should be holistic, considering the total system not just circuit-level constructs such as neural networks and Deep Learning.  I also suggested that one should look to psychology, neural science and brain studies for clues.  In this blog post I will discuss evidences from brain studies that AI practitioners developing intelligent systems should consider.


Human brain has distinct regions each corresponding to specific functions.  For example, occipital lobe (visual cortex) is the visual processing center; somatosensory cortex is mapped to tactile feedback; somatomotor cortex is mapped to proprioception (knowing where the body parts are); parietal lobe is related touch, temperature; and frontal lobe is related to volunteering movements and planning.   Temporal lobe is related to audio-visual recognition, including language and long-term memory, and cerebellum is responsible for sensorimotor coordination but does not initiate actions.  Medulla oblongata is responsible for autonomous functions such breathing, cardiovascular, and reflex.  Finally, the spinal cord connects the central nervous system (brain) to the peripheral nervous system, much like a cable connecting the computer to its peripherals.  Human brain is a learning machine with functional organization, not just a wet, gelatinous blob.



Furthermore, human brain is hierarchical, with more primitive functions, such as emotion and drives in the inner brain, called the limbic lobe, with the more advanced functions assigned to the outer brain.  Scientists have theorized that the brain hierarchy is the result of evolution from reptiles to mammals to humans. The three parts of the brain are therefore named, reptilian (lizard) brain, the limbic (mammal) brain and the neocortex.


What does the brain architecture tell us?

So far we learned in Part 3 that humans behavior (action) is motivated  by needs, and that there is a hierarchy of needs from basic survival to self-actualization.  In this blog post we learned that human brain is hierarchical and functionally organized.  Could there be a correspondence between the two hierarchies?  Are there lessons to be learned from how the brain’s various functional compartments are arranged?  Clearly there is organization to the human brains, and it may provide valuable clues to AI developers for creating human-like intelligent systems.  In Part 5, I will explore the relationships between biological needs, emotion and drives, and how they tie back to intelligence.

(The above article is solely the expressed opinion of the author and does not necessarily reflect the position of his current and past associations)