Summary
This book tried to explain nicely ‘How the mind works?', in a lucid approach. According to the book, the mind is a collection of smaller processes. He calls these processes ‘Agents'.
These agents acts like members of a society. Each has it’s own job to do, each interacts with its neighbor frequently but has some connection with agents at distant. There is a hierarchy like city administrations, where agent at upper level gets orders from above and gets the job done via the subordinate agents. Like for building agent it might use other smaller agents to build a tower using block. But the builder agent itself might be a subpart of higher level agents like Play agent.
Figure: Hierarchy of agents.
The author also discusses about common sense. As adults we regard many things as “common sense", but
Common sense is not a simple thing. Instead, it is an immense society of hardearned practical ideas – of multitudes of life-learned rules and exceptions dispositions and tendencies balances and checks.
Chapter 3 mentions, that not only hierarchical structure are present in this society. There are heterarchies of agents present where some agents are cross connected like rings and loops.
B-Brains: This section in chapter 6 presents layers of abstraction of some agents where one part ( A-Brain) interacts with outer world or experiment with the body. But another part called B-Brain only interact with A-Brain. It doesn’t know anything about the outer world nor about A-brain’s objectives. But it can only control and experiment with A-part. This way, B-brain could learn to play a role somewhat like that of a counselor, psychologist, or management consultant, who can assess a client’s mental strategy without having to understand all the details of that client’s profession. This abstraction layer can be increased by adding maybe another C-Brain.
Figure: Brain hierarchy with abstraction layers.
Concluding thoughts: Although the book tried to explain many concepts about human mind, the explanations are mostly hypothetical. It doesn’t provide much supporting references. But still, it paved some ideas about hierarchical or circular connections between subcomponents where higher level doesn’t care about the goal or other details of lower level. This might be useful in decoding intelligence modeling.
