A lot of the important decisions we make are based on incomplete knowledge. Oftentimes, it’s not that we are making decisions carelessly, but rather that some aspects of a situation are simply unknowable.
Yet we don’t generally freeze like rabbits, unable to make a decision. Somehow we charge onward, continuing to decide on our next move while living with uncertainty.
I wonder whether there is a complex mechanism in the human brain that allows us to do this. Is there some sort of intuitive “probability machine” that we are always running, which assesses the current odds of success in any situation?
If so, I wonder how accurate it is, and what sorts of things can throw it off. We already know that there are obvious answers to that last question. Lack of sleep or too much alcohol can completely shut down our ability to choose the higher probability option for success.
But what about the opposite question? What can we do to increase the accuracy of our internal probability machine? Whatever it is, I’d like to try it!
There was a book about that: Superforecasting: The Art and Science of Prediction
Three fascinating models…
(1) I think followers of Harold Jeffreys supposed that neural circuits were something like little Bayesian hypothesis-testing evidence accumulators that worked on log-of-odds. For comparing only two hypotheses, this means just performing a very simple operation: adding a simple constant, a log-of-likelihood-ratio value. Maybe that’s circumstantial evidence that we sometimes work through the network of options by taking subjects in pairs.
As for what probability to use when we have no evidence, Jeffreys suggested what’s now called the Jeffreys prior.
(2) David R Heise used an impression-formation matrix to represent people as actors trying to maintain social norms (maybe Perlin noise could even simulate “proper deviations” from those trajectories), where he would multiply evidence vectors into a matrix that represented one’s outlook. He could solve for different information about the perceived role of the person, their setting, or associated symbols and emotions, and sometimes the results were unexpected but uncannily accurate. He used a 3D space of “good/bad”, “powerful/weak”, “active/passive”.
[3] Charles Peirce during part of his life advocated the idea of abduction (in addition to deduction and induction) being a mechanism for mostly unconscious symbol association. Wikipedia’s example:
Given…
Rule: All the beans from this bag are white.
Result: These beans [oddly] are white.
=> Case: These beans are from this bag.
[1] https://en.wikipedia.org/wiki/Likelihood-ratio_test
[2] https://books.google.ru/books?id=vMgOoTrI6RYC&pg=PA84&lpg=PA84&dq=david+r+heise+impression+formation+matrix&source=bl&ots=RKwl0HVKoD&sig=ACfU3U3Mqwljmtm-k8U5MAtKbjTYpugYyw&hl=ru&sa=X&ved=2ahUKEwit1MCylJrqAhWEqIsKHTRACTgQ6AEwAHoECAoQAQ#v=onepage&q=david%20r%20heise%20impression%20formation%20matrix&f=false