Friday, February 10, 2012

Psychology & Computation Link

There are interesting parallels between algorithms / computation and Jungian psychology.

Si is introverted sensing, it looks at the current events and "remembers" similar events, objects, sayings from the past. Then, Si is basically a nearest-neighbour approach that is used in Machine Learning. It has no model, the model is the data itself.

Ti is the ultimate modeler. It tries to summarize data, tries to set clear boundaries between definitions so that categorizations and predictions are more accurate. Ti in ML is anything that uses a graphical, analytical, structure based model such as Bayesian Nets, ID3, or already cooked up formula with some missing parameters. Si cannot predict, classify things that it did not see before. Ti can.

Te has logic, measurement, contingency planning. In compsci terms it is a combo of a recursive depth-first search, logic and sensor data processing. It has rudimentary modeling abilities. It is quick so it probably utilizes a cache (hah!).

Ne, Ni are the ultimate non-deterministic generator of possibilities. They are probably "multi-threaded", as they generate many possibilities, sometimes blindly, these many possibilities can be executed in parallel. Even then however, their job might take a long time to finish, which must be why Ni,Ne is known to keep working even when a person is asleep, busy with other things. A "discovery" popping into a scientist's head is simply Ni, Ne finishing its work. There's nothing mysterious about it. Note: Generating possible solutions and verifying them are seperate tasks of course (per our P ≠ NP? discussion), Ne,Ni generate, Ti,Te verify.