#79. Scientific Brainstorming Methods Used Here [engineering]

EN

Red, theory; black, fact

The observer in an improvised eclipse observatory

Explanations of the scientific method all have a spongy spot: “Think of a theory.” So how do you do that? The most advanced form of the scientific method, strong inference, is even worse: “Think of at least two theories to account for the same facts!” Here I am trying to bring some system to the task, engineering the scientific method itself.

Some of these brainstorming methods (A list) reflect the structure of nature and may be considered a guide to what you know; others (B list) reflect the limitations of the human brain and may be considered a guide to what you don’t know but may be prone to assume. What you don’t know but merely assume is a candidate for deletion. An act of theorizing may have an intrinsic deletion-insertion structure, even if what is deleted is subjectively a piece of white space or blur space around a legacy theory that conceals links to other things.

The hope is that all these methods can be condensed and codified somewhat to facilitate the training of an AI theoretician’s assistant.

Each method is followed by the number of at least one post in which it was used. These are not linked because the search engines wouldn’t like so many links on one page. Definitions are in parentheses.

A. Nature.

1. Apply consilience. (Mental schematics of how something works that are useful in one field/category can also be useful in a different field/category.) Apply across (x) species boundaries, as when using research on rats to get insights about diseases in humans. #34, #57, #69, #75
2. Consilience: apply x levels of description. Examples of levels of description are the human world, made of societies, made of individuals, made of cells, made of molecules, made of atoms, made of electrons, protons, and neutrons, made of…? #73, #75
3. Consilience: apply x length scales, timescales. This is basically the usefulness of the Fourier, wavelet, and Laplace transforms. #12, #13, #23
4. Consilience: apply x nature and technology. Example: the parts of the eye correspond to the parts of a camera. #2, #3, #76
5. Match numbers of things in different fields. Example: there are five main neuromodulators in the human brain and there are five dimensions of personality. #69
6. Identify if-then relationships x levels of description. An example is allosteric control of an enzyme’s catalytic site; the primary, secondary, and tertiary structures of proteins are sub-levels of description. #74, #77
7. Use the evolutionary psychology concept. (Human behaviour can be directly explained by evolutionary arguments.) This is identifying if-then relationships x levels of description & timescales. #15-18, #21, etc.
8. Apply recursion. (Apply the same process to the result of its previous application.) #58
9. Re-use a concept from a previous theory. #62, #67
10. Add feedback. The presence of exponential increases or decreases hints at feedback. #74
11. Add hormesis. (Effects can be either increasing or decreasing in different dose ranges of the same substance.) #The hormesis posts in Experimentalist’s Progress
12. Go for cute. (Assume that a design exists.) #70
13. Generalize. Everyday life can be a source of observations you can generalize into a theory. #12, #65, #67, #73
14. Generalize to the next level of description or next-longer timescale. #40
15. Split a concept. For example, you could do this by assigning causation to different synaptic links in the mediating brain pathway, such as in distinguishing between retinal blindness and brain blindness. #36, #37
16. Visualize. #58
17. Visualize with zooming in. #72
18. Apply the divide-and-characterize strategy. This method may be going obsolete. #29
19. Engage in system building. This can involve setting up a matrix of rows and columns and then filling in the cells. The periodic table is an example of one such system from chemistry. #30, #35, #59

B. Human.

1. Apply new concepts published by others. #1
2. Call a spade a spade. Example: chromatin is a polymer and therefore polymer behaviors such as swelling are relevant to cell biology. #66
3. Re-draw entity boundaries. #38
4. Reframe a concept or phenomenon from an unconventional assumption base. #47, #49-51
5. Assume that the past is still happening. #44
6. Provide an explanatory mechanism. #44
7. Don't let the first objection that comes to mind dissuade you from your theory, but start to problem-solve. #45
8. Identify a problem. Problems are the friends of the theoretician. #71
9. Identify the essential properties of something and then ask what else that could be better has those properties. #31
10. Admit variables in stages to a simple core idea (messification, my coinage). This is more gradual and structured than mind mapping and involves visualization with zooming in. #71, #72, #74, #76
11. Add shades of grey to supposed dichotomies. #35, #36, #75
12. Selectively elevate 1-3 facts in search of an explanatory core. #25, #52, #60, #64
13. Take the next step. You can be sure that Nature has—long ago. #14, #75, #77
14. Contemplate the mathematics. #19, #73
15. Avoid perfectionism. #The early physics posts
16. Ask the 5 Ws questions. #77
17. Reverse the conventional explanation. As an example, combustion adds a substance, namely oxygen; it doesn’t subtract a substance, namely phlogiston, as the alchemists believed. #32, #52, The hormesis posts in Experimentalist’s Progress
18. Don't accept conventional explanations as complete. #33
19. Take the math constructs literally, not just as aids to calculation. The heliocentric model of the solar system was literally true, in addition to being an aid to calculation as it was first billed. Modern quantum mechanics is full of such aids to calculation but avoids answering the question: “How and of what is the world made?” #71
20. Build a Rube Goldberg device on paper to explain the phenomenon. Some explanation may be more helpful than none. #68, #74

Multiple eclipse images formed by fortuitous pinhole cameras discovered at a playground