How do we find the ''best'' explanation?

Whenever we set out to learn something new, or we run across information that seems new to us, what makes us trust the information as being truthful? When it comes to matters that are testable through the scientific process, how can we be sure that the explanation being offered, is the best one? 

When we feel that someone is telling us the truth, is it because of confirmation biases that lead us to believing those truths? 

The best explanation should be a factual and reasonable one, but how do we trust facts and reason? Is it because the facts are accepted by the majority? How can we evaluate and trust reason to ensure that it serves us as the ''best'' explanation? 

What are your ideas on this?

(posted this on SF, but know some here don't post there)

At first exposure to new info, it is usually confirmation bias that determines our initial impression. When learning new findings in a field you are already familiar with, older knowledge in that field helps temper bias. Scientific truths are only as good as their ability to predict behavior/new results. Experiment is the only real arbiter. Majority has nothing to do with it, but how consistently findings can be replicated does.

On the other forum where I posted this, an interesting comment came up that someone feels that his ''trust'' comes in if he values the reputation of the scientists who are proposing new ideas, etc. And the reputation of those agreeing with those views. I think that confirmation bias does play a big part too, so agree Syne.

Reputation is just an appeal to authority, and agreement an appeal to consensus. These have no impact on validity, though they are often used as shortcuts for judging new information...especially by laymen.

(Oct 14, 2016 10:09 PM)Leigha Wrote: How can we evaluate and trust reason to ensure that it serves us as the ''best'' explanation? What are your ideas on this?

Charles Sanders Peirce introduced "abduction". This so-called abductive reasoning is "inference to the best explanation". But minus the intellectual loftiness it's just speculations (guesswork) being formed and submitted to account for an observation or acquired data.

The adjective of "best" indicates a pre-established standard for determining why one among multiple, competing hypotheses would "have the most positive qualities". That standard for determining "best" is initially an abstract placeholder; there may be multiple options or alternatives that can be plugged into it.

A (definite) standard must be set before proceeding. The standard selected will be invented by people, it is not a dictum or universal truth discovered lying underneath a rock or bush somewhere; it is not naturally or objectively provided by the universe.

Peirce provided his own standard for what constituted "best": "Facts cannot be explained by a hypothesis more extraordinary than these facts themselves; and of various hypotheses the least extraordinary must be adopted." (Paraphrasing Peirce's discussion, we might say that the best hypothesis is one that is simplest and most natural, is the easiest and cheapest to test, and yet will contribute to our understanding of the widest possible range of facts.) --You Know My Method: A Juxtaposition of Charles S. Peirce and Sherlock Holmes .... Also: ITTBE

Hybrid ideas may be plugged into the placeholder for the standard which will determine "best".

For instance, Peirce's choice might be combined with "the explanation must be coherent with an already existing body of accepted knowledge, theories, views, and approaches". If his usage of "least extraordinary" does not already indirectly embrace this.

Statistical theorems for quantitatively measuring which hypothesis is most likely true could be recruited. Though that arguably bulldozes over the conservative humility of "best". Which does not necessarily in itself make any claim to certitude or perpetual certitude as goal.

Leigha Wrote:How do we find the "best" explanation?

That's an excellent question, one that's basic to the whole philosophy of science.

An earlier less developed version of the following remarks was also posted on SF. They are my take on Peter Lipton's chapter on 'Inference to the Best Explanation' from the Blackwell Companion to the Philosophy of Science. Some of the ideas and examples below come from there. I see that the always philosophically insightful CC has already made similar remarks.

http://www.people.hps.cam.ac.uk/index/lipton/inference

Explanations are involved somehow with the relationship between evidence and theory. Evidence is interpreted as evidence of something, but what kind of relation is this and how do we decide what the 'something' should be? Both in science and in everyday life, people make use of the available evidence to somehow infer what they believe would best explain that evidence. That connection is rarely one of deductive logic, at least when moving in the upward evidence -> hypothesis direction. This process of  'inference to the best explanation' seems closely related to induction and was given the name abduction by Charles Peirce.

Darwin proceeded from evidence like the beaks of his Galapagos finches to what he believed was the best explanation for the variation in beaks (natural selection). Sherlock Holmes interpreted forensic evidence to conclude that the crime was committed by Moriarty. If you walk down a deserted beach and encounter a sand-castle, you will probably conclude that somebody who has now left built it earlier. The street is wet so we conclude that it rained. But despite Holmes announcing that it's all "deduction", in none of these cases does the evidence deductively imply the conclusion. Maybe there's another explanation for the variation in the birds' beaks, maybe somebody other than Moriarty committed the crime, maybe the wind or MR's space aliens are responsible for the sand castle and maybe somebody sprayed the street with a hose.

There's an interesting circularity built into our explanations. Red shifts are taken to be evidence of stellar recession, while stellar recession is taken to be the explanation of the red shifts. This circular loop between evidence and explanation happens all the time in science. It may not be a destructive circularity though. It would be destructive if each side of the relation is taken to justify the other. So where does justification of belief fit into all this? Do observations 'abductively' justify the explanatory hypotheses that we construct based on those observations? Should explanatory hypotheses be understood as somehow justifying the evidence? Or are explanations performing some other function?

What makes one explanation better than another? Should we choose the explanation that is most probable? The explanation that conveys the most understanding? The most mathematically elegant or the simplest explanation? Why does opium put people to sleep? If we explain it by saying that opium has a sleep inducing power, we achieve simplicity and probability, but at the cost of informative power. Great chefs make the best meals, but that doesn't tell us how they do it.

If the goal is to increase understanding, what features of an explanation contribute to understanding? Scope, precision, mechanism and unification are suggested in Lipton's article. Better explanations explain more, they explain with more precision, they tell us how what is observed came about and they unify seemingly dissimilar phenomena.

There's lots more that could be said, see the article above and the article on 'abduction' in the SEP:

http://plato.stanford.edu/entries/abduction/

(Oct 15, 2016 10:23 PM)C C Wrote:

(Oct 14, 2016 10:09 PM)Leigha Wrote: How can we evaluate and trust reason to ensure that it serves us as the ''best'' explanation? What are your ideas on this?

Charles Sanders Peirce introduced "abduction". This so-called abductive reasoning is "inference to the best explanation". But minus the intellectual loftiness it's just speculations (guesswork) being formed and submitted to account for an observation or acquired data.

Explanatory hypotheses seem to emerge from evidence somehow. They don't typically follow deductively, since the hypotheses that we invent to explain the evidence aren't formally implied by that evidence. Yet it appears that there is some logic-like process of inference at work there. Abduction seems to me to be closely related to induction.

Quote:The adjective of "best" indicates a pre-established standard for determining why one among multiple, competing hypotheses would "have the most positive qualities". That standard for determining "best" is initially an abstract placeholder; there may be multiple options or alternatives that can be plugged into it.

I agree.

Quote:A (definite) standard must be set before proceeding. The standard selected will be invented by people, it is not a dictum or universal truth discovered lying underneath a rock or bush somewhere; it is not naturally or objectively provided by the universe.

Context is hugely important.

In science the history of science, and the nature of scientific education, particularly of doctoral-level researchers, shape the context, the stock of concepts, methods and examples of previous explanations, that scientists reach into when they feel the need to explain something. In 1400, if somebody wanted to explain movements of heavenly bodies, they would reach into the geocentric cosmology bag and pull out equants and epicycles. In 1700, they would reach into the new science bag and pull out Newtonian gravitation and Kepler's laws. Today cosmologists reach into their bag and pull out the big-bang and general relativity.

Quote:Peirce provided his own standard for what constituted "best": "Facts cannot be explained by a hypothesis more extraordinary than these facts themselves; and of various hypotheses the least extraordinary must be adopted."

That's interesting. Maybe Peirce is the original source of that annoying slogan that skeptics always toss out: Extraordinary claims (hypotheses) require extraordinary evidence (facts).

Quote:(Paraphrasing Peirce's discussion, we might say that the best hypothesis is one that is simplest and most natural, is the easiest and cheapest to test, and yet will contribute to our understanding of the widest possible range of facts.)

Maybe Peirce was edging towards the idea that explanation needs to reduce the unknown to the known. If we have a phenomenon that we don't understand and want to explain, we need to explain it in terms of entities and principles that are better understood. Otherwise we aren't getting anywhere. (Supernatural and religious explanations often ignore that one.)

There's also the question of probabilities. The SEP gives the example of walking into the kitchen in the morning and finding a plate on the table, bread crumbs, a pat of butter and an empty milk carton. So we conclude that a house-mate made a snack for him/herself during the night. Of course it might have been somebody laying all that stuff out without eating anything, just to make us think somebody had eaten. But we reject that latter conjecture as too contrived and opt for the one that seems most probable. (Conspiracy theories often ignore that one.)

Quote:Hybrid ideas may be plugged into the placeholder for the standard which will determine "best".

For instance, Peirce's choice might be combined with "the explanation must be coherent with an already existing body of accepted knowledge, theories, views, and approaches". If his usage of "least extraordinary" does not already indirectly embrace this.

Statistical theorems for quantitatively measuring which hypothesis is most likely true could be recruited. Though that arguably bulldozes over the conservative humility of "best". Which does not necessarily in itself make any claim to certitude or perpetual certitude as goal.

Again, I'm inclined to agree.

Thanks everyone for your insightful thoughts to this! I'll come back later to respond further to some of your ideas.

Here's one more abstract question, though - Will the ''best'' explanation only be what is best to us, or what is best in general? I may have more knowledge than another person about a particular topic, but he might have more about another. If he doesn't understand the topic to begin with very well, how will he decipher a ''good'' explanation from the ''best'' one? Maybe we need to come to the table with a foundation of knowledge to begin with, before being able to determine a ''best'' explanation, even if it's only ''best'' for us. Do you agree?

(Oct 16, 2016 05:05 PM)Leigha Wrote: Here's one more abstract question, though - Will the ''best'' explanation only be what is best to us, or what is best in general?

I'm not sure what 'best in general' would mean. 'Best' only seems to make sense when there are a set of alternatives. So how is the set of alternatives defined?

We observe something. We invent a hypothesis to explain the observation. That hypothesis will just be one of a set of possible hypotheses that might also be consistent with the observation. Arguably, that set of possible hypotheses might be infinitely large. That infinite set will be reduced considerably if we restrict ourselves to hypotheses that have actually been proposed, as compared to those that might have been but weren't.

So does "best" apply to the big set of possible hypotheses, including the ones that nobody has ever thought of? That would seem to correspond to your "best in general", but how would we ever make judgements about 'best' concerning hypotheses we have never thought of and know nothing about?

Or does "best" apply to the small set of hypotheses that we are actively considering? That would seem to correspond to your "best for us". I think that's how people typically proceed in science and in real life.

Quote:I may have more knowledge than another person about a particular topic, but he might have more about another. If he doesn't understand the topic to begin with very well, how will he decipher a ''good'' explanation from the ''best'' one? Maybe we need to come to the table with a foundation of knowledge to begin with, before being able to determine a ''best'' explanation, even if it's only ''best'' for us. Do you agree?

Yes. I think that hypothesis formation and making selections among explanatory hypotheses as to which ones are best is context-sensitive. In the sciences, it takes place in the context of the history of science and the scientific education that scientists receive in universities. Scientists approach observations with a whole tool-kit of technical concepts, methods and examples of earlier explanations that they deploy in various and often creative ways when they encounter new things that they want to explain.

(Oct 16, 2016 06:17 PM)Yazata Wrote:

(Oct 16, 2016 05:05 PM)Leigha Wrote: Here's one more abstract question, though - Will the ''best'' explanation only be what is best to us, or what is best in general?

I'm not sure what 'best in general' would mean. 'Best' only seems to make sense when there are a set of alternatives. So how is the set of alternatives defined?

We observe something. We invent a hypothesis to explain the observation. That hypothesis will just be one of a set of possible hypotheses that might also be consistent with the observation. Arguably, that set of possible hypotheses might be infinitely large. That infinite set will be reduced considerably if we restrict ourselves to hypotheses that have actually been proposed, as compared to those that might have been but weren't.

So does "best" apply to the big set of possible hypotheses, including the ones that nobody has ever thought of? That would seem to correspond to your "best in general", but how would we ever make judgements about 'best' concerning hypotheses we have never thought of and know nothing about?

Or does "best" apply to the small set of hypotheses that we are actively considering? That would seem to correspond to your "best for us". I think that's how people typically proceed in science and in real life.

Quote:I may have more knowledge than another person about a particular topic, but he might have more about another. If he doesn't understand the topic to begin with very well, how will he decipher a ''good'' explanation from the ''best'' one? Maybe we need to come to the table with a foundation of knowledge to begin with, before being able to determine a ''best'' explanation, even if it's only ''best'' for us. Do you agree?

Yes. I think that hypothesis formation and making selections among explanatory hypotheses as to which ones are best is context-sensitive. In the sciences, it takes place in the context of the history of science and the scientific education that scientists receive in universities. Scientists approach observations with a whole tool-kit of technical concepts, methods and examples of earlier explanations that they deploy in various and often creative ways when they encounter new things that they want to explain.

Good thoughts to this, Yazata. ''Best'' meaning -- the chosen, final answer that the majority accepts as truth, and there are no alternatives to that ''truth.'' (if that makes sense)