Abduction is little more than an attempt to formalise confirmation bias. — Banno

Not at all. Abduction is the use of the creative imagination in formulating testable hypotheses that might best explain the observed facts. An abductive hypothesis is always provisional—open to rigorous testing, and thus quite the opposite of confirmation bias.

Do you reject everything science teaches? — Relativist

Of course not. A rejection of one way that philosophers have claimed science works is not a rejection of science. That scientific theory is developed through abduction is a theory about scientific method. Pointing out the problems with that theory is not pointing out problems with what scientists do, but with what philosophers claim that scientists do.

Abduction is the use of the creative imagination in formulating testable hypotheses that might best explain the observed facts. — Janus

There's that word "best" again. It hides that the criteria being used are things such as parsimony, coherence, and predictive success, normative concerns. Why not drop the pretence of "abduction" as a seperate rational process and look instead at the basis that scientists use for choosing between rival theories.

An abductive hypothesis is always provisional—open to rigorous testing, and thus quite the opposite of confirmation bias. — Janus

Notice that testing is a seperate process to abduction - one adduces the "best" explanation and then tests it. Abduction is not necessary for testing an hypothesis.

There's that word "best" again. It hides that the criteria being used are things such as parsimony, coherence, and predictive success, normative concerns. Why not drop the pretence of "abduction" as a seperate rational process and look instead at the basis that scientists use for choosing between rival theories. — Banno

I don't think you understand what 'abduction' means in the context of science. It is the use of the imagination to come up with what seems to be the most fitting explanatory hypotheses. Of course testing is a separate issue. I only used the word 'best' to underline the fact that in hypothesizing the desirable aim is always to come up with what, consistent with whatever criteria, seems to be the best explanation possible given whatever limits of information and imagination are in play.

Testing is an entirely separate issue—it comes after the hypothesizing. To put it concisely 'abduction" simply refers to the process of forming hypotheses.

Notice that testing is a seperate process to abduction - one adduces the "best" explanation and then tests it. Abduction is not necessary for testing an hypothesis. — Banno

As I say above I haven't said or implied that testing is a part of the abductive process. Of course abduction is not necessary for testing an hypothesis—it is necessary for coming up with an hypothesis to be tested.

I don't think you understand what 'abduction' means in the context of science. — Janus

Yep. Just so. Do you?

It is the use of the imagination to come up with what seems to be the most fitting explanatory hypotheses. — Janus

Ok, that's it's proposed use - how does it manage to do this?

to come up with what, consistent with whatever criteria, seems to be the best explanation possible — Janus

But there is no method for doing this - only what someone claims to be the "best" hypothesis.

To put it concisely 'abduction" simply refers to the process of forming hypotheses. — Janus

Well, if that is all it is, then it doesn't tell us which to choose among the many - which is "best"...


So we have deduction, and formal definitions of validity. We have induction, which consists in the claim that if all previous A's were B's, then we might well infer that all subsequent A's will be B's. As Hume pointed out it's formally invalid, but we do it anyway. Then we have a something labeled "abduction", that allows us to infer the best hypothesis, without telling us what "best" is. It's another name for selecting the hypothesis one prefers, without giving any reason.

There's more. Any explanation can be claimed to be the best, since no criteria are set out in the notion of abduction. But for any body of evidence, there are indefinitely many hypotheses that could explain it. Every test of a hypothesis involves multiple auxiliary assumptions. Any “failure” could be blamed on the main hypothesis or the auxiliary assumptions. Apply Duhem and Quine and the notion falls apart.

To put it concisely 'abduction" simply refers to the process of forming hypotheses.
— Janus
Well, if that is all it is, then it doesn't tell us which to choose among the many - which is "best"... — Banno

There is no certain way to ascertain which is "best". That's why I said "seems best'. The only way to ascertain whether a hypothesis is a good one is to test it, which means seeing whether the predictions that are entailed by the hypothesis are observed.

You mentioned parsimony, coherence, and predictive success—the first two would presumably be in mind when forming rigorous hypotheses. Coherence would be relative to consistency with existing accepted scientific understanding.

In any case, to refer to my original response—I was disagreeing with the assertion that abduction has anything to to with confirmation bias, and I say this is not so because hypotheses are to be tested, not accepted on account of their "feeling right" or whatever.

That said, feeling right might come into play in choosing between rival hypotheses, but acceptance of one over the other would be contingent on entailed predictions being observed, as well as coherence with current accepted scientific belief and understanding.

I was disagreeing with the assertion that abduction has anything to to with confirmation bias, and I say this is not so because hypotheses are to be tested, not accepted on account of their "feeling right" or whatever. — Janus

Indeed, and my reply was to reaffirm that the testing of an hypothesis is not part of performing an abduction. Abducting is choosing the "best" hypothesis, on the basis of one's preferences - the very meaning of confirmation bias - the tendency to interpret a situation so as to confirm one's preexisting attitudes.

It's presented as "given some evidence, infer the hypothesis that would best explain it" where "best" is left ill-defined. This leaves it entirely open to arbitrarily inferring any explanation to be the best. — Banno

It’s not the explanation that’s the best, it’s the hypothesis that’s a good one and worth testing. A strawman from the king of strawmen. You are waving words around as if they were arguments.

Indeed, at its heart, it remains unclear what abduction amounts to; and as such, it is ineligible as a grounding for rational discourse. — Banno

Abduction is brainstorming.

Brainstorming is a creativity technique in which a group of people interact to suggest ideas spontaneously in response to a prompt. Stress is typically placed on the volume and variety of ideas, including ideas that may seem outlandish or "off-the-wall". Ideas are noted down during the activity, but not assessed or critiqued until later.

Is it rational? Yes, of course. It’s a method that works to generate new ideas that can be evaluated and justified. It can be very effective, as I know from my own career. It would be irrational not to use it under the appropriate circumstances.

:wink: I wish you would think.

Here's a thing - what does "abduction" mean? Even the SEP article can't say. So now you claim it's just making up an hypothesis. So why not just call it "hypothesising"? Why the new name?

Becasue Peirce liked triads, and wanted to add something to deduction and induction to make it a trinity. A very poor bit of abduction...

If you, any of you, think you have a clear notion of what abduction is, and why it is useful, set it out! There's be a Doctorate in it for you.

Of course not. A rejection of one way that philosophers have claimed science works is not a rejection of science. That scientific theory is developed through abduction is a theory about scientific method. Pointing out the problems with that theory is not pointing out problems with what scientists do, but with what philosophers claim that scientists do. — Banno

OK, then you know what abduction is, and claim that science doesn't actually use abduction. Tell me what science actually does that lends it credibility, that is lacking with abduction.

You didn't answer my question about conspiracy theories- are they credible? If not, why not?

That's not what was said. But if you can set out what you think abduction is, and how it works, just go ahead and do so.

I don't think anyone here can. I think folk have been told that what they are doing is abduction, but not looked to see what that meant; and when they do look, they will see the idea is hollow.

Show how I am mistaken.

Here's a thing - what does "abduction" mean? Even the SEP article can't say. So now you claim it's just making up an hypothesis. So why not just call it "hypothesising"? Why the new name — Banno

My preferred approaches to philosophy of science find the concept of abduction problematic for a number of reasons. First, abduction is too rationalistic; real science is more anarchic. Second, abduction misses the paradigm-dependence of hypothesis generation. And third, abduction isn’t a universal logic but a practice-specific activity embedded in forms of life. Maybe this is at least somewhat consistent with your objections.

If you, any of you, think you have a clear notion of what abduction is, and why it is useful, set it out! There's be a Doctorate in it for you. — Banno

I’ve already answered that question. I recognize you don’t like my answer.

OK, I'll bite. I'll give a general idea of abduction.

Abduction entails drawing a non-necessary inference from a set of data (intended to be all available, relevant data), that consists of an explanatory hypothesis for that data - one that is deemed to explain the data better* than alternatives.

The inference is defeasible- it can be falsified by new, relevant data (previously overlooked or newly discovered) that is inconsistent with the hypothesis. Alternatively, it can be supplanted by a new hypothesis that demonstrably provides a superior* explanation.

This is consistent with the de facto means by which science advances through theory development, verification, revision, and replacement.

________________
*I'm deferring discussion of what makes an explanation better/superior.
________________

I could go into more details, but this suffices as a general description. Describe what what you disagree with, and provide what you consider a correct description.

Pretty much. Abduction was perhaps an attempt to form an algorithmic method for science, back in the days when such things were thought important. Where you are presumably coming to this from a phenomenal logical perspective, mine is more to do with the approaches of Popper, Quine and eventually Fayerabend. Perhaps what we have in common is the rejection of algorithmic processes.

you really haven’t given a validation of abduction. If you really believe that the whole of abduction is just brainstorming, then you’re missing quite a bit. Check out the SEP article perhaps.

Indeed, and my reply was to reaffirm that the testing of an hypothesis is not part of performing an abduction. Abducting is choosing the "best" hypothesis, on the basis of one's preferences - the very meaning of confirmation bias - the tendency to interpret a situation so as to confirm one's preexisting attitudes. — Banno

I don't think it's fair to say that the hypotheses that are chosen are merely a matter of one's preferences—to repeat, they should be consistent with current scientific theory and understanding. If you want to call that confirmation bias, then you'd better apply that judgement to the whole of science.

Confirmation bias consists in ignoring evidence that tells against one's preferred beliefs, and scientific practice should be the opposite of that—it should involve actively trying to falsify current accepted belief and theory and attempting to find better, more comprehensive hypotheses.

I had a quick look at the SEP article, and I didn't find it helpful—firstly because I don't think there is a significant difference between the two kinds of abduction it [purports to identify, and secondly because it seems to conflate the process of imagining hypotheses with a purported tendency to believe in them.

As I understand it, abduction is simply the creative imagining of hypotheses—and the next step would be to test them rigorously—if someone doesn't want to do that, but wants to cling to their "per hypothesis" that has nothing to do with abduction per se. So, abduction is not confirmation bias in any necessary sense, although of course abductive reasoning might be used to confirm biases.

you really haven’t given a validation of abduction. — Banno

It doesn't need to be validated. What would you replace it with?

One can simply create a way of thinking, and affirm it by living by that method. It's legitimate because it's a way of life that works for many. It's like religion, in that sense. Doesn't make it a very worthwhile way of thinking though... that whole inner outer world thing has lead so many thinkers astray, from Plato to Kant, and even still more beyond them like Dennet.

Do you reject everything science teaches? Scientific theory is developed through abduction, and it has proved successful. — Relativist

That seems a rather strange, if not perverse, response from someone who thinks we should be guided by the science as to what to do about human-induced global warming. — Janus

Hume's attack is not on science here, and it is not on morality when he points out that you can't derive an ought from an is. As I have pointed out elsewhere, the two arguments are very similar, and what he is attacking in each case is a form of argument. Let me put it this way, he is not attacking rationality, he is attacking rationalism. He is attacking dogmatism. And when you are unable to defeat his logic, and also unable to accept his arguments, it is dogmatism that you are showing.

Yes, I believe in science, and I hope it will continue to provide a guide to the future and tools for the present for whatever future we may have. But I do not pretend that belief and hope are "rational" while unable to provide any coherent rationale. I also believe in trying to be truthful in communication, and kind to other people, and I wish you would too, but I do not pretend that it is rational, because it isn't.

Rationality cannot tell us how to live, or even what to think; to answer these sorts of question requires something else. it requires caring about something.

Abduction entails drawing a non-necessary inference from a set of data (intended to be all available, relevant data), that consists of an explanatory hypothesis for that data - one that is deemed to explain the data better* than alternatives. — Relativist

Cool. Nice and clean. Good stuff.

A few issues.

"Better" - an improvement on "best", but suffering the same ambiguity. If abduction is going to tell us which of the innumerable possible explanatory hypotheses to choose, then we need more than an asterisk and a deference. We need the basis for that choice. Otherwise abduction falls to underdetermination, to the Duhem–Quine problem.

Whence normatively? Deductions are preferred when valid. Inductions are preferred on the basis of Bayesian statistics. What of abduction? If it is statistical success, then isn't it just anther name for induction? If it's some pragmatic or parsimonious, isn't that just an appeal to aesthetics? to what you prefer?

So it comes down to how you cash out better/superior.

And hence my original point, that whatever criteria you choose, you are subsequently just reinforcing that choice.

—to repeat, they should be consistent with current scientific theory and understanding. If you want to call that confirmation bias, then you'd better apply that judgement to the whole of science. — Janus

Well, yes -

Second, abduction misses the paradigm-dependence of hypothesis generation. — Joshs

Conformation of the current scientific theory. Feyerabend would have a party here.

Better" - an improvement on "best", but suffering the same ambiguity. If abduction is going to tell us which of the innumerable possible explanatory hypotheses to choose, then we need more than an asterisk and a deference. We need the basis for that choice. Otherwise abduction falls to underdetermination, to the Duhem–Quine problem. — Banno

If a conclusion were "determined" (not underdetermined) it would be a deduction- a conclusion that follows necessarily.

Of course, as you said, deduction would be preferred, but in real life (including science) we rarely have sufficient information to make a deduction. But there is often sufficient information to support some hypotheses more than others.

So it comes down to how you cash out better/superior.

And hence my original point, that whatever criteria you choose, you are subsequently just reinforcing that choice. — Banno

Methodology is indeed key. Some basics: explanatory scope and power, parsimony, more plausible than alternatives (consistent with more facts that are commonly accepted), fewer ad hoc assumptions (ad hoc suppositions are assumptions that are not entailed by the data and other commonly accepted facts). Biases entail ad hoc assumptions. It also entails consideration of other hypotheses.

Ideally, an abductive conclusion ought to be only as specific as the information warrants, otherwise it will include ad hoc assumptions.

Finally, the level of certainty ought to tied to the strength of the case. For example, consider a jury verdict based on a preponderance of evidence vs one based on "beyond reasonable doubt". A chosen "best" explanation may still be (arguably) unlikely. There's always the risk of choosing "the best of a bad lot"- which would tend to be the case when the data is sparse.

It's useful to solicit and receive feedback from others with divergent views. This can help identify overlooked, relevant facts, challenge assumptions that are ad hoc or reflect bias, and identify alternative hypotheses for comparing.

Your claim that abductive reasoning entails "reinforcing that choice" is false. That would be a corruption of abduction, such as occurs with conspiracy theories. The criteria I listed are sufficient to reject conspiracy theories. They depend on biased (ad hoc) assumptions, cherry-pick facts, instead of considering all relevant facts, reject or rationalize facts that are inconsistent with the theory, and a person's conviction toward a theory tends to gain strength by seeking endorsement others with similar biases, while alternative hypotheses are rejected on the basis of bias and/or an unjustified faith in the conspiracy theory.

Accepting science means you treat the body of scientific information as true, despite the fact that it is possibly false. IOW, the mere fact that a scientific "fact" is possibly false is not sufficient reason to withhold judgement on it.

To be consistent, apply this more broadly. It's reasonable and rational to draw conclusions based on incomplete information. That's what abductive reasoning is all about. I discussed this in my last 2 replies to Banno (1. Here and 2.here).

Accepting science means you treat the body of scientific information as true, despite the fact that it is possibly false — Relativist

No it doesn't. Scientists are not all equally scrupulous, and are subject to peer pressure, the persuasion of big pharma et cetera, and the need to get funding. Some science is biased and some is slapdash, and some is bullshit. It's not supposed to be religion where you just believe what the high priests say.

Accepting science means you treat the body of scientific information as true, despite the fact that it is possibly false — Relativist

No it doesn't. — unenlightened

Then explain by what you meant by "I believe in science."

Scientists are not all equally scrupulous, and are subject to peer pressure, the persuasion of big pharma et cetera, and the need to get funding. Some science is biased and some is slapdash, and some is bullshit. It's not supposed to be religion where you just believe what the high priests say. — unenlightened

False equivalence. Science is not equivalent to what individual scientists say. I'm referring to commonly accepted theory. How we deal with the potential for bias by scientists is another matter.

Then explain by what you meant by "I believe in science." — Relativist

What I believe is that science is a sceptical endeavour, that progresses by means of demonstration. which is to say, that I expect scientists not to put their trust too lightly in the work of others, but require experiments to be repeated, and findings to be demonstrated, and theories to be treated as provisional whenever their scope is extended.

Science is not equivalent to what individual scientists say. I'm referring to commonly accepted theory. — Relativist

And individual scientists do not talk about commonly accepted theory? Really, what do you imagine needs your stalwart defence here? Are you having a battle to see who understands science better? Enough already!

Science is not equivalent to what individual scientists say. I'm referring to commonly accepted theory.
— Relativist

And individual scientists do not talk about commonly accepted theory? — unenlightened

Irrelevant to the point I made: you accept some things as true, despite the possibility it is false.

Really, what do you imagine needs your stalwart defence here? Are you having a battle to see who understands science better? Enough already!

Not all all. Up to now, you seem to have been arguing that if a statement is POSSIBLY false, then it cannot be assumed true. That is what I was challenging.

What I believe is that science is a sceptical endeavour, that progresses by means of demonstration. which is to say, that I expect scientists not to put their trust too lightly in the work of others, but require experiments to be repeated, and findings to be demonstrated, and theories to be treated as provisional whenever their scope is extended. — unenlightened

:up:

----------

As to Peirce’s devised notion of abduction, I’ll butt in for a sec.:

Abduction is as worthless as pure guesswork when divorced from that ideal, else standard, by which that abducted is measured. That standard being simply the maximal explanatory power of the given abduction to account for those givens it seeks to explain—this especially in comparison to all other alternative explanations.

As a relatively well-known example in science, Einstein’s imaginings of traveling at the speed of light revealed to Einstein that something was amiss with Maxwell’s equations, so Einstein abducted a new explanation for how things worked, for which he then devised a new form of mathematics to properly express. None of this would have been of any scientific significance without the given abduction holding maximal explanatory power for the relevant known data. And field tests were done which empirically validated that light does indeed bend due to gravity. Yet the Theory of Relativity is as of today in partial conflict with the relevant paradigm of quantum mechanics: they each describe the physical universe using fundamentally incompatible frameworks. We hold onto both because each holds a tremendous, maximal, explanatory power for the data obtained within each branch but, because they are fundamentally incompatible, we already know that either one of the two or both are in some way(s) mistaken. The Theory of Relatively is, again, resultant of an abduction that currently has maximal explanatory power within its field of study. This is not to then say that new abductions could not eventually surface which will hold yet greater explanatory power than does the ToR. (the accumulated data, always empirical, remains unaltered, but the explanations for it can on occasion come in the form of paradigm shifts).

Same ideal standard of maximal explanatory power can, for another example, be applied to abductions regarding metaphysics, and not just the physical world which the empirical sciences study.

In overview: the explanatory power of ideas and theories is often enough an overlooked essential measure of an idea’s/theory’s worth. Same can be said for the Theory of Evolution via Natural Selection. Nothing comes close by comparison in terms of explanatory power as regards the diversity of lifeforms given all the data we know of. (I know, I know, it has that one contender of “The omni-creator deity did it” but, to be forthright, to many if not most this explanation for everything that exists does not come close to providing any understanding for what exists. So its explanatory power is very weak, if at all there. This even when granting the hypothetical of such a being.)

Abduction is no more rational than any imagining is. And I can’t fathom how Hume’s arguments again rationalism wherein sound deductions rule the land would in any manner be changed by it. It’s the explanatory power of that which has been abducted which reason states gives the abduction credibility in likely (but not necessarily) being true. This as is the case with ToR.

...it should involve actively trying to falsify current accepted belief and theory and attempting to find better, more comprehensive hypotheses. — Janus

Should. But should it?

We have before us quite different notions of abduction. Sometimes it is talked of as the process of forming an hypothesis. We know that, for any set of observations, there are innumerable possible explanations. Simply having available a range of hypotheses is insufficient. We must choose between them.

In the supplement, Peirce's version of abduction is taken from this quote:

The surprising fact, C, is observed.
But if A were true, C would be a matter of course.
Hence, there is reason to suspect that A is true. — SEP article

In reality, we do not have just A - we have alternate hypotheses, each of which explains C. And we have the possibility that C is incorrect. C is also theory laden - observations dependent on our prior presumptions as to what it is we are observing.

We are never in a position to say that this hypothesis is categorically better than that one.

So the SEP account brings in ABD1:

Given evidence E and candidate explanations H1,…, Hn of E, infer the truth of that Hi which best explains E. — SEP article

And to the problem of which hypothesis is best. Abduction does not tell us. It instead brings in the auxiliaries of simplicity and coherence, aesthetic preferences that remain unclarified within the context of abduction. So again, abduction amounts to choosing the hypothesis that looks good to you.

Feyerabend examines examples in detail, such as the tower argument, early telescopic observations, or Brownian motion, to show that science often progresses by insisting on an explanation that is not the simplest or most complete. An example - Copernicus's heliocentric model was both less accurate and more complex than the Ptolemaic alternative. The Copernican case showed that the heliocentric theory was adopted despite not being more accurate, and actually being more complex, contradicting the idea that science always progresses by choosing simpler, more predictive theories.

One example amongst many.

Now Feyerabend used such examples to show that science does not conform to the model proposed by Popper - that falsification, while useful, is very far from the full story. The case is even worse with Abduction, which remains ill-defined and obscure.

Science in real life is much, much messier than the descriptions given by Popper or Peirce. And you are right that the SEP article does not set out the notion of abduction well - but we need add that, that is a problem for abduction, it remains ill defined. The term papers over the issues of method, rather than explicating them.

My own view is of science as a social phenomenon rather than a logical one. There is no firm method that underpins the practice of science, but rather a set of attitudes that involve openness to criticism, open discussion, join examination of evidence, collaboration - the basic liberal attitudes of which we are so in need.

This is perhaps the point of intersection between myself and .

If a conclusion were "determined" (not underdetermined) it would be a deduction- a conclusion that follows necessarily.

Of course, as you said, deduction would be preferred, but in real life (including science) we rarely have sufficient information to make a deduction. But there is often sufficient information to support some hypotheses more than others. — Relativist

Quite right. We need pay close attention to that last bit. We may indeed support one hypotheses over another, but the "why" cannot be based solely on information - there must also be a preference. A mere list of facts is insufficient to decide between competing hypotheses.

That's again the problem with abduction - it doesn't set out why we should prefer one hypotheses amongst the many.

See the examples from Feyerabend given in my reply to @Janus, just above. An examination of the history of science shows that it does not follow the supposed prescription you provide; and indeed, that scientific progress is dependent on breaking those conventions.

The point I would press here is again that what makes science work is not a series of logical rules, but a group of sociological rules. It's not a special type of logic - induction or abduction - that makes science effective, but the open interplay between scientists.

A neat example that supports the hypothesis that "abduction" - understood as accepting the best hypothesis - is central to scientific method. That's what confirmation bias entails. Popper taught us to look also for examples that run counter to our expectations - to look for falsifications. Let's do that.

I gave the example of the comparison between the Copernican and Ptolemaic systems, above. The first renderings of the Copernican system were at best no simpler or more accurate than those of the then quite well developed Ptolemaic account.

Brownian motion was noticed in pollen grains in 1827. For decades, it was accounted for as at first evidence of a vital force in the pollen, then when it was demonstrated in non-organic particles, as evidence of eddies in the fluid. It wasn't until Einstein provided a statistical analysis of Brownian Motion as the result of collisions with other particles that it was linked to molecular theory - the maths made Brownian motion a testable, quantitative prediction of molecular theory. Feyerabend pointed out that the acceptance of this explanation is thus theory-laden, dependent on the availability of a mathematical interpretation in order to allow empirical confirmation.

Bringing this back to your post, what this shows is that what counts as providing the greater explanatory power is dependent on the ad hoc and auxiliary hypotheses employed. There is no simple way to compare competing hypotheses, since each hypothesis brings with it a differing account of what observation is relevant.

The surprising fact, Brownian motion, is observed.
But if there are eddies in the fluid, Then Brownian motion would be a matter of course.
Hence, there is reason to suspect that there are eddies in the fluid.

Abduction here leads away from the better answer!


Considering relativity again, the procession of Mercury was explained by an otherwise unseen additional planet - which was even named "Vulcan". It had to exist, becasue it was the best explanation for the observed phenomena - exactly in line with Abduction:

The surprising fact, the procession of Mercury, is observed.
But if Vulcan were true, the procession of Mercury would be a matter of course.
Hence, there is reason to suspect that Vulcan exists.

But this abduction was mistaken!

What these examples show is not just that abduction is sometimes mistaken, but that it leads to a lack of progress, and that other, wildly differing background assumptions are instead needed to progress our understanding.

Notice that in each case, abduction leads to the confirmation of the accepted paradigm, where what was needed was a change to that very paradigm. Abduction as a counterproductive process.

A nod to @Joshs.