10

It claims there are three options of which none of them are satisfying.

Circular argument doesn't prove anything because it's just when the premise is the same as the conclusion.

x ∵ x

Infinite Regress isn't clear on why we'd need to do this let alone even possible to implement so it also has no proofs.

x ∵ ... (never ending chain)

Foundationalism is different in that it's actually useful. Mathematics is its epitome. Math has tons of proofs. Math isn't only cohesive but also adhesive. What more do we want?

y ∵ x; where x is presumably true

As for it's criticism: Why start at claim x?

Because presumably, it's either true (via self-evidence or empirical research) or it doesn't matter because sometimes we only want to see where the premise takes us.

Maybe an interesting take on my question is: how do we know it's a trilemma?

QWERTY_dw
  • 701
  • 1
  • 6
  • 10
  • 3
    We do not need to "really know", it is an argument against those who claim that we can know (typically, foundationalsts), so their premises are accepted for the purposes of reductio. And those are more than enough to make an exhaustive list of alternatives, however it is itemized. Agrippa's version had 5 items instead of 3. – Conifold Jul 17 '19 at 21:45
  • As a consideration... sometimes we don't want to see where the premise takes us. Following a premise takes time and effort. While mathematics may have taken an approach which rewards foundationalism, it can be disconcerting in other disciplines, such as martial arts, where there's a distinct possibility that your entire martial art was founded on an assumption that was wrong. People who arrive at that conclusion often feel like they've wasted 10 or 20 years on an art. – Cort Ammon Jul 17 '19 at 23:27
  • See also https://philosophy.stackexchange.com/questions/7325/why-is-the-m%C3%BCnchhausen-trilemma-an-unsolved-problem – J.G. Jul 18 '19 at 10:33
  • 1
    It's "its", not "it's". – tomasz Jul 18 '19 at 12:46
  • 2
    I'd argue that all three forms can support a proof and deliver some value. But none of them is satisfactory, in that none of them is undeniably true, but require some compromise. The first one can prove a theory to be coherent. The second one can prove a theory to be without holes, since it stands on an endless tower of regression. None is really worse than standing on a random assumption. – Falco Jul 18 '19 at 15:36
  • 1
    @Falco Can you give an example of a circular proof and an infinite proof? – QWERTY_dw Jul 18 '19 at 16:51
  • 1
    @Falco But there's an important distinction between "random assumption" and "presumably true assumption". – QWERTY_dw Jul 18 '19 at 16:53
  • @user40358 As far as I understand it, this distinction does not exist for a radical skeptic. Either something is provably true, or it is uncertain - there is no "probably true" for the purist. – Falco Jul 18 '19 at 16:56
  • 1
    @Falco That's okay with me. Because radical skeptics can object to basically anything. They're like that annoying facetious kid at the back of the classroom. Their arguments are not useful, marginal and absurd as far as I'm concerned. – QWERTY_dw Jul 18 '19 at 17:02
  • @user40358 but solving the trilemma would mean beating radical skeptics on their own turf – Falco Jul 18 '19 at 17:12
  • I struggle to find examples which are neither trivial nor to big. Modern physics theories could be an example of a system of circular proofs - empirical evidence sets boundaries, but I've could formulate a million theories to fulfill all known evidence, but the scientific community chooses the most simple theory, which is coherent. Although the line is blurry, if I choose any statement of a circular proof as an axiom it becomes a foundational assumption based proof. But usually the theory is not standing on the foundation of experiments, but rather experiments inspire a theory and can test it. – Falco Jul 18 '19 at 17:27
  • Also keep in mind an axiomatic system (foundationalism) is necessarily limited in domain of application, i.e. by definition it does not give ultimate answers. – christo183 Jul 19 '19 at 08:53
  • The literal way that we know that it is a trilemma is that it has three parts. If what you really means is how we know that the third one is valid, that is a different question. – polcott Jul 23 '20 at 21:14
  • The Peano axioms define operations on the defined set of natural numbers forming mutually inter-locking self-defining semantic tautologies. Successor(Successor(0)) > Successor(0) is known to be necessarily true and thus impossibly false on this basis. There is no infinite regress. The knowledge tree is acyclic. – polcott Jan 17 '21 at 21:44
  • Agrippa's/Münchaussen's trilemma is a good argument, undermines in one go the entirety of the philosophical enterprise. Genius at another level. – Agent Smith Mar 01 '23 at 07:41

6 Answers6

14

You have misunderstood the point of the Munchausen Trilemma. It plays a key role in the process of philosophy showing that none of our beliefs are justified knowledge, per the standards of "reasoning".

Most people hold that they have knowledge and beliefs based on justified reasons, and that beliefs SHOULD be justified, and knowledge isn't knowledge unless it is justified somehow. The term for this is the Principle of Sufficient Reason: https://plato.stanford.edu/entries/sufficient-reason/

In its first usage, by Spinoza and Leibniz, the PSR basically called for a proof, before one could satisfy it.

Another key term in the history of knowledge is Justified True Belief -- the criteria that knowledge is only knowledge if one has both justifications, AND it is true. https://plato.stanford.edu/entries/knowledge-analysis/

However, most empiricists since Locke have been indirect realists -- holding that the world is only inferred, not known directly. And if we cannot know the world directly, then one cannot EVER know of a belief about the world is TRUE or not, and JTB is unachievable. https://www.iep.utm.edu/perc-obj/#H2

In practice today, most "reasoning" people follow a softened version of the PSR -- where "sufficient" reasons are just "supporting justifications are stronger than refutations".

However, one can apply the Munchausen trilemma to all "supporting justifications" to challenge what THEIR support is. And the answer, in every case, because we cannot complete an infinite series, will lead to and UNJUSTIFIED supposition, or a circular argument.

If beliefs need to be justified to be reasonably held, then all those justifications, to be held reasonably, must themselves be justified. But they are not and cannot be. The Munchausen Trilemma refutes all claims to be "reasonable" or hold "justified beliefs" based on our current standards of knowledge or reasoning.

The response among many philosophers has been to embrace larger networks of supporting assumptions, this is the coherence reply to this dilemma. https://www.iep.utm.edu/coherent/ The reasoning is that while a simple circle might be a fallacy, a complex web of justifications is not. The name of the Munchausen Trilemma, which critiques the circularity of coherentism, is a ridicule of this claim. Baron von Munchausen cannot pull himself out of the mud, or pull his horse out of the mud, but by pulling on his hair, then lifting himself this way, he was able indirectly to lift his horse through the stirrups! Make the circle big and complex enough, and one can do the apparently impossible with it.

If no beliefs are justified by our current form of rationality/reasoning, then philosophy and knowledge must both become non-rationalist. The response to this has varied between Europe and the US. In Europe, it has been an embrace of radical relativity -- IE postmodernism. In the US it has typically lead to pragmatism, where one can justify beliefs based on the pragmatic criteria of their utility/effectiveness.

Elaborating on the consequences of the pragmatic re-evaluation of reasoning and rationalism. This involves accepting that truth is not an absolute -- but only "approximately true", and that reasoning is not itself justified, or always valid, hence logic problems such as circularity, or unjustified assumptions, are not fatal problems for a assertion or justification.

Hence flawed justifications, such as the Thomist "5 Proofs" of God -- showing they are not actually "Proofs" does not refute Thomism as a POV. One needs to evaluate its pragmatic effectiveness/value in explaining the world, and the revival of Thomism in recent years shows it is pragmatically useful.

And while Descartes' Foundationalism is actually questionable (there have been challenges on selfhood, much more widely accepted challenges on God, and in particular challenges to the trustworthiness of our reasoning) he still assembled a pretty good foundational case, and much modern thinking is still indebted to him due to the usefulness of his argument.

The cohenrentist response to Munchausen's Trilemma may be logically circular, but the argument that we HAVE to be circular to build a worldview at all, is a highly useful and powerful one anyway.

Similarly, the entire basis of the validity of the scientific/empirical method to attain knowledge is an effectiveness argument -- IE it uses empirical justifications to validate empiricism -- in fairly explicit circularity. Yet science is demonstrably useful, hence the logically invalid justification is -- pragmatically still valid.

Dcleve
  • 13,610
  • 1
  • 14
  • 54
  • We can know that some expressions of language stipulate relations between conceptual classes and this stipulated relation is itself an element of the set of knowledge. The axiomatic argument of the Munchausen’s trilemma fails to acknowledge that brand new knowledge is created by stipulating relations between conceptual classes. – polcott Jul 23 '20 at 18:15
  • Thus there only can exists useful fictions. – Rieke Dec 14 '22 at 18:13
  • "none of our beliefs are justified knowledge" so maybe "dogs" are a kind of "office building" ??? – polcott Mar 01 '23 at 01:52
12

Circular argument

We know it's a trilemma because the argument is founded on logic and proofs, and all proofs will end in either circular logic, infinite regression, or a foundational assumption.

Infinite regress

You can always break a proof into parts. Those parts get simpler and simpler. Keep breaking them up long enough, and all parts will eventually become circular logic, infinite regression, or foundational assumptions.

Foundational assumptions

We know it's a trilemma.

All kidding aside, we find this pattern is generally true. Logical arguments end up in one of these buckets. Sometimes you can change which bucket it ends up in (such as calculus, which permits us to replace some infinite regression arguments like Zeno's paradox with foundational assumptions about how limits behave), but they end up in one of these buckets in the end.

The main value of the trilemma is to provoke thought. Its to make you think about what must come of rational thought. If it does so, then it has done its job. Proving it "right" however, is more difficult. The whole point of the trilemma is that people who believe in it are not satisfied with the result of any attempt at a proof.

To prove it, you would have to select what concept of "proof" you wish to explore and what tools you are willing to consider in the proof thereof. Those are rather personal, which means there's not a one size fits all answer, besides the satirical version I wrote above.

I would argue that you have a clear opinion as to what proofs you find appealing. You appear to find the foundational ones appealing while the others feel like "not an argument." Consider, however, the infamous Turtles argument:

The following anecdote is told of William James. [...] After a lecture on cosmology and the structure of the solar system, James was accosted by a little old lady.

"Your theory that the sun is the centre of the solar system, and the earth is a ball which rotates around it has a very convincing ring to it, Mr. James, but it's wrong. I've got a better theory," said the little old lady.

"And what is that, madam?" inquired James politely.

"That we live on a crust of earth which is on the back of a giant turtle."

Not wishing to demolish this absurd little theory by bringing to bear the masses of scientific evidence he had at his command, James decided to gently dissuade his opponent by making her see some of the inadequacies of her position.

"If your theory is correct, madam," he asked, "what does this turtle stand on?"

"You're a very clever man, Mr. James, and that's a very good question," replied the little old lady, "but I have an answer to it. And it's this: The first turtle stands on the back of a second, far larger, turtle, who stands directly under him."

"But what does this second turtle stand on?" persisted James patiently.

To this, the little old lady crowed triumphantly,

"It's no use, Mr. James—it's turtles all the way down."

Now you don't have to agree with the opinions in this story, but you must admit that there something worth calling a "proof" here, and it ends in infinite regress. As a thought experiment, how would you go about responding to this? Would you be inclined to tell the little old lady that it is not valid to use infinite regression in proofs?

Consider how we model numbers with Peano arithmetic. We always have an axiom of induction. We may call that simply a foundational axiom, but if we look into why we think it is a valid axiom, it starts to look an aweful lot like an infinite regression argument. We just tucked it away inside a foundational axiom so that we didn't have to worry about it polluting the rest of the proofs. So, perhaps, in a way, such fundamentals of arithemtic simply prove that infinite regress is a valid method of proof in some carefully constructed circumstances!

Community
  • 1
Cort Ammon
  • 17,775
  • 23
  • 59
  • I think the relation between infinite regress and induction is only superficial. If some statement of the natural numbers can be proven by induction, then if you give me a particular natural number*, the proof the statement for this particular number has only a finite number of steps. Can you elaborate how your motivation of induction leads to infinite regress? – Discrete lizard Jul 18 '19 at 07:09
  • *Note that by this I mean a number from some finite subset of the naturals (e.g. a number of which the digits fit in a comment), which is something very different from any natural number. Indeed, to prove the claim for the latter you would need induction... It seems that the circular argument comes closer to my belief in induction. – Discrete lizard Jul 18 '19 at 07:09
  • @Discretelizard The axiom of induction is itself the infinite regress, just tucked away in an axiom. It says that I can prove that for some function f, f(n) is true for every natural number. Why? Because f(0) is true, and if f(0) is true then f(1) is true, and if f(1) is true then f(2) is true, and if f(2) is true then f(3) is true, and we'll just use that logic over and over to prove f(n) is true for all n. – Cort Ammon Jul 18 '19 at 15:26
  • Once you accept the axiom of induction as valid, then the infinite regression gets squirreled away in an axiom which states that if f(0) is true and f(n+1) is true if f(n) is true, then f(n) is true for all real numbers. Now it's finite... but the logic behind it was infinite. – Cort Ammon Jul 18 '19 at 15:27
  • (I may not have been clear. Just in case, the particular formulation of induction I am referring to is the second order axiom in Peano arithmetic. As far as I know, other forms of induction tend to validate themselves using the Peano arithmetic axioms) – Cort Ammon Jul 18 '19 at 15:30
  • 1
    You write "Because f(0) is true, and if f(0) is true then f(1) is true, and if f(1) is true then f(2) is true, and if f(2) is true then f(3) is true," and this indeed leeds to requiring an infinite number of statements for the full proof. But none of the statements you have written down depends on a statement that has not. In particular, if I pick a number, say 1000, then the method is in fact a valid proof for all numbers up to 1000, without needing induction as an axiom. This is different from the e.g. the turtle example, where we have proven nothing if we stop early. – Discrete lizard Jul 18 '19 at 16:49
  • So, I think this justification of induction is different from the examples of infinite regress listed here. From what I understood of the definition of infinite regress here, I was under the impression that infinite regress is not just an argument requiring an infinite number of statements, but one where each statement depends on one not yet proven. – Discrete lizard Jul 18 '19 at 16:49
  • @Discretelizard You say "valid proof for all numbers up to 1000." The axiom of induction is "valid proof for all numbers," period (without using an infinite amount of paper to write down all the individual proofs). However, I do see a subtle distinction that you may be getting at. Perhaps it's that you prove f(n) using f(n-1), rather than using f(n) to prove f(n+1). Might that be the difference? One rages forward proving things in its wake, while the other dives towards f(0), blindly trusting that it will get there eventually! – Cort Ammon Jul 18 '19 at 18:08
  • Yes, the point of the axiom of induction is that you can prove something about an infinitely large set, you don't need it for a specific finite set. "Perhaps it's that you prove f(n) using f(n-1), rather than using f(n) to prove f(n+1). Might that be the difference?" Yes, exactly. – Discrete lizard Jul 18 '19 at 18:15
  • 1
    @CortAmmon "if f(0) is true and f(n+1) is true if f(n) is true, then f(n) is true for all real numbers." Induction works only on well-ordered sets, which the set of real numbers is not. – Acccumulation Jul 18 '19 at 20:58
  • The semantic meaning of the "product" of decimal integers 10 x 40 is defined by its algorithm to be 400. – polcott Jul 24 '20 at 18:23
5

The trilemma is about justification of a given proposition. Any justification, so the story goes, takes ultimately one of these forms if faced with skepticism.

Therefore, the third option is about people who answer to the question "But how do you know that x really is true" dogmatically, e.g. with "Because it is", "Because I say so", etc.

Ultimately, the answers "It is self-evident" and "Empirical research shows it" are, if taken as an end-point of justification, arbitrary since the skeptical objections "Why do you assume self-evidence here, I do not follow" and "What are the standards of this research, why should this be a particularly good standard for the truth of the proposition" are still valid.

Sure, we accept that for statements about the world, empirical research is probably the gold standard and brings us as close to the truth as we might get, but that's not the setting in which the trilemma is supposed to work.

It is about forms of justification, and ending the chain of justification at an arbitrary point is dogmatic, no matter how well-justified the point itself might be. If I claim that this justification was unquestionable and ultimately justifies the whole chain as true, I am dogmatic.

Philip Klöcking
  • 13,815
  • 3
  • 40
  • 72
  • 'We know nothing' is an untenable proposition though. – Joshua Jul 18 '19 at 19:29
  • 2
    @Joshua And nobody ever claimed as much, nor is the trilemma. This is all about apodictic certainty and how any justification of such would run into these forms. The Münchhausen Trilemma objects against logical necessity, not pragmatic reasonability. – Philip Klöcking Jul 18 '19 at 19:59
  • I am absolutely certain of your existence as well as mine. – Joshua Jul 18 '19 at 20:03
  • 1
    You shouldn't be. ;) – Katie Kilian Jul 18 '19 at 20:51
  • 1
    @Joshua See, the question is not about your conviction, but your justification thereof. Thus, you can be absolutely certain of whatever you want, as long as you do not or cannot further justify this conviction, you effectively fall under the third case. – Philip Klöcking Jul 19 '19 at 09:18
  • I picked the example I did because you would not be able to construct the hypothetical world in which it is not true. – Joshua Jul 19 '19 at 19:05
3

The first and main point to understand about the trilemma is that it is an argument. As such, its function is to convince other human beings, at least to the extent that they are rational.

The trilemma is an argument about knowledge and for this reason is often misunderstood as proving the impossibility of any knowledge.

This, however, is a logical impossibility. Either the Trilemma is itself knowledge or it isn't. It couldn't possibly be knowledge, however, since if it was, it would be a counterexample to its own assertion.

And if it isn't knowledge, then, why should we care about it?

We need to care because the Trilemma is effective as a rational argument against a certain idea of science as objective knowledge the physical world.

Any idea of science as knowledge of the real world which assumes that subjectivity can and should be entirely left out of science is indeed condemned to an infinite regress. And a good theoretical model of this is indeed the theory of Justified True Belief.

Justified True Belief either leaves you not knowing whether a justified belief is really true, because of the circularity of the theory, or requires that you produce a second-level justification to justify all the elements coming into the initial justification, and then of course to produce a third-level justification to justify all the elements coming into the second-level justification etc., ad infinitum.

Justified True Belief actually describes quite well the history of science, where each generation of scientists produces a new, more detailed justification without being able to say whether the last science is actual knowledge of the real world.

The Trilemma is thus effective against any Justified True Belief view of science as knowledge of the real world. It is effective, however, only to the extent that it is a convincing argument. It is still not, and won't ever be, knowledge.

As a convincing argument, it probably has some effect, up to a point, on how people think about their own practice, scientists in particular.

However, the fact that it is a convincing argument about the absence of knowledge of the real world, in particular scientific knowledge, doesn't say anything at all as to the possibility of knowledge or indeed about whether we know at least some things or not.

Suppose you know X. What the Trilemma says is that you cannot justify that you know X. However, what the Trilemma doesn't do, emphatically, is to prove that you don't know X. Again, the Trilemma is not knowledge and therefore isn't knowledge that you don't know X. It is not even an argument that you don't know X. It is only the argument that you cannot "satisfactorily" justify that you know X. So, if you know X, of course you do, and the Trilemma be damned.

Still, whether you really know X or not is of no interest to anybody but you. What matters to other people is to know themselves whether you know X. However, the Trilemma is this very convincing argument that whatever you will ever say, other people won't be convinced that you know X. And why should they?

Speakpigeon
  • 7,364
  • 1
  • 11
  • 26
1

The dogmatic argument, which rests on accepted precepts which are merely asserted rather than defended https://en.wikipedia.org/wiki/M%C3%BCnchhausen_trilemma

Much of knowledge is stipulated relations between finite strings. The only way that we know that "dogs" are "animals" and not "office buildings" is on the basis of the set of properties that have been assigned to give those terms their meaning.

The entire body of analytic knowledge is only known to be true on the basis of these stipulated relations between finite strings. It is widely accepted that analytic truth is verified entirely on the basis of the meanings of the terms alone.

This is the same sort of proof that I had in Geometry 50 years ago. One of the proof steps was to cite a definition. We can say that we know that "dogs" do not have "windows" on the basis of the definition of a "dogs" and a "windows". It is not merely asserted that "dogs" do not have "windows" it is conclusively proven.

polcott
  • 481
  • 1
  • 3
  • 20
1

Let's turn the trilemma on itself just for the fun of it.

  1. Either the trilemma is itself satisfactorily justified or it is not.
  2. If it is not, then either no justification is needed, else it is unjustified and of no consequence to justification of other claims.
  3. If no justification is needed, then this is same for other claims and the trilemma is of no consequence.
  4. If it is justified, then either it is justified through one of the three options, or some other way.
  5. If it is justified through one of the three options a) circularity, b) infinity, c) foundation, then at least this option is a satisfactory justification for a claim and the trilemma is of no consequence.
  6. If it is justified through some other way, then this way is a satisfactory justification and again the trilemma is of no consequence
  7. Either way the trilemma is of no consequence to the satisfactory justification of claims.

QED

Nikos M.
  • 2,676
  • 1
  • 11
  • 19