54

Intro

The scientific method is a key process of how we acquire knowledge and may shape our understanding of the world. If I am not mistaken, this method has been defined several times during our history.

The scientific method is not free from disadvantages. I don't want to talk about practical issues such as the statistical bias caused by the fact that work that demonstrates effect gets more easily published than work that demonstrates absence of effect. I am asking about the fundamentals of the scientific method.

Broad Question

I am asking whether anything better than the current scientific method could theoretically be achieved? Is there any reason why the current method has to be the best method for acquiring knowledge or can we imagine anything better?

Going into the specific

Typically, I am curious about the process of proving the hypotheses wrong. If I am not mistaken, no hypothesis can be proven correct, we can only prove hypothesis (uncountable) wrong. Such methodology very much follows the method of hypothesis testing in statistics. Taken from my memory of the work I read from Elliot Sober, a long time ago, this method leads to the important issue that one can only discard all the hypotheses that (s)he can imagine, but can never be sure that he would not have missed some hypotheses. One can never prove anything right. One can only prove things wrong.

Can't we imagine a method that would not be based on rejecting hypothesis? Is there a fundamental restriction in how one can acquire knowledge that forces us to use this seemingly sup-optimal method?

16
  • 8
    Why is it sub-optimal? You note only, from Sober, that scientific method can't reach certainty, because nothing can. Is there some other reason it's sub-optimal? If not, then the answer is very short: no. Apr 16, 2015 at 11:34
  • 15
    Traditional Chinese Medicine, I assure you, does not do a better job than scientific methods do on that particular scale: achieving certainty. Science does not achieve certain knowledge, but nothing else does better. This is because scientific methods ARE the methods that achieve such certainty. If TCM were able to generate accurate knowledge, its methods would BE scientific (and are to the extent that it does). Apr 16, 2015 at 20:36
  • 3
    "The scientific method is a key process of how we acquire knowledge..", um, well, no. Technically, it's how we acquire empirical knowledge. The scientific method is nigh-useless for the acquisition of non-empirical knowledge and as a consequence, we use other methods/approaches for those. For instance: Philosophy. Apr 16, 2015 at 20:45
  • 6
    Actually, the scientific method cannot even prove something wrong. It can only quantify the probability with which a theory agrees with observation. There are some agreed-upon probabilities after which people call a theory false (the 5 σ from particle physics for example), but that's not an absolute falseness either.
    – Turion
    Apr 16, 2015 at 23:15
  • 3
    Just as an observation, it would probably help to focus your question on the "Precise" aspect of it and to simply not try at this stage to try to treat your "precise" problem as a more general problem just yet. Your question is "Falsificationism does not yield definitive knowledge because it always only resolves logically formed hypotheses negatively. Is there any kind of epistemology that avoids this problem?"
    – Paul Ross
    Apr 17, 2015 at 10:39

16 Answers 16

33

(Edit: this answer is now split into two parts, thanks to a lengthy discussion with Rex Kerr. I made my original answer on a very specific reading of the scientific method. He had a very different reading, which came to a different but very related outcome. I've tried to capture that in the first part. The second part is my original answer, for those who wish to use the stricter reading)

There are at least two extremes as to how one can define the scientific method. One is a process, one is more of a set of principles and a goal. The process is well defined as:

  • Observe something interesting
  • Formulate a hypothesis that you think would model this interesting thing better than existing model.
  • Run a series of independent tests of the hypothesis.
  • Statistically demonstrate that the original model (often called the null-hypothesis) predicts the outcomes of the tests to be highly unlikely.
  • Reject the null hypothesis (assuming the data backs your claim)
  • Demonstrate that your new model does a better job of predicting the statistical results.

This is what I was taught the scientific method was in high school. If that is the version you are after, skip ahead to the second part, which explicitly targets that reading.

However, there is another more fluid reading which also exists. The statistical requirement is relaxed, because it can cause trouble. However, there is a focus on both the elimination of hypotheses through testing and the preference towards hypotheses which are testable. This reading of the scientific method is a very general direction, so the alternatives are equally general.

Science is a very deductive learning approach. It depends upon one writing a hypothesis in highly objective terms and then testing it. There are many situations where deductive learning does not work. Procedural learning is often viewed as an alternative approach. Consider the case of an athlete. They collect large amounts of information from scientific approaches, but the final bit that takes them from a "good athlete" to a "great athlete" is all "feel." There may be no written hypotheses. There may be no statistical testing. Yet, the mind absolutely learns in this way. Thus, procedural learning like this would be a valid alternative method. In fact, many Chinese martial arts focus almost entirely on procedural learning because it is so hard to learn deductively.

Which reading of the scientific method you want to use is up to you. What follows is written entirely from the perspective of a strict statistically valid approach to the scientific method. Much of what has made science great is its ability to build upon previous hypotheses. While statistical rigor is a nicety for rejecting hypotheses, it becomes essential for building hypothesis which can support others.


Finding an alternative to the scientific method depends on you deciding what you want out of a method. You will never find a better tool than the scientific method at it's game. However, if that game is not what you really want it to be, there are alternatives.

The most visible example of this I have seen is western medicine compared to Traditional Chinese Medicine. They developed very different approaches, and yet both appear to yield results. TCM actually does work on line along your questioning: it is not fundamentally built off of rejecting hypothesis.

Let's look at the scientific method, and see if we can make some headway. There are two major features of the scientific method which stand out as "interesting" for this line of thinking:

  • The scientific method is highly steeped in the language of statistics.
  • The scientific method seeks objective theories.
  • The scientific method tests theories.

These are traditionally seen as strengths. However, they can also be seen as weaknesses (like all good superheros, their strength is their weakness.. that's what makes them interesting).

The scientific method is completely and utterly useless without statistics. This means any singular event is completely beyond its reasoning. It cannot provide answers to topics such as "the purpose of your life" because there is only 1 you, and N=1 means there is no statistics.

Related to this, the scientific method strives to be objective. It always tries to remove the observer from the picture. This is very valuable, because it ensures that your discoveries are applicable to others. However, it also proves to be tricky in many situations. Social studies in particular have great difficulties with the scientific method because it is so very difficult to make good tests that keep the observer out of the loop. As an example, TCM claims that acupuncture works. Those who have tried it, claim it works with uncanny success. However, science has had fits trying to find any effect of acupuncture beyond the infamous cop-out "the placebo effect." The issue is that it is almost impossible to develop an effective control to measure against because the acupuncture practitioner knows if they are doing it right or not. Whether you believe acupuncture works or not depends heavily on whether you accept results which lack a solid control to ensure objectivity.

Finally, science tests its theories. This sounds absurd, because it seems so obvious that you should test them. However, a theory is not accepted at all until it is tested. The result is that anyone with a theory must expend the resources to do the testing before science will do anything with it. Other approaches get away with a different style: you use a theory once you have it, and you test it when you get an opportunity to do so. The tests can also be dangerous. (Edit: I had a reference to the LHC and potential to create black holes here, but it was too contentious. Instead, it has been replaced with a hypothetical example) Consider a hypothetical particle physics experiment. The scientist is rather confident that their theory is correct. They begin experimenting, after calculating that they would like 100 samples to do statistics on. Generally speaking, they are finding their theory holds out for test after test. However, on tests which disagree with their hypothesis (which happens in the scientific method due to noise), the observer notices a burst of energy from the test apparatus. That burst becomes stronger and more dangerous with every data point that disagrees with their hypothesis. At some point, the scientist decides to cut the experiment short, because they are uncomfortable putting their life at risk to finish the test. By the strictest reading of the scientific method, that data cannot be analyzed because it is tainted with the scientist's choice to cut the tests off early. This might induce biases because the scientist is more likely to cut them off faster if the results look good for their theory. Other methodologies are capable of using this data (including the intuition of that scientist, who will not try the exact same experiment again).

Seeing that the strengths and weaknesses of science are so confounded, it is up to each individual to decide if those are ideal for them. There are many others, none so visibly different from the scientific method than that of TCM. As described to me in a lecture, the difference is in the approach towards healing the human body:

  • Western medicine tears the body apart into components, develops hypotheses about these components, then builds them up. At each step, it develops testable hypotheses, and tests them. From there, it finds things which may provide results, and tests those.
  • TCM starts with the body as a whole, finds things that cause good results, then develops testable theories about why the results occurred.

The end result is that much of TCM is doctor-centric. A doctor finds out what works well for them, and suggests it to others. The focus is less on rejecting bad hypotheses, and more on finding new good hypotheses. TCM relies more on natural attrition to weed out the hypothesis, rather than actively trying to disprove them.

Can I claim one is better than the other? I'm not sure if I can. However, I do feel comfortable claiming that they are different, and that a remarkably large number of individuals consider one better than the other in both directions. It's simply another way to approach things.

20
  • 14
    Western medicine is not a good example of scientific methodology. I say this because you should differentiate between biology, pharmacology, testing, and how medicine is practiced by doctors on a day to day basis. Doctors diagnosis symptoms in a patient and often mis-diagnosis symptoms. They rely on their experience and 'best guess' more times than often. Apr 16, 2015 at 8:22
  • 4
    @SwamiVishwananda I use western medicine as an example because it is in that fuzzy region where things break down, thus there is a useful comparison to other methods (like TCM). If I were to concentrate on a region where the scientific methodology is at its best (such as physics), there would be little to no competition to the scientific methodology, so it would be hard to find a methodology good enough to be deemed "an alternative to the scientific method."
    – Cort Ammon
    Apr 16, 2015 at 15:03
  • 3
    One fun fact about the LHC. Scientists have known since long before the LHC was designed that cosmic rays produce explosions in the upper atmosphere of Earth that are millions of times stronger than those produced in the LHC. Thus, we knew before LHC was built that if the LHC were capable of destroying the Earth, it would have already been destroyed a long time ago. That sounds like a pretty good safeguard. Apr 16, 2015 at 16:28
  • 10
    @CortAmmon - Also, the acupuncture point is almost equally bad. You can of course test the hypothesis that being attended by an acupuncturist will help your pain, and compare that against being attended by various other people doing various other things. If TCM says "acupuncture provides relief from XYZ" and science says "acupuncture, delivered in the traditional TCM way, provides relief from XYZ but we can't find strong support for the hypothesis that it works for the reasons the practitioners say it does", that is hardly showing that the scientific method has limitations.
    – Rex Kerr
    Apr 16, 2015 at 23:00
  • 8
    Due to science, modern TCM practitioners have largely abandoned qi, meridians, yin and yang, energy flow etc. as explanatory frameworks, so it seems odd to use TCM as an example of a realistic alternative to science. There is also little agreement, even amongst TCM practitioners, of diagnoses or treatments. "A doctor finds out what works well for them, and suggests it to others" is, funnily enough, also the method of witch doctors and faith healers... which is not a credible alternative to science.
    – bain
    Apr 17, 2015 at 19:49
28

If I am not mistaken, no hypothesis can be proven correct, we can only prove hypothesis wrong.

This is known as "falsificationism". It viewed with much scepticism by today's philosophers of science. The author you mention, Elliot Sober, has suggested that it be retired, deriding it as "Popper's f-word" (referring to Karl Popper, whose own views on this subject shifted somewhat over time).

It is untrue in the exact way you described it, because if a hypothesis A is wrong, then there is another hypothesis, called A-is-wrong, that is correct. So if it is possible to prove some hypotheses wrong, it must also be possible to prove some hypotheses correct.

But the real problem is that you are speaking in terms of absolute certainty, so you are crediting the scientific method with powers no method could possess.

All evidence has multiple (perhaps infinitely many) possible explanations. One interpretation is always available: the evidence might be flawed, in which case it can be ignored.

So all knowledge is conditional. Outside of mathematics, knowledge is conditional on unreliable evidence. But even mathematical deduction produces theorems that are true only because other theorems on which they depend have been shown to be true, and ultimately the whole edifice rests on basic assumptions (axioms) that are simply assumed to be true. Or rather, all mathematical truth is conditional on the truth of the axioms.

So we need a way to compute the certainty of a deduction based on the certainty of the facts that deduction relies on. That's what probability theory is.

Not for nothing did Laplace observe that the whole system of human knowledge is tied up with probability.

15
  • 4
    Comments from downvoters always welcome! Apr 16, 2015 at 16:22
  • 7
    Your paragraph with A-is-wrong gets tricky in situations where there can be infinitely many hypothesis. It may require infinitely many rejections of hypotheses A, A2, A3, etc. before proving A-is-wrong is correct. Some bounded problems can show that there are not infinitely many hypotheses, but in other cases, it is remarkably hard to bound them as such.
    – Cort Ammon
    Apr 16, 2015 at 21:30
  • 7
    Even in mathematics, there's also the possibility that everyone examining the proof so far has missed a mistake.
    – Dan Bryant
    Apr 16, 2015 at 21:32
  • 1
    @DanBryant - indeed, and that situation can last a long time; Euclid wrote the textbook on geometrical proofs that was required reading for all educated people for about 2300 years, but in the late 1800s people were still discovering axioms that he had inadvertently relied on without stating. Apr 16, 2015 at 22:17
  • 3
    @DanielEarwicker Ahh, I think I see what I was getting wrong. I was not looking at the trivial version, where the mere act of rejecting A is accepting Anything but A. Often the hypotheses constructed in such way are less valuable than one would want (arbitrarially valueless in some cases). I was looking at the case of "if you reject enough hypotheses, you eventually run out of false hypotheses," which is also phrased as "when you have proven all that is impossible, what is left, no matter how improbable, must be the truth."
    – Cort Ammon
    Apr 16, 2015 at 22:21
17

You are starting from the hypothesis that your understanding scientific method is correct and complete, and that everyone else has the same understanding. Neither of which is sustainable from the evidence here.

There are no 'weaknesses' with scientific method, publishing and grants are issues of personality not science, as even scientists have personalities. You are correct in suggesting a hypothesis can not be 'proven' true but that is its the primary strength, and not an inherent weakness as you imply.

At its simplest, scientific method is a process for producing useful explanations for how the universe works. It is a stunningly simple process with only four steps:

  1. observe a phenomenon
  2. produce hypothesis to explain it
  3. use hypothesis to predict previously unseen phenomena
  4. experiment to observe unseen phenomena.

If phenomena matches predictions, then hypothesis is useful because it describes what is there AND it led to new knowledge. If phenomena not observed or doesn't match predictions, then either experiment is insufficient or hypothesis is not useful.

It has been suggested that it should be match reality, but this is not correct (or at least is misleading) as demonstrated by both Relativistic and Quantum physics, where the theories were completely outlandish when proposed and it was years before experimentation could confirm their predictions.

It does not matter who comes up with the idea, nor does it matter who performs the experiment, so it is free from bias in that respect. Also, anybody can show that any idea is incomplete if they come up with the appropriate experiment that shows predicted results are not there or not correct, but this does not make the hypothesis false (since it was never 'true' to begin with) it simple puts limits on its usefulness. Newtons laws of motion will put a satellite in space, so they are clearly useful, but if you want accurate GPS then you need to use the improvements and refinements proposed by Einsteins theories on relativity.

Footnote - Please notice that there is no mention of the 'S' word anywhere in this answer. Contrary to popular belief, it is not a core feature, it is just a very useful tool, one of many.

2
  • What do you mean by statistics not being a core feature? (is that the S word?). I'm not claiming it is nor it isn't, I'm just curious about the wording of the footnote. Aug 29, 2016 at 16:36
  • 2
    The OP was drawing the inference that because something was a certain way in statistics, so it must be in science. I was pointing out that science would happen in the same way even if statistics were never discovered.
    – Paul Smith
    Aug 30, 2016 at 13:59
13

The scientific method is simply a method for ranking theories. Logic and Theology are other methods.

When using Logic to rank theories, we are asking the question "Which theory makes the most sense?"

When using Theology to rank theories, we are asking the question "Which theory most closely matches my Holy Scripture?"

When using Science to rank theories, we are asking the question "Which theory most closely matches reality?"

The way science performs this ranking is simply by comparing the predictions of the theory to reality, i.e. performing experiments. Theories that match reality better are considered better than those that do not. People talk about "falsification", but that is really just an extreme form of ranking. Falsification is not at all necessary to the scientific method; all you need are multiple theories that experiments can distinguish. Science can rank the two theories as "better" and "not as good" without necessarily throwing either one out.

For example, the theory "Newton's Laws of Motion" has been superseded by "Einstein's Theory of Special Relativity", since Einstein's theory more closely matches reality. But that doesn't mean we throw out Newton's Laws. They are fine approximations, and in most practical situations, the two theories are indistinguishable.

Science cannot answer the question "Is the Earth 4.5 billion years old or is it much newer than that, but created to appear exactly as if it were 4.5 billion years old?" By definition, those two theories will match any reality equally well, so comparing them to reality cannot rank them differently.

However, the question "Mr. Smith has kidney stones. Will acupuncture be an effective means of treating him?" is one that science can help with. The two theories "Acupuncture is effective in treating kidney stones" and "Acupuncture is not effective in treating kidney stones" are distinct and will match reality with different degrees of effectiveness.

Furthermore, theories have inherent suppositions: the theory "Acupuncture is effective in treating kidney stones" presupposes the theory "There exists some mechanism by which acupuncture could affect kidney stones." We can test this theory by searching for such a mechanism. We can test it further by categorizing types of mechanisms that could exist and exploring other effects the existence of such mechanisms would have. Any such mechanism would have other effects which could be detected with other sorts of experiments.

One twist of the scientific method is that it sometimes does not provide the same ranking that Logic or Theology would. For example, Logic tells us that Newtonian Relativity is the most sensible theory, and Einstein's Theory of Relativity is much less logical. Logic would also tell you that Quantum Mechanics is one of the silliest and least logical theories ever devised. And yet Science tells us that these two illogical theories match reality better than the more logical counterparts.

If you want to rank theories because you want to make predictions about what will happen in reality, then Science would be the way to go. If that is not your goal, then you should pick the method that best matches your goal.

1
  • And always be careful what goals you choose, because you might achieve them.
    – Scott Rowe
    Jul 8, 2022 at 1:43
10

I think that you are confusing some things. Although statistics is a tool, among a wide choice of tools, used in the scientific method, it is not in itself the scientific method. It has also been my understanding that you use the scientific method to prove - not disprove - a hypothesis. In science, proof lies in the assertion, not the negation.

Stephen Jay Gould in his book Hen's Teeth and Horse's Toes: Further Reflections on Natural History, specifically Chapter 19: Evolution as Fact and Theory says:

In the American vernacular, 'theory' often means 'imperfect fact'--part of a hierarchy of confidence running downhill from fact to theory to hypothesis to guess...In science, 'fact' can only mean 'confirmed to such a degree that it would be perverse to withhold provisional assent.' I suppose apples might start to rise tomorrow, but the possibility does not merit equal time in a physics classroom...Evolutionists have been clear about this distinction between fact and theory from the very beginning, if only because we have always acknowledged how far we are from completely understanding the mechanisms (theory) by which evolution (fact) occurred. Darwin continually emphasized the difference between his two great and separate accomplishments: establishing the fact of evolution, and proposing a theory--natural selection--to explain the mechanisms of evolution.

So, first you have to define what is the scientific method. The first step in the scientific method is to observe and collect facts in the natural world. For instance, using the classical myth, Newton observed an apple falling from a tree. He then observed that all the apples he watched fell from the tree (none went up in space, none hovered). He then observed other items also fell to the earth. He then collected data as to rates of falling speed, etc.

He then developed a hypothesis - gravitation - that would explain the mechanism as to how things always fell. He then developed a mathematical formula that showed how gravitation could be applied to all objects. He then confirmed that his hypothesis was valid in new events (subsequent objects that fell did so at the same rate predicted by his formula).

As his hypothesis could be tested repeatedly in the real world it and became generally accepted it became theory. If his hypothesis had not been able to predict future events, it would have remained a hypothesis and never a theory. Note: Newton never tested the hypothesis that objects fell because the gods pushed them down. He never tried to disprove the theory that objects fell because the gods pushed them down, he only tried to prove his theory.

3
  • Your answer would be better if you remove the incorrect notion of science proving things. Newton did not prove his theory of gravity correct; indeed that would have been impossible given that it is wrong (false). Instead he (and others) collected evidence (empirical data) supporting his theory.
    – hkBst
    Mar 1, 2018 at 14:55
  • 2
    @hkBst Before he came up with the theory, he had a hypothesis based on observation (what goes up must come down), after testing his hypothesis against more observations he came up with the theory.. He then tested his theory to see if he could predict future observations... We know that his theory are not completely correct now based on the fact that we know that mathematics has its limits in explaining the physical world and also Einstein's theory. But it's a good approximation. His theory did not 'prove' anything. It only explained in a better way how we can measure and predict events. Mar 3, 2018 at 7:53
  • 1
    @hkBst We do not what the physical sensual universe 'is'. That is the realm of metaphysics is, not physics. Mar 3, 2018 at 7:55
5

What would being a serious alternative to the scientific method imply? To be useful, it would need to make precise and observable predictions about the material world. However, if you make predictions like that, common sense implies that it CAN be observed if the predictions actually come true, and that it SHOULD be observed, at least to check whether you fooled yourself or not. But with that, you pretty much have the basic idea of the scientific method.

Sure you could question if a prediction needs to be observable and precise to be useful. If I understand Cort Ammon's answer correctly, that's what he suggests about traditional Chinese medicine. He makes the valid point that predictions concerning human beings can hardly be precise, and that's certainly a huge handicap for sociology, psychology and medicine.

But any form of medicine that claims to be useful implicitly makes the prediction that it is able to help the majority of patients. Is that a precise prediction? (With precise I mean that is possible to check objectively if a prediction came true.) Probably not, because it is also very hard to predict precisely how the illness would have progressed without interference. And if the patient feels somewhat better that might only be a result of the placebo effect.

You might be inclined to trust in a controversial theory like acupuncture (or homeopathy), but intellectual honesty would demand that you at least consider the possibility of being wrong. Even if acupuncture were all right in principle, the therapist you choose might not know what he's doing, or even be a devious fraud. Moreover, maybe only a part of a traditional form of medicine might be "the real thing", the other part might still be the result of superstition and wishful thinking. How can you distinguish between those?

If the criterium of "somehow feeling satisfied" after an intervention that promises benevolent yet inprecise consequences is not enough to evaluate a non-scientific claim, what is?

I see three possible answers. (They also apply to magical rituals and similar stuff.)

a) You blindly trust in tradition and authority. History of mankind tells us that might not always be a good idea - not in politics, and not in science.

b) You trust what feels emotionally most appealing to you. That is not exacly a promising strategy to avoid wishful thinking.

c) You apply philosophical or poetical criteria. Sadly, history has shown that this can be quite misleading. For example the ancient greek astronomers were convinced that the planets moved in circles because they considered circles to be the most perfect shapes in geometry. But modern observation has discovered that nature dared to choose a less beautiful shape: ellipses.

The amazing thing about science is that it provides an objective way to check its own claims. You suspect that a scientist is a fool who is telling bullshit? You don't have to trust his authority, you can go and check for yourself if he is telling the truth. (Admittedly, you better not bother trying to repeat elaborate experiments like the LHC in your garage.)

In logic and mathematics claims can be checked objectively as well. But they do not provide direct knowledge about the material world. (For example, with mathematics alone you cannot decide if Euclidian or non-Euclidian geometry is the correct description of the universe, although both are "true" in a mathematical sense, i.e. logically developed out of different sets of basic assumptions.)

To conclude, I cannot see an alternative to the scientific method that has the same ability to filter out incorrect or misleading descriptions of the physical reality.

2
  • 2
    Psst, I hate to break it to you, but they're very slightly wiggly ellipses due to the gravitational influence of other planets, etc..
    – Rex Kerr
    Apr 17, 2015 at 1:42
  • 2
    Oh dear! Please don't shatter my beautiful delusions by mentioning crude reality! ;)
    – elias_d
    Apr 17, 2015 at 5:04
4

You asked:

I am asking whether anything better than the current scientific method could theoretically be achieved? Is there any reason why the current method has to be the best method for acquiring knowledge or can we imagine anything better?

Why certainly!

Imagine omniscient authority OA. OA could be a person imbued with omniscience, a deity available for questioning, an Oracle that is never wrong, Deep Thought (the supercomputer from the Hitchhiker's Guide), etc.

What would be better in terms of 'correctness' would certainly be the Consult the OA Method! Experiments are difficult and time consuming, science takes so much work. If OA is accurate, reliable, and indeed omniscient, then surely this is a 'better' method than the possibly-mistaken Scientific Method.

Various supposed OAs have arisen through history, and you will typically see some parallels with the scientific process among their followers (reliance on it, using it as a way to know truths about reality, trusting in it). Examples include the Oracle of Delphi, Moses, Jesus, Mohammad, "The Bible", Joseph Smith, L. Ron Hubbard, etc. These methods have so far fallen short for a number of reasons; their omniscience is dubious, the existence of falsified statements/claims, or the limited availability of the source.

However, I could in theory imagine an Omniscient Authority that could just immediately answer all of our questions with full knowledge and truth, which would be 'better' at determining the truth and also easier than the Scientific Method.

1
  • You can imagine quite a bit.
    – Scott Rowe
    Jul 8, 2022 at 1:47
4

I am asking whether anything better than the current scientific method could >theoretically be achieved? Is there any reason why the current method has to be >the best method for acquiring knowledge or can we imagine anything better?

This is hard to answer until you define "better at WHAT?".

But perhaps you mean something like "better at finding useful intellectual models of the real world". (Versus, say, finding more emotionally engaging explanations of the events around us; there are definitely approaches which are better at that than the scientific method, like religion and sometimes philosophy).

The answer I offer is: for practical purposes, it's rather unlikely.

But before this answer makes sense, we have to knock the Scientific Method off any pedastal of purity, any idea that it's a pure logic based procedure for discovering truth.

The practice of the scientific method is really a collection of pragmatic heuristics which have proven to be fairly efficient at winnowing hypotheses and reducing human errors though filtering what scientists accept and reject. No one heuristic or even fixed set of heuristics IS the scientific method which has proven itself effective (despite textbooks trying to simplify and regularize) - they are just tools.

(Heuristics: give more weight and trust to theories and observations which have been reproduced by other scientists; which explain more existing data; which require the fewest new assumptions; which successfully predict a new observation and especially when the prediction differs from previous theories; which come from a respected researcher with a demonstrated track record; which have been published in a quality peer reviewed journal; etc - some of these are hallowed as closer to 'pure' scientific method, but all of them inform the social process of collectively building ever more accurate and useful models of the real world, ie: science).

Not every tool is used in every case (try reproducing the Big Bang 10 times and see how difference there is), and there is some real fuzziness. The set of tools can be and often is tweaked, while staying within the practice of science. So let's look deeper.

The core behind the most important of those pragmatic heuristics is this underlying meta-heuristic: clearly observe the discrepancies between your models (theories) and the real world, and continuously adjust and expand your models to reduce that discrepancy. Put another way, science uses the measurable and observable aspects of the real world as THE standard against which all theories must be tested, and uses a negative feedback loop to successively minimize divergences.

When you use an approach to understanding and modeling the real world (or something else) without this negative feedback loop, THEN you fall outside the realm of science. We all do that - poets, lovers, even a businessman trusting their (unexamined much less unproven) intuition.

In this light, your question could be: Is there an approach to creating more accurate models of the real world, than the scientific method of paying attention to the differences between theories (or beliefs) and the real world and working systematically (using various heuristics as tools) to minimize that discrepancy?

And like science, the pragmatic (and not purely logically derived) answer is: Not likely.

It's like asking whether there's a better way to land more darts on the bull's eye than the approach of noticing how far darts are landing from it and correcting for the factors that cause the inaccuracies (negative feedback). Well, one might occasionally hit the bulls eye by tossing darts while blindfolded and unaware of how close the darts are landing, it's not likely to be a good strategy in the long run.

And this is a falsifiable hypothesis (another of those heuristics): just come up with a method which does a better job than the fuzzy but much practiced scientific method :-)

2

The scientific method allows us to make a model of reality which appears to work to a great degree of accuracy. There are, however, two major limitations: we can only model what occurs when there are many observable occurrences; and we can only precisely model what will occur when there are finite possibilities. This means we cannot answer questions such as: "What am I?"; "Was there existence before time?"; "Where, exactly, is a given particle in phase-space?"; and "Is this painting good?".

The first limitation seems straight-forward and generally the methods humanity follow here are alternative belief systems ...from the religious belief, "God was before time."; through the subjective belief "I think this painting is one of the best I've ever seen."; to the logical belief "I am a complex system of self references that happens to produce an entity capable of pondering such things.".

The second limitation seems at first to pose little threat, after all when are there infinite possibilities? However it seems that quantum physics gets limitations here (exposed as the uncertainty principle). A mathematical interpretation of the likelihood of unobservable states is that of a negative probability; this is akin to defining a region of phase-space in which the particle in question could be anywhere-when.

All our experimental methodology is based on the principle of the statistical stabilization (the law of large numbers). All experiments are prepared in such circumstances that relative frequencies must stabilize. This is the result of our cognitive evolution. In the process of evolution the brain extracted from the chaotic and (lawless) reality phenomena which satisfy the principle of the statistical stabilization (repeatability in the average). These and only these phenomena are considered by the brain as real physical phenomena. Negative probabilities give the possibility to extend the range of physical phenomena by considering phenomena which violate the principle of the statistical stabilization.

"Interpretations of probability." A. Khrennikov (1999)

A nice illustration of negative probabilities in the quantum world is described by Johannes Koelman in his blog post "Quantum Casino - Less Than Zero Chance".

Even with these limitations there is still the question of whether we can know anything (Scepticism), or even if we can know that (Pyrrhonian scepticism). However it is also often forgotten that the scientific method acknowledges that it's results may be incorrect and may revise what it has previously asserted upon observations of previously unobserved phenomena (hence adhering to Fallibilism).

Lastly there is always the possibility that there is actually no such thing as truth or that any assertion of truth is shorthand for reporting an infinite regress or even just syntactic sugar (Deflationism).

Therefore, it is my opinion that we cannot answer the question posed - it could be the case that superior knowledge acquisition methods are theoretically possible, it may even be the case that the scientific method is a stepping stone toward such methods, or it may be that what the culmination of human experience has produced is merely a reflective mask over the persona of reality, or even that the search itself is what makes reality manifest as it does for us.

Another couple of questions the scientific method cannot answer may enlighten us, "Does reality exist when unobserved?", and "Does reality abide by the same laws when unobserved as when observed?".

2
  • 1
    I mostly agree, but why can we only model what will occur when there are finite possibilities? Even the simplest real-valued one-variable model has, formally, infinite possibilities, but that's no problem whatsoever. You can test an infinite family of models, as long as you have some way to describe how well your data supports or undermines the model which doesn't require you, the scientist, to do infinite work. In fact, this is standard fare in Bayesian statistics.
    – Rex Kerr
    Apr 17, 2015 at 17:21
  • 1
    Thanks, I have qualified the statement with the word "precisely" as this is the limitation, not that we cannot model in such cases. To be honest I'm not entirely sure this is still quite correct ...I may edit further later if I can think it through clearly. In relation to the uncertainty principle the lack of precision is a mathematical result of the fact that the pair of properties in question are conjugate variables, furthermore since they are conjugate variables any model for one must therefore implicitly model the other. Apr 18, 2015 at 6:42
2

I've seen some confusion about formalizing the way that science, through the language of maths, is turned into "knowledge", so I would like to express the way I see it first.

The only absolute and universal truth we can achieve (if any) is the one that is driven by logic, and that is what mathematics is about. The subtle point lays in the fact that mathematics never actually say anything about the real world, and most importantly never states the assumptions (hypothesis) it uses are true, it just states that

hypothesis => thesis

is true as an implication. This is called theorem.

For example, we can think that if a natural number is divisible by 4, then it most certainly is divisible by 2. In this case

hypothesis: (all the axioms of natural numbers) + "n is divisible by 4"
thesis: n is divisible by 2

Every single human being on earth will agree that the implication from hypothesis to thesis is true, but that does not mean that the hypothesis itself is true.

This is basically why with mathematics only, we can understand little about the real world in term of real knowledge.

So this is where science comes into play. Science actually has the really difficult job to find the set of hypothesis that are most likely to correspond to how reality is, so that using the language and truths of mathematics, one can then derive from them a theory that can be use to predict future outcomes of a particular phenomenon.

This "finding the correct hypothesis" is actually a very difficult problem, and I like to think of it like this:

Reality is like a game being played on a chess board by some unknown rules. Science tries to understand those rules by just looking at the moves that are made.

Since the rules can be arbitrarily complicated, it is easy to see that science can never be 100% sure of anything, it can only be sure that the rules are not a certain set (hence the hypothesis can be only denied).

So the question "is there a better way than science to achieve knowledge" can be seen like "is there a better way to understand the rules of the game than looking at the moves?"

I can think of two possible alternative ways:

  1. you ask who made them
  2. you intuitively reach knowledge

The thing is, I can see a lot of problems in both approaches.

Let's suppose we are actually able to talk to God (approach 1) and he explains to us all the rules of the game, in a way that perfectly matches all our observations so far. How do we know he is not lying? We can't. So that is, too, conditional knowledge, therefore that too is "science".

Let's now suppose one reaches perfect knowledge by intuition. In this case, that person would just "know" future moves without knowing the rules, not because someone told him/her (otherwise that would be still case 1), bust just because he/she knows. This might be an alternative way to science, but the main problem is that is not communicable. There cannot be any class, teacher, book that teaches you intuitive knowledge, because it is based on nothing.

So is this a "way" to achieve knowledge? I fear not, by definition it seems more like an "event" that you happen to experience that cannot be related to any cause/effect phenomenon that could be studied by science.

In conclusion, I feel like "seeking for knowledge" itself is science under a different name, as a matter of definition.

In fact, the word "science" comes from latin "scentia" which means exactly "knowledge". The different evolution of the two words went along different paths and nowadays "science" sounds more like a discipline, a subject, but they were and they are ultimately, the same thing.

2

Typically, I am curious about the process of proving the hypotheses wrong. If I am not mistaken, no hypothesis can be proven correct, we can only prove hypothesis wrong.

The hypothesis that you refer to is its statistical definition. As someone else pointed out, any scientific statement is a hypothesis. When you reject the null-hypothesis you are accepting the alternative hypothesis.

Now, what statistical tests cannot say is which of the infinite alternate hypotheses is actually true; in other words our alternative hypothesis is always general such as m ≠ µ. There are one-tailed tests that are more stringent than this and address the direction of the inequality (either m > µ or m < µ).

This is so for statistical tests because each of them can only test if a certain observation fits a given model. There are many popular models which are the basis of the common statistical tests. Note that if you intend to test if a certain observation follows Poisson process, then you are trying to validate an observation as a case of a certain model instead of the rejection based tests.

However you should note that it is difficult to prove that a certain data follows a model because you need to verify each and every parameter. Two distributions are considered same only if all of their moments match. This is essential because there can be an infinite number of functions that produce a certain shape (a function can be expressed as McLaurin series and if all moments are same the functions ought to be the same).
Now what we regularly do is to determine if something that we observed is not just because of a random measurement error and that is why we aim to reject the null hypothesis which assumes that the data is an outcome of a Gaussian distribution model (You can see this answer in in Biology.SE for details).

I would say that you can validate a possibility instead of reject infinite possibilities if you know the underlying model and have enough data. This is how the prediction-validation-correction method works. In certain cases the model can be built using basic principles instead of inferring from the data. Finally, all models have assumptions and you need to be sure if your experiments satisfy these assumptions or not; if not you should revise the model.

Intuition or "gut feeling" is no alternative method

Intuition cannot be called a method because there is no set protocol for it. And there is no way to replicate it. Biologically speaking, intuitive guess is basically a case of applying multiple statistical tests (sub-consciously, we can say). Intuition works only when you have a good deal of prior information (sub-consciously or consciously).

2

Empiricism, in its most bald and simplest version - in Hegel, this is his notion of Sense-Certainty cannot say anything other that here it is, or there it was; one cannot move from 'a bottle' to even 'one bottle' and then to 'one'.

To do this, requires what traditionally is called induction or abstraction; and it's these two notions that sustains the traditional concept of theory-making in the scientific method when philosophically thought in a positive manner (not the negative manner of Popper - falsification).

But, as Deutsch notes in one of his popular books, this is neccessary but not sufficient to describe the scientific method.

0

To get a more positive answer based on current facts, and less criticism of the premise, we can look at what are the methodologies in the 'sciences' that are not 'Popperian'?

Two examples stand out for me -- anthropology and the dialectical systems like those of Marx and Freud. I would suggest that these two offer two alternative criteria for ranking hypotheses, and, if nothing else give us guidance as to which hypotheses to test, which is a major component of scientific inquiry that Popperians just ignore.

The component of anthropology that defies falsificationism is the tradition of handling data as a story, which comes from the original habit of handling stories as data. There is an internal human criterion for what is and what is not a good story, which I would attribute to an instinct against being deceived. Everyone in science uses this criterion, but only anthropologists and those favorably disposed toward them from other social sciences dare to consider this a basic part of the scientific process.

The component of dialectics that defies falsificationism is the continual iteration of refinement that makes the theory infinitely flexible. At its worst Freudianism is never wrong, the error is always in the application, and the failures to apply are always analyzed in terms of how to better apply the theory in the future. Only when there is a real and deep disagreement in the community of practitioners does anyone fall back on actually testing the theory, usually by borrowing ideas from literature or from the other, more data-driven, schools of psychology. At its best, only modifications of the theory are allowed that recapture the current theory in its entirety and simply shift focus.

I would propose that these two stances are captured in Kuhn's broader theory of science. They are things that he sees going on in scientific 'revolutions'. When the current paradigm loses traction, it is necessary to develop alternative groundwork, and to winnow these contenders down or merge them into a cohesive candidate for a new paradigm.

The process that seems to go on, to my eye, for developing new groundwork is very much a dialectical process, weak candidates are refined and folded in an iterative process that resembles dialectical development, until they have a certain internal texture that allows them to gain adherents.

The process for winnowing alternatives, again, to my eye, strongly depends upon our sense of story, and runs very much like the process of historiography or anthropology.

I would argue that there is space for a process that more fluidly combines these three approaches and takes them all equally seriously. If Kuhn is right, from a historical perspective, science already really does so. But it ignores its implicit dependency on the storytelling process and disparages dialectical refinement, admitting their relevance only when it is in crisis.

To me, such a process would be Kuhnian evolution writ small, and would resemble metaphorically something like sculpture. One chooses a medium, one makes or cuts the bulk to be sculpted, and then one refines that bulk to produce a recognizable work. Traditional sculpture resembles what Kuhn describes: almost all of the work is refinement. But other varieties exist where the bulk is molded or welded, and most of the work is not in cutting away excess, but in the middle phase of building up. And still others exist, from homey decoupage-in-relief, to outre 'found' art, where the primary activity is gathering the medium.

2
  • 1
    It would be helpful to add links or explanations about Kuhn's ideas and the methods from anthropology and dialetics. The way it is is very vague to me.
    – ivbc
    Jun 11, 2017 at 2:13
  • Freud, Marx, Kuhn and Popper are already huge names that anyone can Google. I am really sure that adding the links would help absolutely no one.
    – user9166
    Jun 12, 2017 at 16:25
0

I can think of two alternate processes that are used to attain truth where science is not suitable to the application.

The process that this thread follows is the Socratic method - arguments and counter arguments that apply logic, some degree of evidence and examples to prove a point. Law uses this method, forensic science can contribute to evidence but ultimately the Socratic method is the process used to determine guilt or innocence. You can prove someone’s blood is on the scene with science but you cannot prove why it is there or prove that the killer had intent, the Socratic method is the best process to attain truth in this scenario.

In the design/planning field research by design is a common approach to determine the best possible outcome . Essentially multiple scenarios are designed and each is assessed based on an establish framework. This could be used to determine the best approach for densification of a town or for cleaning a river. Science is limited in that it generally focuses on one variable and complex systems like cities and environments are chaotic.

Science has

0

Science is a human endeavour, which, like all human endeavours is flawed. By definition then, it can be improved. Kuhn wrote about paradigm shift in science. Paradigm shift is necessary because of inertia within a scientific field. If there were less inertia, paradigm shift would not take so long and progress would be smoother. If the scientific method is sufficiently improved in the future will it be a qualitatively different method? The recent development of a vaccine in record time shows that increased resources, accelerated regulatory steps, and accelerated clinical rollout can reduce inertia. So I think that a better way is likely but it will be founded in the scientific method.

-1

One method is solipsism, or radical skepticism. You can only believe what you personally experience - in this moment. What is it you want of a method? The scientific method is a poor method if one is attempting to solve the problem of other minds. The more you measure the world, at finer and finer scales, the less inclined you may be to regard anything - even another person as possessing "a mind of its own."

http://plato.stanford.edu/entries/other-minds/

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .