t f p g+ YouTube icon

Problems with Defining Science Using the Falsification Criterion

Bookmark and Share

November 27, 2011 Tags: Science & Worldviews
Problems with Defining Science Using the Falsification Criterion

Today's entry was written by Steven Benner. Please note the views expressed here are those of the author, not necessarily of BioLogos. You can read more about what we believe here.

This blog is the second entry in a series by Steven Benner. The first post can be found here. Throughout the series, Benner discusses the nature of scientific progress and the difficulty of defining what is and is not science. Discussion questions are included at the bottom of each post.

As we began to see in my last post, falsifiability is not a particularly useful tool for distinguishing scientific and nonscientific propositions. Take a simple law-like proposition that philosophers of science like to discuss: "All emeralds are green". We may regard this proposition as scientific because we can conceive of an observation that falsifies it. We might observe an emerald that is not green. Hence, we might conclude that the proposition is "scientific" under the falsifiability demarcation criterion. Not a particularly interesting law, of course. But perhaps we can be satisfied that we are doing "real science".

We can even express this in a syllogism that classical Greek philosophers would recognize (it is called the contrapositive). If an observed X is an emerald and X is not green, then the proposition is false.

Unfortunately, things are not so simple in the real world of science. It turns out that whether or not an emerald is observed to be green depends on how it is observed, and who is doing the observing. For example, an emerald may be observed to fluoresce a red color when observed under ultraviolet light.

No problem, you say.

The proposition can be changed to read: "All emeralds are green when examined under white light". But even then, the falsification effort does not work if the observer has red-green colorblindness. We must further modify the proposition to read: "All emeralds are green when examined under white light by an observer who does not have red-green colorblindness".

These modifications of the original law constitute ad hoc "auxiliary propositions". We invent them to explain away an observation that would otherwise appear to be falsifying. This type of thing seems to defeat the demarcation.

We might say: Fine, scientists are not allowed to modify, ad hoc, propositions so that they survive observations that apparently falsify them. Unfortunately, it is not justifiable. We should protect propositions from certain contradicting observations. For example, we should not discard a theory if the observing instrument was broken at the time that it generated an allegedly contradicting observation.

The creation of auxiliary propositions ad hoc must be done if one does not want to be paralyzed in building models for reality by the exigencies of real world experimentations and observations. Not surprisingly, the history that created the empowerments cited above is littered with ad hoc auxiliary propositions.

These problems become only worse when we move to more complex (and more interesting) scientific propositions, where many propositions are logically connected, including many that we do not even know that we are presuming. For example, it is widely accepted today that the Earth was formed about 4.5 billion years ago and that life has been present on Earth for the majority of the time since it formed.

No problem today, perhaps, but it was a problem in the 19th century as evolutionary theory was being developed. Biologists required hundreds of millions of years of Earth history to explain the observed diversity of life under a model of gradual evolution. Darwin himself used the record of sedimentary rocks to suggest that the Earth was at least 300 million years old.

Some famous physicists disagreed, including the physicist William Thompson, also known as Lord Kelvin. Kelvin was the physicist who helped develop the laws of thermodynamics, one of the most robust sets of laws that physics has ever produced. Kelvin's contributions to physics were so significant that the absolute temperature scale is named for him. We measure temperatures from absolute zero using "the Kelvin", not "the degree Fahrenheit" or even "the degree Celsius".

Starting in 1862 and for forty years thereafter, Kelvin used his understanding of thermodynamics to argue that the Earth could not possibly be as old as evolutionists required. Why? Because the Sun could not possibly be so old. Even if the Sun were made of the best coal possible, said Kelvin, it could produce heat at its current rate for only about a thousand years. Kelvin thus held that the laws of physics disproved the model of common descent by gradual evolution, key to Darwinian evolution as a theory.

Today, we know that the Sun generates its energy by nuclear fusion and radioactive decay, not by burning coal. This involves the conversion of matter to energy under Einstein's famous e = mc2 equation. Kelvin knew nothing of either. Nor, however, did the evolutionists who stubbornly continued to believe in evolution, despite its having been "falsified" by physics. Illustrating the challenge in defining science as a string of falsifiable propositions, Kelvin's falsification had to be ignored to get the correct answer about the age of the Earth.

The exhortations to "think outside the box" or to "challenge authority" do not serve us here. Had we put Darwin and Kelvin on a stage to have a grand debate, they would never have arrived at e = mc2. Indeed, had someone suggested in 1880 that the Earth was very old and the physicists were incorrect because atoms could fuse to give new atoms with a net conversion of matter into energy, they would have been dismissed as heretics by both camps.

I review these examples of abstract and real science to disrupt the comfortable lessons that you may have been taught about "the scientific method". I also want to emphasize the importance of a more sophisticated understanding of processes within science before people make what they think are "scientific" arguments as they argue in the public square. Once this comfortable complacency is disrupted, we can rebuild a more realistic view of how scientists actually generate empowering knowledge about the real world.

In my next post, we will consider cultural factors that influence how scientists do their work.

Discussion Questions: Dr. Benner suggests that falsifiability is not a comprehensive criterion for determining whether a question is or is not scientific. What other criteria might be important in defining science? Benner also argues that the current state of knowledge often sets the stage for a new discovery, which could not have been made before. Can you think of other examples from the history of science where new ideas became "ripe" for discovery? Do we hold too tightly to our current state of knowledge as being absolutely true?

Steven Benner is a Distinguished Fellow of the Foundation for Applied Molecular Evolution in Gainesville, FL. He received his doctorate in chemistry from Harvard University. Benner and his group of researchers initiated synthetic biology as a field and invented dynamic combinatorial chemistry, which is currently being used in pharmaceutical development.

< Previous post in series Next post in series >

View the archived discussion of this post

This article is now closed for new comments. The archived comments are shown below.

Page 1 of 1   1
gingoro - #66271

November 27th 2011

I don’t think that your example illustrates a problem with the idea that in general scientific facts and theories need to falsifiable.  I consider that in the initial stages of an undeveloped science that all that one may be able to do is collect observations and possibly organize them into taxonomies.  Yet such stamp collecting is part of science an important part.

Darwin could have found something that would directly have falsified his theory and that would have been much more devastating.  In modern day discussions the proverbial rabbit in the Precambrian would be such a devastating falsification.  (By the way I have heard holders of the YEC position argue that Noah only took a few species with him on the boat and that the diversity we see now is the result of evolution that has occurred in the intervening 6000 years.  Duh.)

To my mind the real issue is what one does when something falsifies the current best theory.  Very often scientists continue to use the existing theory until something better comes along.  For example relativity and quantum mechanics are in conflict yet we continue to use both theories
as they are the best we have.  In fact Newton’s laws seem quite adequate
for landings on the moon etc.  Part of the greatness of leading
scientists is to “know” when to “ignore” a falsification and when such a
falsification must be immediately dealt with.

Of course falsification is not sufficient as a demarcation for science.  After all when debugging complex computer codes one makes lots of hypotheses’ and then arranges to test those hypotheses until the source of the bug is located.  No one would call such an activity science although falsification occurs with great regularity.
Dave W

James R - #66272

November 27th 2011

I agree with gingoro.

Falsification, by itself, may be an inadequate demarcation criterion, but that scientific hypotheses must be potentially falsifiable cannot be in doubt.  It is the job of natural science to account for nature; if a hypothesis cannot be squared with the observation of nature, the hypothesis must be dumped.

Of course, all due care must be taken in applying this general principle.  It is possible that a measurement that seems to refute a hypothesis was made sloppily and is inaccurate.  And if the overwhelming number of measurements seems to support the hypothesis, and only a few seem to contradict it, it is possible that some explanation may be forthcoming which will vindicate the hypothesis—a suitable rider or qualification that will explain the apparently incompatible data in terms of another law which can be shown to be compatible with the hypothesis.  No one is suggesting that a major scientific theory such as evolution should be thrown out because the sequence data from just one genome does not fit the expectations. 

Nonetheless, there comes a point where the number of adjustments that need to be made to the hypothesis in order to “save” it from observational data becomes so great, that the hypothesis can only be rescued by grafting onto it a patchwork of conceptions that have nothing to do with the original idea, and the hypothesis ends up surviving in a monstrous, distorted form.  At that point, scientists should be looking for a simpler, more coherent hypothesis.

The problem with fantasies like “string theory” and “multiverse” is that the arguments in their favor are purely theoretical; they are accepted, where they are accepted (and there is much resistance to both of the examples), because a certain group of scientists finds them mathematically elegant or satisfying or aesthetically pleasing or something of the sort.  But many theories might be internally coherent and to some minds pleasing; what makes science science, and not philosophy or literary criticism or something else, is that theories are tested against nature.  We should all distrust theories that are so purely mathematical and conceptual that they have no touchpoint with nature.  One way of determining whether or not a theory has a touchpoint with nature is to ask whether it can generate falsifiable hypotheses. 

It is often said, for example, by evolutionary theorists, that a fossil of a Cambrian rabbit would destroy evolutionary theory.  And so it would—provided the finding was properly verified.  There is no conceivable adjustment of traditional evolutionary biology which could absorb Cambrian rabbits as the product of some law quite compatible with gradual diversification of a few simple original forms. 

The same applies to those creationists who insist that six-day creationism is a legitimate scientific alternative to evolution.  If they want it to be counted as scientific, it should be at least potentially falsifiable by the appeal to evidence.  But there is no imaginable evidence which a six-day creationist would ever allow to falsify the position; always some contrived adjustment would be made in order to preserve the position, because the position is not ultimately held for its quality as science, but because it is believed to be guaranteed true by God.  I’m not opposing creationism as such here, but merely denying its claim to be scientific.  It’s precisely the criterion of falsifiability which enables us to see why that sort of literalist creationism is not and cannot be science.  No risk of falsification, no science. 

Thus, while I don’t worship falsification as the be-all and end-all, I am deeply suspicious of those who preach against falsification.  I always suspect them of preaching against it because they wish to offer theories which have very little empirical testability, and pass them off as good science.  I can’t think of any version of modern natural science in which falsifiability wouldn’t play a significant role, even if by itself it is not a satisfactory demarcation criterion.

HornSpiel - #66283

November 27th 2011

The problem with fantasies like “string theory” and “multiverse” is that the arguments in their favor are purely theoretical;

I would not call theories based on mathematical models fantasies. These models do in fact do make predictions that can be theoretically tested in the laboratory, even if we do not yet have the aparatus to do the testing. Einstein’s equations of relativity were such in their time. Certainly these new theories will ultimately stand or fall by their practical/empirical validity not philosophical (or theological for that matter) elegance. That is what science does.

If the multiverse turned out to be true, would that threaten your faith? It shouldn’t, no more that the Big Bang. The question would then become, Where did the Multiverse come from? The search for naturalistic explanations to observed facts should never be opposed by Christians. If and when I do feel uneasy by such efforts, I find it means I was relying on those facts for my belief rather than actual time praying and experiencing God.

James R - #66296

November 29th 2011


I was not calling string theory and multiverse fantasies because of their
mathematical character.  I was calling them fantasies because they are
wholly imaginary, until they are anchored by empirical testing.  That is,
they are systems of the mind’s eye only at this point.

Nothing that “turned out to be true” would threaten my faith, since whatever
is true must have come from God.  However, there is strong reason to
believe, based on the sort of person who generally champions the multiverse,
that the multiverse view is popular because it eliminates the near-inevitability of inferring
design from the fine-tuning of the universe—the only universe which we
know for certain to exist.  I am not saying that there are no honest
theoretical grounds for postulating a multiverse, but as all careful
students of the history and sociology of science know, theoretical arguments
are not the only motivation that scientists have for supporting the models
that they do.  And what applies to scientists applies with even greater force to science popularizers and science groupies of the sort that swarm the internet, who are frequently motivated more by secular humanist ideology than by the love of

I don’t say that Christians should oppose inquiries into possible natural
causes of anything.  However, there is no reason for a Christian to assume
that a wholly natural account will always be available.  It’s fine to 
propose and investigate possible naturalistic accounts of the origin of life; it’s wrong to
assume that life must have originated by wholly natural means.

I do not think that “praying and experiencing God” are any more reliable
ways of being sure of God’s existence than arguments based on natural
phenomena.  Psalm 19, Romans 1 and other passages in the Bible stress that
knowledge of God can come from nature as well as from “religious experience.”  And epistemologically speaking, inward certainties that one has experienced God (e.g., “I feel that God is guiding me in my choice to quit my job as an accountant and go to medical school”) are deeply problematic:  wishes, hopes and fears (along with drugs, chemical imbalances in the brain, psychological conditioning, etc.) can delude a person into “experiencing” things that have no objective referent.  And of course, there is the problem of comparative religion:  many Hindus are as certain that Krishna has spoken to them as Christians are that Jesus has spoken to them.  In short, “subjective” proofs of the existence of God are no more reliable than “objective” proofs (e.g., teleological arguments, cosmological arguments).  I personally find teleological arguments much more persuasive than the word of my neighbor that God talked to him last night.  I have no idea of the psycho-medical history of my neighbor, whereas I can examine for myself the data from which people infer design, and the inference seems pretty persuasive.

tufty22 - #66291

November 28th 2011

i would also not agree that theories based heavily in mathematics are fantasies.  we need these aesthetic formulas to lead us to greater progression in the lab.  a failure is one more possibility removed and one step toward a sound discovery.  i also think you may be omitting the most important aspect of theoretical exploration and it’s touchpoints to nature.

we are nature at its finest.  why shouldn’t we argue truth from aesthetics or emotional connection within science?  it seems to me that without these humanistic anomalies there would be no flame under the ass of scientists.  no wonder when looking at the stars.  no love.

tufty22 - #66292

November 28th 2011

really, i can’t say (colloqial word for buttocks) in here?  hmm i guess all the kids reading this would be offended.

test sentence:  i rode a stubborn ass through the desert.
Jon Garvey - #66275

November 27th 2011

The trouble with the Cambrian rabbit example, it seems to me, is that any theory that required that degree of anomaly to be falsified would be problematic in itself.

It’s akin to saying that Einstein’s theory would be falsified by anyone who could teleport a rock from Mars to the earth faster than the speed of light. It would indeed do so, but setting the bar that high would suggest the theory was running scared.

Darwin’s own falsification - “if it could be shown beyond doubt that a change could not occur by infinitesimal small steps…” (or words to that effect) sounds more modest but is equally impossible. You could even propose a telephone turning into an orang-utan that way.

gingoro - #66277

November 27th 2011

Jon I was trying to differentiate between a falsification that is intrinsic to evolutionary theory and an extrinsic falsification.  Where an extrinisc falsification is of the kind where studies in another science (physics in this case) provide conflicts such as the world not having sufficient age for evolution to occur.  I would see the problems that Newtonian mechanics faced being mainly of the intrinsic kind in that not all observations in physics could be explained by Newton’s theory.  I’m not sure that I am explaining my thought here very well but I differentiate between conflicts within a particular field in science and conflicts with other areas of science.  I’d tend to rate those within a given field as more important than those across fields.
Dave W

Jon Garvey - #66279

November 27th 2011

Hmmm… you lost me a bit there. Are you saying that you can fool some of the people all of the time internally, but only all of the people some of the time externally…? Probably time I got to bed.

Seriously, I’m not exactly sure if this is where the post is coming from, but historical theories are intrinsically difficult to falsify, because you can’t reproduce the events, and data is necessarily incomplete. You can only take what data there is and make the best inference.

By the same token historical theories ought, then, to be treated with rather more caution than operational theories - calling them “established fact”, for example, is as irresponsible as claiming to have discovered the definitive version of history.

So the Cambrian rabbit would not falsify evolution, as you could claim that it’s evidence that the Cambrian explosion had even bigger megatonnage than we thought, and that mammalian evolution occurred much earlier than had been concluded from the deficiency of the fossil record before the Eocene, which deficiency will no doubt be filled by future discoveries. Rabbits clearly evolved, as you don’t find them in the Precambrian rocks.

In such a scenario, what other inference could there be if evolution is the only naturalistic game in town?

James R - #66297

November 29th 2011


I’m a big fan of yours, so you must forgive me if the only thing I can find to disagree with you about is something small.  (And after I voice that disagreement, I will proceed to a rant of my own which is not intended as a refutation of anything you have said above.)

Yes, it’s possible that Darwinian theory could be rescued from a rabbit fossil that was “too early” by postulating that the ancestors of rabbits—presumably reptiles—were around much earlier than previously thought, but that we have unfortunately lost all fossil remains of those earlier reptiles.  But this is clearly special pleading, as the usual excuse for why we don’t have more pre-Cambrian fossils—that soft tissues don’t preserve well—would not apply to reptiles, which have bones.  There is no reason to think that the emergence of the first land vertebrates (the precondition of rabbits) is going to be pushed back much further than it already has been.  Maybe 5 or 10 million years, but not 50 or 100 million years.  So if you find a rabbit fossil before there are any reptile fossils, or, worse, before there are any amphibian fossils, any gradualistic theory of evolution is dead. 

One could still hold out for a hopeful monster—a rabbit that was the son of a sea squirt of something of that sort.  But no biologist would take this seriously as a possibility; the sea-squirt has no delivery system to preserve and develop a rabbit embryo; and in any case, the rabbit would drown as soon as it was born a mile underwater, ending its evolutionary line.

Of course, you know all of this.  You were showing how a Darwinian might argue in order to preserve his hypothesis against seriously damaging evidence.  And you were perfectly right; this is how he would argue, and how he would be compelled to argue, give his premises.  But the argument is pathetically weak.  A Cambrian rabbit would be a colossal problem that would leave Darwinists everywhere red-faced and scrambling.

The problem with the Cambrian rabbit example lies not in the logic.  It lies in the fact (overlooked by both pro- and anti-evolutionists) that it says nothing about mechanisms.  For example, we might never find a Cambrian rabbit, and therefore we might conclude that evolution is the best explanation for the morphologically logical sequence of fossils.  That would tell us nothing about how the sequence was generated.  Darwinian theory could still be almost entirely a false description. 

Thus, when the scientist folds his arms, glares, and says:  “Find me a Cambrian rabbit!  You can’t, can you?  Therefore, Darwin was right!” the scientist is entirely missing the point.  Darwin did not claim to have invented the idea of common descent.  His contribution was not “evolution” but natural selection acting on small variations as the mechanism of evolution.  The non-finding of the Cambrian rabbit merely makes evolution likely and renders possible the Darwinian mechanism; it doesn’t go one inch toward verifying that mechanism.

The problem with much evolutionary apologetic from atheist and TE Darwinists, on this site and elsewhere, is that it focuses on trying to establish the supposed fact of evolution, based on allegedly transitional fossils and genomic similarities and so on; but even a rock-solid demonstration of the fact of evolution would not verify that it occurred due to the means proposed by Darwin and his neo-Darwinian successors.  The Cambrian rabbit could falsify both the Darwinian mechanism and evolution generally; but the non-occurrence of the Cambrian rabbit cannot verify the Darwinian mechanism.  That’s why the “Find me a Cambrian rabbit!” defense is irrelevant.

Every serious investigator agrees that we are not likely to find a Cambrian rabbit.  Doubt about Darwin (outside of the fundamentalist world) does not focus on the hopes of finding a fossil out of sequence.  The problem with Darwinian explanation is that even if all the fossils are perfectly in sequence, it is still an unlikely candidate for the cause.  You don’t hear Shapiro, Newman, Margulis, etc. talking about Cambrian rabbits.  They don’t need the Cambrian rabbit to show the deep inherent weaknesses of Darwinian theory.
But again, I am sure you know all this, as I am sure that gingoro knows all this.  Perhaps, though, my exposition will be of use to others.

penman - #66317

November 30th 2011

At last someone’s mentioned Shapiro!

Jon Garvey has a helpful summary of his recent book “Evolution: A View from the 21st Century” on Jon’s “Hump of the Camel” website. And James R. is perfectly correct that one doesn’t have to find a rabbit in the Cambrian to entertain plausible doubts about the “omnipotence” of natural selection as evolutionary mechanism. Shapiro manages it without a rabbit in sight (Cambrian or otherwise).

Personally I have no serious doubts about the genealogy of life over geological time, but mechanism has always bothered me, & it’s nice to see a Shapiro stirring the pot of ideas…..

Jon Garvey - #66276

November 27th 2011

But to have a go at the study question: ‘Can you think of other examples from the history of science where new ideas became “ripe” for discovery?’

I’m fascinated how the ability to conceptualise in science seems so dependant on the current state of technology, or even sociology. For example, in the age of mechanics, the universe was mechanical. In the age of European Progress, evolution was seen as the move towards perfection. In the age of chemistry, life was explained by chemistry. In the age of computers, DNA is seen as a code. Now, of course, we begin to see cells through the eyes of nano-technology.

Does that mean that new technology opens our eyes to previously hidden truths, or does it actually restrict is from looking at things in any other way?

Roger A. Sawtelle - #66298

November 29th 2011


The problem today is that philosophy has not kept pace with science.

Philosophy still favors mechanistic world views.

James R - #66325

November 30th 2011


“Philosophy still favors mechanistic world views.”

I cannot think of very many philosophy professors today who still hold to a mechanistic world view.  In fact, I can’t think of any examples at all, except (maybe) Dan Dennett.  Do you have evidence for this claim?  Or are we into a vocabulary problem again, with you using “mechanistic” in a non-standard way?

Page 1 of 1   1