by Jonathan Y. H. Sim
Science comes from the Latin word, “scientia,” which means: knowledge. As a noun, it refers to the body of knowledge that we, humans, have discovered. As a verb, it refers to the process of knowledge-creation: from observing, hypothesis-making, experimenting, and even predicting.
Humanity has come a very long way with Science. We know more about the universe and about ourselves than ever before. We have extended our lifespans, decreased the mortality rates, and found ways and means to tackle diseases and famines.
Yet, we humans are funny creatures. Despite all the advances in the Sciences, we have reached a curious point in our history where more and more people are increasingly sceptical of Science. Around the world, many parents are refusing to vaccinate their children out of the fear that these vaccines would cause autism in them. More people are rejecting the fact that the Earth is round, on the grounds that there is no evidence that it is round, but that all evidence points to it being flat.
Part of the scepticism is rooted in the fact that the very enterprise of science, of knowledge-making, has become so sophisticated, that it has become incredibly difficult for the average person to understand it well. Most people get the wrong idea that Science is about definite answers, proven facts that are not open to disputes. And they get disillusioned when the news (or articles on the Internet) constantly feeds them with contradictory headlines about Science. Some days, you’d see headlines like, “Too little sleep causes cancer.” Another day, you’ll see a contradictory headline like, “Too much sleep causes cancer.” And a few days later, you might see yet another contradictory headline like, “Sleeping 8 hours a night causes cancer.”
If Science is about things that have been proven, why are we getting articles like this? It’s easy to become sceptical when we are constantly bombarded by such headlines.
But another reason for that scepticism comes from pseudo-science (fake science). Unlike conventional academic Science, pseudo-science is incredibly easy to understand. And that is greatly aided by the fact that people often claim to observe immediate results. This is not something we see in the laboratories or hospitals. To add oil to the fire, sceptics sometimes argue that money-loving corporations are pulling the strings in conventional Science, preventing us laypeople from having access to the best that science has to offer.
Who’s right? Who’s wrong?
I will not use this chapter to tell you what is good science or pseudo-science. Even that can be very problematic. “Science” is a problematic term, just like the word, “questioning.” We use it, but we don’t know what it really means. Even scientists across the entire spectrum of scientific disciplines cannot fully agree either. So, if we don’t have a good definition of science, how are we able to tell good science apart from pseudo-science?
In this chapter, I will expose you to the questioning processes of the Physical Sciences, so that you can develop a better grasp of the rigorous process underlying the enterprise of knowledge-making.
1. What is Science?
Richard Feynman defined Science as a process of investigating and explaining observed phenomenon. Another scientist by the name of Hobson, says that it’s a process for proposing, testing and refining ideas. These definitions, unfortunately, are too broad and general to be useful in identifying good science.
In the early 1900s, physicists tried to define what makes science science, and what makes it good science. At that time, physics was regarded as the top science. However, in that process of defining what science is, their definition of science was too physics-centred. This was a huge problem for all the other scientific disciplines, as they all investigate nature in very different ways.
Biology took the greatest hit from this physics-centred definition. Unlike physics and chemistry, much of biology did not have universal laws and theories, and consequently, biology was regarded more as an inferior science, or worse, pseudo-science. The social sciences took a big hit as well, and were immediately regarded as pseudo-science simply because it wasn’t anything like physics at all. Till this day, there are many prejudices against the social sciences as a pseudo-science. As a consequence of such a definition, many scientific disciplines starting having “physics-envy,” and tried to conform their processes to that of physics. But this creates more problems for those disciplines as it hinders their ability to effectively study phenomenon as well as they should.
Over the decades, scientists have since becomes more open and understanding of the diverse nature of the Physical Sciences, expanding the definition of Science away from that initial physics-centred definition, to one that is more encompassing of its diversity. A typical modern criteria for what makes for a good scientific theory looks like this:
(1) The theory must be able to described observed/experimental data with error considerations
(2) The theory must be open to falsification
(3) The theory must have predictive power
(4) The theory cannot claim 100% certainty, and is only fallible.
This seems like a good working criteria that includes all the scientific disciplines. However, good scientists sometimes do not operate this way. And more interestingly for us, a good deal of pseudo-science have since tried to appear legitimate by conforming to such a criteria of good science. This goes beyond the standard definition of pseudo-science, where you start with the conclusion and work your way to prove it by finding evidence that proves your conclusion (while ignoring evidence to the contrary).
A good example are the UFOlogists. These are well-meaning people who have dedicated their lives to searching for signs of aliens on Earth. They have been mocked and laughed at by the scientific community for decades, and have since stepped up their game to meet the rigours of science, developing theories that explain observed data with error considerations, and are open to falsification. Yet, no scientist will take their work seriously because mere conformance to a criteria for good science is not enough.
What is important to good science is the process. It is not about the results, it is not about the answers.
2. The Good, The Bad, and the Ugly of Science
To help us navigate through the ambiguity in the word, “Science,” as a good philosopher, I would like to distinguish a few categories of “Science” to help us better appreciate and understand the multiple senses that are at play when we use that word.
(1) There is good science, which follows rigorous processes that I will elaborate later.
(2) And there’s bad science, where the process is sloppy, and often prone to inductive errors.
(3) There is pseudo-science, which pretends to be scientific even though it doesn’t follow the rigours of science.
(4) There are things out there that’s just non-science. It has nothing to do with science and doesn’t make scientific claims either. E.g. the statement, “french fries are delicious,” is a non-scientific claim. However, it becomes bad or pseudo-science if I start to claim that “french fries are good for health.”
(5) Finally, there is also misreported (or misunderstood) science, which has the potential to transform good science into bad science or even pseudo-science. The source of this problem stems largely from news sources. More often than not, journalists and editors do not have the scientific training to properly understand and interpret journal papers. But even if the journalist has a background in science, his/her training in that particular field in the sciences is not sufficient to enable that journalist to understand things beyond that field.
To be fair, sometimes, the exaggeration comes not from the journalists but from the scientists themselves. Sociologists of science have found researchers behaving like this because it works so well in helping them receive the much-needed funding. This, however, is a rather myopic move that contributes to science scepticism around the globe.
Let me cite two examples.
In December 2017, Kristensen et al published a paper entitled, “Ibuprofen alters human testicular physiology to produce a state of compensated hypogonadism.” Essentially, the paper found that the use of Ibuprofen (a pain killer) affects the hormonal levels in men, leading to a reduction of testosterone.
However, on 9 January 2018, CNN reported this article instead as: “Ibuprofen linked to male infertility, study says.” This is a far cry from what the paper originally discussed. But wait, it gets better! The study got distorted even further. Other news websites, like Metro UK, reports: “Popping ibuprofen could make your balls shrivel up.” Nowhere in the original study did it ever mention infertility or even the shrinking of a man’s testicles, yet somehow, credible news sources, like CNN, can report something so incredibly distorted.
Here’s another hilarious example. In February 2018, Kageyama et al published a paper entitled, “Spontaneous hair follicle germ (HFG) formation in vitro, enabling the large-scale production of HFGs for regenerative medicine.” All the paper says is that they found a chemical compound that promotes hair growth on mice, which could potentially be used in humans. One journalist did some research and discovered that the same chemical compound is found in McDonalds french fries. Thus, days after the paper was published, credible news websites, like Newsweek, started reporting such headlines: “Chemical in McDonald’s Fries Could Cure Baldness, Study Says.” This article went viral, and poor Professor Kageyama received calls and e-mails from all over the world from bald men asking just how much french fries they needed to eat to cure their baldness.
How could news like this lead to pseudo-science? Well, I could set up a hair treatment clinic offering a special french fry therapy. If you dared to challenge my treatment as some kind of bogus nonsense, all I have to do is to point you to the paper by Kageyama et al, and some of these credible news websites as my proof. And if you are not a specialist in bioengineering or biomaterials, you wouldn’t have the knowledge or resources to challenge it. Chances are, you won’t even read the scientific paper, but simply acknowledge it on the grounds that it has passed the stringent process of peer review, and is thus a fact. And thus, I can spread my pseudo-scientific treatment all over the world, offering french fries as a cure for baldness, with Science appearing to back me up.
This is, in fact, what is happening with a lot of pseudo-science these days. If you challenge a pseudo-scientific treatment, the founder could potentially point you to a few scientific publications as proof.
In which case, not only is misreported/misunderstood science really damaging, but it is what’s giving weight to a new breed of pseudo-science where the general public have difficulty differentiating from Science itself!
3. Negative Questioning: Learning Good Science from the Mistakes of Bad Science
One way to learn about what makes good science good is to study the ways in which we fall into error in committing bad science. Here, I’d like to cover four of the most common mistakes: (1) hasty generalisation; (2) unrepresentative sampling; (3) survival bias; (4) and confusing correlation for causation.
3.1 Hasty Generalisation
Have you ever found yourself saying something like this?
I’ve encountered a taxi driver who was really rude today. I can’t believe how taxi drivers are all rude people! Where are their manners?
Well, unless you have encountered every single taxi driver that’s out there, you shouldn’t believe it either! This is an example of an inductive fallacy, known as Hasty Generalisation. Here we are assuming – based on observations of a very small sample – a general principle that applies to the bigger group.
The problem with a small sample is that we can never be sure about it representing the rest of the population. Even if you encountered 30 rude taxi drivers, they might only represent 1% (or less) of the taxi driver population. It is entirely possible that you were just unlucky to have encountered all the rude drivers that’s out there (maybe they live in your neighbourhood). It is also possible that confirmation bias (remember this?) is at work, so we only pay attention to taxi drivers who are rude, and forget about the ones who are polite.
You cannot properly draw a valid conclusion from a small observational sample. What makes good science good is the fact that science pays careful attention to the sample size in an observation or an experiment. If a scientist did a limited study on 20 subjects, the scientist is careful to be clear that what is true in the findings is only limited to the 20 that he/she observed, and is only willing to extend that claim if and only if he/she has had a chance to sample a much larger population.
3.2 Unrepresentative Sampling
Let’s imagine that you have ordered ten crates of grapes from a supplier to make wine. You had asked for the freshest grapes that are ripe and ready for wine making. But this is the first time you’re ordering from this supplier. How do you know that the supplier has given you what you had asked for? How do you know that the supplier didn’t cheat you by hiding rotten grapes?
One way would be to check every single grape in all ten crates. But that’s almost impossible. It’s too time consuming.
You could check all the grapes in the first and second crates. But it’s possible that the supplier might have guessed that you’d check it this way, and hid rotten grapes in the last few crates. In which case, your sampling would never encounter the rotten grapes at all.
You could check the top layer of grapes in all ten crates, but it’s also possible that the supplier had hidden all the rotten grapes at the bottom of each crate. Again, your sampling would never encounter the rotten grapes at all. And you’d go away with the mistaken impression that all your crates of grapes are fresh and fine.
This mistake is known as unrepresentative sampling, and it is a methodological error where the subjects sampled for the observation or experiment are not representative of the general population that’s being considered. One solution would be to randomly sample a few bunches from the top, middle, and bottom layers of every crate. By being thorough in your sampling, you can be sure that what you have been observing is not a biased or skewed set.
Here’s another example to illustrate the point. If a scientist wishes to study the effects of alcohol consumption on the kidneys, the scientist must be careful of the presence and proportion of alcoholics in the study, as they consume far more alcohol than the general population (unless of course, this is a study on the effects of alcoholism, then that would be a different matter). So if we have a sampling that has a high proportion of alcoholics, our findings would be skewed because of them. The results, therefore, would be true of alcoholics, and not of the general population.
3.3 Survival Bias
Imagine this scenario: It is World War I, and you are the chief aviation engineer in the air force. A squadron of planes have just returned from a bombing mission. These are the survivors. The rest have been shot down. To reduce the number of casualties, the general has commanded you to examine the damage on the planes that have returned, so that you can recommend to improve the armour. This way more planes can survive enemy fire and return in the next wave of attacks.
You look at the planes and the only clue you have are the bullet holes on the planes. Based on your examination of the bullet holes, will you recommend to add more armour to the area with:
(a) 10 bullet holes?
(b) 5 bullet holes?
(c) 1 bullet hole?
(d) or 0 bullet holes?
Most of you might probably say that you’d upgrade the armour in the area with 10 bullet holes. If your answer is (a), (b), or (c), you have committed Survival Bias!
In fact, the answer is quite the opposite. The area with 10 bullet holes (or more) means that that particular area is not very essential to the plane. There’s no need to add more armour because bullets going through that area are unlike to cause the plane to crash.
Areas with 1 or 5 bullet holes might be important. But they are not as important as the area with ZERO bullet holes. Those planes survived because they were not shot at that area. The planes that crashed from the sky had probably been hit just once at that area and did not survive.
What happened here is that we have forgotten about the non-survivors. An examination based purely on the survivors is an unrepresentative sampling.
Survival bias is another cognitive bias that we easily fall prey to just like confirmation bias. Survival bias arises when we overlook the fact that a sample we are examining had previously made it past a certain selection process. It’s called survival bias because we are merely looking at the survivors, and not those that failed to “survive” the selection process prior to the study. This bias is more devious than the error of unrepresentative sampling because we can be blissfully unaware of the lack of representation due to the absence of other subjects that might comprise our population of study. And so we end up going away with the false belief that we have done a study that is quite representative of the general population. As a consequence, we may end up with overly optimistic beliefs, having ignored (or forgotten) the failures that didn’t “survive.”
3.4 Confusing Correlation for Causation
There is a story of a farmer who observed that whenever the rooster crows, the sun would rise. He intrigued by this relation. So he spent one month observing the rooster before sunrise. And without fail, whenever the rooster crows, he would see the sun rise. After observing this phenomenon for an entire month, the farmer was quite convinced that the crow of the rooster causes the sun to rise.
But we all know (I hope you do) that the crow of a rooster does not cause the sun to rise.
What this story illustrates is an error of inductive reasoning – an inductive fallacy – of seeing two events occurring, and jumping to the conclusion that one causes the other. That A causes B.
But the causal relation could be in the reverse direction, i.e. that B causes A. The rooster’s crow didn’t cause the sun to rise. The rising sun compels the rooster to crow.
Most of the time, however, we are simply mistaken about the causal relation. There’s no causal relation. It’s just a coincidence. It can be hard to believe that two events can occur with absolutely no causal relation, by pure random chance, especially when we can see how the two are relevant or possibly related to each other.
Yet, reality tends to be stranger than fiction, and there are far too many happy and unhappy coincidences happening all the time. There is a website that I love, known as “Spurious Correlations” that demonstrates how two completely unrelated events can correlate so well. Yes, the world is strange.
So we should be careful not to confuse correlation for causation. You bought a new pair of shoes, and now you’re experiencing some aches in your legs after running, There could be a causal relation, but it could also be due to other reasons. Confirmation bias is what’s at work behind the scenes of our minds, making us think that the shoes are our prime suspect. This is precisely how a number of pseudo-scientific theories arise.
One way to overcome the urge of drawing causal links where there may not be any is to consider the list of alternative explanations: what else could have caused the phenomenon?
Here’s an example of a typical pseudo-scientific claim. Imagine you have a friend who comes up to you and tells you that she’s very certain that the magical quartz crystal she puts beside her pillow every night cures her of her regular headaches. She seems to be confusing correlation for causation, thinking that the crystal causes a cure from headaches. Are there alternative explanations? Plenty! One possible explanation for why her headaches are gone every night is that sleep can get rid of headaches. Most aches and pains (and other problems) go away the next day when we awake thanks to the wonders of sleep. Another possible reason might be the placebo effect from believing so strongly in the healing properties of the crystal. Yet another explanation is confirmation bias, where in the absence of pain killers, she’s compelled to find some explanation for why the headache is gone, and chose the crystal as it stood out of the ordinary. With confirmation bias at work, she selectively observes (or remembers) moments where the crystal is present when her headaches are cured.
Let’s try one more example of pseudo-science. There is a very popular self-help book with a strong following. In that book, it claims that if you want something, the secret is to want it intensely. The Universe will know it and make that secret wish of yours come true. Is it really true that wishing really hard for something can cause the Universe to make it come true? Many people who read that book swear by it, claiming that it works. They could be confusing correlation for causation. So let’s consider some alternative explanations. For starters, if you want something so badly, you’d naturally start working towards it – consciously or unconsciously. So it’s not that the Universe is listening to your wishes, but that we’re putting in effort to make it happen.
Now, I love this example a lot because it highlights an additional point. What happens if you tell someone, “I tried wanting something so hard, wishing so intently every day and every night. But it never came true.” If you believe in that book, what would you say? Probably: “It didn’t come true because you didn’t wish hard enough. Try harder!” There is no way to test if this theory is rubbish. Every time it fails, someone can always tell you that you didn’t try hard enough. This brings me to my next point.
4. Unfalsifiable Statements and the Importance of Falsifiability
The trouble with unfalsifiable statements is that there is no way to test if it’s ever wrong. It’s always right.
It’s like trying to prove that our lives are not guided by the hands of destiny. Everything is destiny. If you’re single and alone, it’s destiny. If you found someone in your life, it’s destiny too. If you broke-up, it’s also destiny. Everything you do is destiny, If we’re not aware of the problems of unfalsifiable statements, we might think that this theory about destiny is true, and become fervent believers of it. But destiny isn’t just true, it’s so far beyond true that there is no way to test it at all. To be clear, the reason why it’s not testable has nothing to do with us trying to deal with something in the spiritual realm (even theories about the spiritual realm can be expressed in ways that provide the potential to be tested and proven wrong).
This is not a feature, but a bug! It is a problem of flawed reasoning. With a little bit of creativity, you could generate a ton of unfalsifiable theories about the world. It will always be true, and no one could ever prove it wrong. Any and every attempt either proves it right, or you’d be told that you didn’t try hard enough – try again.
This was a huge problem in the early days of Science in the 19th and early 20th century. Scientists were coming up with theories involving invisible entities that couldn’t be tested, and this annoyed a number of philosophers and scientists.
Their solution? Hypotheses must be testable! The famous philosopher of science, Karl Popper went a step further. He argued that seeking confirmation to prove a hypothesis is not sufficient. As long as you do this, you’re answers are only inductively true, and do not have the kind of deductive certainty as 1+1=2.
Just in case you got a bit lost with the terms, “induction” and “deduction,” let me explain. Induction is a process of inferring a general claim (or theory) based on an observation of samples. For example, if the bus comes at 8am every morning for the past two weeks, it is reasonable to conclude by induction that the bus will come at 8am tomorrow morning. Similarly, if all I see around me are swans that are white, it is reasonable to conclude by induction that all swans are white… until we encounter a black swan! For centuries, most of Europe were quite certain that swans are white until they came to Australia where they encountered black swans for the first time. This was sensational news that rocked Europe.
The point is, theories are right and continue to be right until the day we encounter evidence proving that they’re wrong. They are never 100% true, only tentatively true until proven wrong.
Deduction, on the other hand, is way more certain than induction. The rule of deduction is that as long as the premises are true, and if the conclusion follows from the premises, the conclusion is necessarily 100% true. The usual example cited is this: “All humans are mortal. Socrates is a human. Therefore Socrates is mortal.” The only way you can debunk a deductive conclusion is to first demonstrate that at least one of the premises are false. In which case, the conclusion becomes uncertain.
How do we get deductive 100% certainty in science? Popper says that we can get them through falsification.
Premise 1: If the hypothesis, “all swans are white” is true, there will not be any non-white swans.
Premise 2: Here is an observation of a black (non-white) swan.
Conclusion: Therefore, the hypothesis, “all swans are white,” is false.
While we may never be sure about what’s right, we can be very very certain about things that have been falsified. It seems like the advancement of science is not about what we know about the world, but a treasure trove of all the things we know is not about the world. This may seem like a very counter-intuitive view of science, but don’t be too quick to dismiss this. Like the hypothesis, “all swans are white,” we just can’t be sure when a theory is wrong until that black swan evidence surface. It could happen tomorrow, or one century later. We could never know.
Thus, in Popper’s view, a good scientist doesn’t just try to make new discoveries about the world. A good scientist does whatever he/she can to falsify what we know, in the hope of relearning, rediscovering something new that we didn’t know before. And if a theory continues to stand strong despite numerous attempts at falsification, we can be sure that that theory is a really good one.
For us to effectively falsify theories, we need to be clear about the way we formulate hypotheses. The problem with unfalsifiable hypotheses like destiny, is that there is no way to prove it wrong. A lot of unfalsifiable hypotheses stem from the statements being vague and ambiguous. By leaving things vague and ambiguous, we allow for a multiplicity of interpretations, and so a variety of data could be subsumed under these interpretations, thus leaving little or no room for a chance to be proven false.
Consider this hypothesis:
Hypothesis 1: There will be war.
It’s so broad and generic, so vague and ambiguous, there is no way to falsify this hypothesis. War could happen tomorrow, next week, next year, or next century anywhere in the world (or in the galaxy, for that matter). If there isn’t war now, there could be war in the future. There is just no possibility for this statement to ever be wrong. And so, if I say this today and war breaks out 400 years later, I am a fortune teller!
If vagueness and ambiguity is the enemy, then what we need is to strengthen our hypothesis with clarity. Let’s consider this hypothesis:
Hypothesis 2: War will break out by the end of this week.
Hypothesis 2 is better than hypothesis 1, as we now have a deadline. If war doesn’t break out by the end of this week, it is clear that Hypothesis 2 will be proven wrong.
However, the hypothesis still lacks precision. What kind of war are we referring to? Nuclear, conventional, chemical, cyber, or are we talking about a trade war, or a war of words? Where will the war take place? Physically in a particular area on Earth? Or are we referring to a cyber war that will occur online? There are too many interpretations, and far too many empirical data could potentially be considered confirmation for the hypothesis, rather than as disconfirmation.
If I wrote hypothesis 2 thinking that a nuclear war will happen in the Korean peninsula, but some kind of strange penguin civil war breaks out in Antarctica, I’m still right – that weird penguin war is still a confirmation of my hypothesis.
So we need to do better than this. More clarity is needed to expel the vagueness and ambiguity in the hypothesis. Can you see how the conceptual tools in the previous chapter is relevant here? So let’s try this hypothesis:
Hypothesis 3: Nuclear war will break out by the end of this week in the Korean peninsula.
This is much better. Now, we have a deadline, a place, and the type of war. There are three ways in which my hypothesis could be tested and proven wrong.
Of course, we can continue to push ourselves to do much better. But the point is that good scientific theories demand precision. The problem with Science is that in some cases, it is very difficult for scientists to formulate something as precise as Hypothesis 3, so Hypothesis 2 is tolerated until more can be known.
5. Confounders and Conclusions
Let’s talk about Science without really talking about Science. We’re gonna be ghost hunters! This might seem completely unrelated to Science, but this will be a useful exercise for us, as a lot of new research in the Sciences deal with the unknown, with things that we’ve never encountered before. How do you even know what to test for, or how to run an experiment, if you don’t quite know what you are dealing with?
How might we design an experiment to test for ghosts? Well, I’m no ghost expert, but at least from the movies and stories I’ve heard, ghosts tend to mess around with electro-magnetic signals. I don’t know if they generate their own electro-magnetic waves, or if their presence causes an alteration to existing waves. Nonetheless, I could use a sophisticated electro-magnetic wave detector and bring it to a well-known haunted location.
Suppose this happened: While standing in the haunted location, my detector produces a positive reading, a sudden sharp spike on the meter (let’s say from 0dB to 100dB and back in 2s), or the meter displays an increase for a period of time (100db for 5 minutes), can we conclude that we have found ghosts?
Not quite. There could be confounders, things that mess up our readings. There are many alternative explanations for why my detector could produce a positive signal. The occurrence of a solar flare, someone nearby could be sending out a message over radio frequencies, or I might have just received a call on my mobile phone. These are enough to interfere with my detector.
A good scientist will do whatever’s possible to eliminate these confounders. Laboratories are always preferred where possible as these are highly controlled environments where we can eliminate as many confounders as we can. But there aren’t haunted laboratories to visit, and we might chase ghosts away if we were to build a laboratory here. So what else could we do?
Suppose we were able to eliminate those external confounders, e.g. we have installed thick lead walls around the haunted area and now we remotely monitor the detector without having to be physically present. Now, we find our detector sensing a 100dB electro-magnetic wave despite all that we’ve done. Is that a sign of a ghost? Maybe it is, but it could just be the presence of other confounders that we have not considered. So again, we need to go back to the drawing board and consider the alternative explanations (it could have been caused by the remote monitoring of the detector, as it wirelessly sends data to us in a faraway location).
Let’s turn the scenario around. What happens if you DON’T find anything? Can you conclude that there are no ghosts? No, not at all! It could be that the ghosts just aren’t there when you were doing the study. Humans are often more terrifying than ghosts, so maybe they ran away when we came to the haunted site! Or it could be that they don’t effect or alter electro-magnetic waves.
But you also cannot conclude that ghosts do not respond to electro-magnetic readings. You don’t even know if they exist, so how can you be sure of any positive property of ghosts? You also cannot conclude that ghosts that respond to electro-magnetic waves do not exist. You don’t know this either. They might exist, but they might be somewhere else. Just as how kangaroos are found only in Australia, so maybe those ghosts exist in a particular part of the world.
What you can conclude is this: We have not found ghosts that produce electro-magnetic effects.
Just as how we require our hypothesis to be as clear and precise as possible, notice how good science demands that the conclusions ought to be as clear and precise. For that matter, especially when we are dealing with the unknown, good science yields very modest conclusions. A telling sign of pseudo-science or bad science is the exaggerated claim like, “Ibuprofen causes male infertility.” That requires a huge jump in logic from the experiment that was done, to such a conclusion.
We should be extra careful and aware of this, because even scientists err in this way, writing conclusions that do not immediately follow from the experiment. If a scientist did a study on the effects of caffeine on mice, the scientist can only validly conclude, based on the experiment, about the effects on mice. Anything about humans is only speculative and not demonstrated at all by the experiment. We have to be extra careful about that.
Let’s review the way we approached this ghost-detecting experiment. To put it simply, if we find positive readings on our detector, it could be confirming evidence to the hypothesis, “Ghosts exist.” If we find nothing, can we conclude that the lack of evidence is disconfirmation of the hypothesis that “Ghosts exist”?
A similar situation would be if you lost your keys. You’re quite sure you lost it in your house. You searched high and low, but couldn’t find it. You tell your mother (or your spouse) that you searched the house but couldn’t find it. Chances are, your mother (or spouse) would say: “You didn’t search hard enough. Try again.”
Similarly, if you told your mother (or your Principal Investigator) that you couldn’t find ghosts, she’d probably tell you to try harder. In that case, what would suffice as disconfirming evidence for the hypothesis, “Ghosts exist?” Or is this statement unfalsifiable?
What if my hypothesis was “Ghosts do not exist”? It’s clear that any positive reading on my detector could potentially function as disconfirmation of the hypothesis. But if we do not find ghosts from our experiments, can the absence of evidence confirm the hypothesis that “Ghosts do not exist”?
I’ll leave you to think about this.
6. The Absence of Evidence
If you’re unfamiliar with Singapore, Singapore has the harshest laws against the trafficking of drugs. If you traffic drugs above a certain amount, the death penalty awaits you. In fact, if you ever flown on a plane to Singapore, some pilots will, just before landing, announce that if you are carrying drugs, the death penalty awaits and – Welcome to Singapore!
Let’s now imagine this scenario. You’re not a drug trafficker, nor are you a drug abuser. You’re a good citizen, and you’re travelling alone from some country to Singapore (whether it’s a return trip from a holiday, or you’re visiting as a tourist, or something). Just after you’ve stamped your passport, a customs officer asks to check your carry-on bag. To your horror, the officer found a packet of drugs in your bag. It contains an amount that warrants the death penalty. You have no idea how it got there. And unfortunately for you, while you were on the plane, you reached into your bag to grab your headphones, and accidentally left your fingerprints on that packet of drugs (there are no other fingerprints). So there’s two positive evidence linking you to the drugs: it’s in your bag, and your fingerprints are on it.
How do you prove that you’re innocent? (Before reading on, do consider what you can do to prove your innocence.)
You might say, “If you check my background, I am clean. I have no criminal records.” But that’s true for every criminal caught for the very first time. It doesn’t prove that you’re innocent.
You might say, “Check the CCTV of the previous airport, on the plane, and here.” If the officer finds a video footage of someone planting the drugs, you’re innocent! But, what if the officer finds no such thing? Does it mean that you’re innocent? Not at all!
You might say, “If you check my bank account, there’s no monetary evidence that I’ve been paid to do this.” True, but it might be that the drug syndicate would only credit money into your account later, or they might pay you in cash. So it doesn’t prove you’re innocent.
You might be tempted to suggest undergoing a polygraph (lie-detector) test. But that’s pseudo-science, and doesn’t actually determine whether you’re telling the truth or not.
Try as you might, it’s really really difficult to prove that you are innocent.
The point of this exercise is to highlight to you that not all claims are of equal weight. Some claims are much harder to prove, some are easy to prove. All I need to prove that you are guilty of a crime, is just one piece of evidence. A fingerprint, a trace of your hair, or a video recording of you doing the deed, etc. But to prove that you are innocent, I need a whole lot more. The problem with proving innocence is that you are trying to use the fact that you didn’t do any crime (the lack of such evidence) as evidence of your innocence. That is an incredibly difficult task!
And this brings me to an important saying:
The absence of evidence is NOT the evidence of absence.
The absence of evidence can be interpreted either way. The lack of video evidence showing someone planting drugs in your carry-on bag could mean that you are guilty (since no one was caught planting it), or it could mean that you’re innocent (since it could also mean that someone planted it in an area without cameras).
When we are trying to confirm or disconfirm a hypothesis, we must question and be aware of the type of claim that we are dealing with and what we’re up against.
Thus far, I’ve talked about how the absence of evidence cannot confirm the existence of ghosts nor can it prove that you are not a drug trafficker. But here’s a different scenario: I have a bowl of noodles. My hypothesis is, “This bowl does not contain fishball noodles.” My evidence? There are no fishballs in my noodles! It would be insane to say in this context that, “The absence of evidence is not the evidence of absence!” Clearly there are no fishballs in my bowl, and in this instance, the absence is confirmation that this is not a bowl of fishball noodles.
Why is the bowl of non-fishball noodles different from ghosts and drugs? In the case of the noodles, I’m only concerned with one particular bowl, and fishballs are quite tangible unlike ghosts. So, I can search my bowl very thoroughly and not find fishballs. But in the case of ghosts, not only am I dealing with the scope of where I’m searching (a room, a neighbourhood, or the universe), but I am unsure what would even qualify as evidence for ghosts (Electro-magnetic waves? Heat signatures?). I can exhaust each possibility, but there will still be many many possibilities left. Thus, we cannot validly conclude the absence as confirmation for the hypothesis, “There are no ghosts.” You’d just be asked to try harder.
As I’ve shown above, we can be easily confused as to when the absence of evidence can or cannot be evidence of absence. This becomes a huge problem when we’re dealing with science scepticism or pseudo-science, especially since a lot of the controversies deal with issues of absence. Some claims are harder to prove than others. But because we are unaware of this, we can stumble when we’re investigating.
7. Scientific Consensus and the Burden of Proof
If the absence of evidence is not the evidence of absence, why are we so sure that unicorns don’t exist? Clearly there’s no evidence that unicorns exist, but at the same time, there’s also no evidence that unicorns don’t exist either. Does this mean that science cannot answer this question?
More importantly, why are we more inclined to think that it’s rational to believe that unicorns don’t exist?
To prove that unicorns exist, we only need one piece of evidence. To prove that unicorns don’t exist, we need a whole lot more evidence (just like the drug trafficking case earlier). Considering how it’s a lot easier to prove that unicorns exist (since you only need one piece of evidence), the burden of proof falls on those who claim that unicorns exist to prove to the rest of the world of their existence. Yet, to this day, no one has come forward to present real evidence of the existence of unicorns.
On the other hand, there are more scientists out there who have tried to search for unicorns in their own way and found none. Some have even examined supposed proof of unicorns, but found them to be a hoax. From a scientific perspective, there is scientific consensus that there are no unicorns.
This doesn’t mean that the scientists possess the truth, that they’re 100% correct. It just means that it’s more reasonable to believe them that unicorns don’t exist, since the people who are supposed to prove it (since it’s easier to prove it) have not done so, and that the scientific community have tried to find it but found none. To prove all these people wrong, one has an uphill and difficult – but not impossible – task of convincing the rest.
How do we know that something like climate change is indeed happening now? We didn’t come to such a conclusion because one scientist went out to test for it. Instead, thousands and thousands of scientists independently did their own thing, within their disciplines, and in the process found significant changes in the climate to produce the phenomena they studied. This was a consensus that arrived from independent research. Thus, the burden of proof thus falls on the people who deny it. They need to show why all these peoples’ research are wrong.
The same applies for new controversial issues in science. One study isn’t enough to settle the matter. Does coffee cause cancer? Do wifi signals cause cancer? We see articles like this all the time with contradictory answers. More time is needed to research on these issues before the scientific community arrives as a consensus.
Why am I bothering to talk about the burden of proof? Well, when we are challenging a claim, or when we encounter someone challenging a claim, we need to assess who has the burden of proof. One trick often used by promoters of pseudo-science (or people who just want to win arguments) is that they will unfairly shift the burden.
Conspiracy theorists have been claiming that Mark Zuckerberg, the founder of Facebook, is a shape-shifting lizard person whose secret mission is to destroy humanity. This sounds crazy, but it’s even crazier that people would believe it. Nonetheless, if you say, “I don’t believe that Mark Zuckerberg is a lizard person.”, they might reply:
“What’s your proof that he’s NOT a lizard person?”
A question like this will probably leave you stuck, unable to respond. How do you even answer a question like this?
In such unfair debates, the silence is taken as a defeat. The conspiracy theorist has won the argument, thinking that by forcing you into a corner where you can’t respond, you have failed to defend your claim. But that’s only because the conspiracy theorist had unfairly shifted the burden to you. Proving that Mark Zuckerberg is not a lizard is a much harder claim to prove than that he is a lizard. So it’s important to be vigilant of such unfair shifting.
As I have shown in the preceding sections, what separates good science from bad or pseudo-science stems from the kinds of rigour needed to formulate clear hypotheses open to testing and disconfirmation, to the rigours of carefully distinguishing causation from correlation, to being careful about how we select our sampling to avoid skewing or bias, to checking and eliminating confounders in the process, and lastly, to be cautious in formulating a conclusion to ensure that it immediately follows from the experiment, rather than making a big leap of logic. At the same time, it is also important for us to distinguish between claims: some are harder to confirm/disconfirm than other claims, and so much care is needed to question if we are dealing with a potentially unfalsifiable hypothesis, or if the burden of proof is unfairly shifted to us.
Far too many pseudo-science are marketed as scientific, especially in the area of wellness. So it’s not sufficient to refer to a standard definition of science (vs. pseudo-science). We should be assessing the process by which they arrived at their results. And similarly, we should be strive to be as rigorous as we can in our own knowledge-making processes so that we can avoid creating pseudo-scientific myths for ourselves.