The limits of science


In a lot of ways Science has become the religion of the day. We can’t go more than a day to hear / see / read a "news" story about "scientists" saying something about something important. We can’t help but feel dazzled, confused, perplexed, overwhelmed by these announcements. And we have discussions with others:

  • I heard that scientists say that X cures cancer.
  • Didn’t you hear? Scientist announced that X causes cancer How is this really different from saying "my shaman said"?

The core principles of science

I’m going to argue that scientific thinking is really in a different league – but it also has many limitations. The core ideas of scientific thinking are:

  • We can observe things and those observations have some relation to the objective reality
  • This objective reality is probabilistically deterministic (ie. if we flip a coin it will land either heads or tails but it generally won’t turn into a unicorn and fly away)
  • Based on our observations we can construct models / hypothesis about cause and effect. While there can be an infinite amount of hypothesis, we assume that reality has a simple and elegant structure and thus prefer the hypothesis which explains the largest category of events with the minimal assumptions
  • Scientific hypothesis need to be testable (also called "falsifiable") – ie. they need to be forward looking / have predictive power. For example if I say "object alway fall downwards" then you or I can take a ball, a key, a rock, etc and verify that indeed, when I let go of it, it falls towards the ground. However such positive outcomes have little value (they increase the likelihood that the hypothesis / theory is true only by a small amount) and the most valuable things are falsifications – when somebody makes a prediction based on the theory but then a different outcome is observed – which most likely means that the theory is false (unless there was an error in the experiment).

This looks an awful lot like "the ten commandments of the science religion", doesn’t it? Because there is no reason to believe them intrinsically. There are only two things in favor of this mindset:

This is how our world seems to work – and thus these principles seem intrinsically true to a lot of us. Then again this "gut feeling" is no different from being convinced that a given kind of deity governs our lives

What is different from every other religion/philosophy however is that it contains the framework to extend it. You just observe, theorize and then try to falsify your theory. Voilà! You’re adding to the scientific knowledge. All other religions are closed while the religion of science is open.

Science is not "truth"

You might have observed that the principles – as described in the previous section – have somewhat of a wishy-washy nature. I keep using words like "likely" and "usually".

For example I said "if we flip a coin it will land either heads or tails but it generally won’t turn into a unicorn and fly away". I can’t say for certain that it won’t turn into a unicorn, but so far nobody reported a case of this happening, but we observed the coin landing heads or tails a lot of times so we’ll assign a very small probability to something else happening.

Scientific results are always probabilistic, but that is just life: if I go out tomorrow I might get hit by a car, but I most probably won’t. Note that other religions generally avoid saying anything about "this life/world" and reserve their proclamations for "the other/after life". At least science is trying to make some predictions, even though they sometimes turn out to be false.

This bears repeating: Science is not "truth". There is no such thing as a "scientific fact". Our scientific knowledge is simply a collection of theories which we failed to falsify – as of yet. Now, some of those theories have been around for a long time and have been found true in many situations and it’s prudent to do to act as if they were the absolute truth, but we must accept the fact that there is always the small possibility that they might turn out to be false.

To pound some more this idea: all the above is true even if we would "do science perfectly". However we are humans: subject to our biases, feelings and other motivations.

Science is not math

(a couple of words about induction)

Related to "there is no scientific truth" is the mirage of mathematics in science, which goes something like this: "we all know that 2+2 is 4, this scientist is using a mathematical formula, so whatever comes out of the formula must be true".

No!

Mathematics and science use very different ways of reasoning:

In math we use deduction: we state some ground rules ("axioms") and from those we deduce (prove) all the other statements. All the deduced statements are "true" (if we didn’t do any mistakes) since they are derived from the axioms which we assume to be true. This doesn’t mean however that it necessarily has any relation to the objective reality. (And just a side-note: even if we stay within the bounds of such imaginary systems – ie. don’t try to apply it to the "real world" – we will hit some fundamental limitations ^1)

Science however uses induction: if I see that both an apple and a rock is failing down, I guess that both "falling down" events happen because of the same cause. Such "rules of thumbs" ^2 work generally but can give the "wrong" results sometimes (as opposed to the rules of deduction which are always correct if we accept the initial axioms). Of course for a scientist these cases of "unexpected outcome" are the most interesting ones since they signal an opportunity to learn something new, but they can be most distressing if we think of science as "the source of truth".

So how does science use mathematics? It is just a precise way to describe hypothesis / theories and to manipulate those. Thus we might say that the maximum distance of a ball thrown is described by the equation v^2 / g and we can make predictions about the distance of a ball will travel before throwing it (and verify after the fact that prediction was – mostly – correct) but the fact that we formulated the theory in mathematical term does not make it fundamentally more true than any other scientific theory. It still is "the best theory we have for now which isn’t refuted by evidence".

Other limitations of science

Coming back to the idea "science is not the absolute truth even if we would "do science perfectly":

We don’t do science perfectly:

  • We have biases and when we have a theory we might not look "hard enough" for evidence to refute it
  • The academic structure is not set up to encourage verification of results: publishing replication results or negative results is discouraged (may scientific journals don’t even accept them for publication)
  • Science is entirely probabilistic (it only tells you what probably is true), however statistics (the branch of mathematics which deals with probabilities) is complicated and it’s easy to make mistakes when judging what constitutes probably true (Is Most Published Research Wrong?)
  • In some very interesting fields it is hard to conduct experiments (more on this in the next section)

And as if the above wasn’t enough, there is the problem of communicating the results (the step where "there is a 60% probability that eating a bit of chocolate improves bone density in white women over 50" turns into "Science Fact! Women should eat chocolate daily!").

To give some uplifting news: these issues are known and people are trying to work on it. There have been scientific journals set up to publish replicating studies and/or studies with negative results. There is a movement to encourage researchers to pre-register their studies (to state what data they’ll collect and how they will analyze it) and publish the results even if they are negative. Finally some journals require scientists to give a "simplified abstract" when publishing the research which can be adapted by journalists easier.

However these are hard issues and we can help out by not having unrealistic expectations.

Words about human-centric fields

A couple of final words about the science in fields like health (mental and physical) or economics: these are the fields which are the most important to us but where the difficulties presented multiply:

  • Generally we can’t do random double-bind trials where we select a group of people and infect them with AIDS or make them live on less than $1 a day (you know, because we value human life)
  • This means that we can only study people who already are in this situation but that makes it very likely that we confuse cause and effect
  • Even if the experiment is non-intrusive (or we’re doing an observational study) it is very hard to get a diverse set of participants (ie. most psychology experiments are done on young white males in the US – no wonder that they don’t replicate across the world)

Again, people are working on addressing these issues, but it is just one more reason not to believe the "fad" articles published daily and shared virally on social media.

Science is not perfect, but it’s the best that we have.


Leave a Reply

Your email address will not be published.