What is this Science thing you speak of?
Teach me how to talk Science.
In our previous post (see here), we learned that Science is not just a collection of facts. Science is best understood as a process and the scientific method is the single most consistently reliable and successful way of acquiring knowledge.
Despite that, often when talking about a well established scientific theory, we might hear someone say, ‘well, that’s just a theory’ as if it’s a bad thing.
If we’re going to learn about the scientific method, we’re going to have to speak the same language.
Words like fact, theory, hypothesis, and law mean something totally different to a scientist than the way they’re used in everyday speech. To allow us to share a common vocabulary, let’s take a look at each of these words and attempt to define them rigorously.
What are facts?
Facts are observations confirmed to a degree such that it would be perverse to withhold provisional assent. It’s provisional because Science does not claim absolute certainty, it approaches truth asymptotically. This epistemic humility shields science from becoming unchallenged dogma.
What kind of observations?
Precisely measured and methodologically recorded observations. They may be direct or indirect. Sometimes indirect observations are more reliable than direct observations.
For instance, it is a fact that the fossil record holds hundreds of transitional species even according to its strictest definition, and that both microevolution and macroevolution have been directly and indirectly observed.
And a hypothesis would be?
A hypothesis is a proposed explanation for a given set of facts. A good hypothesis must have strong predictability power and must be, in potentia, falsifiable. Put simply, it must have testable predictions that have a capacity to be proven otherwise.
For instance, Einstein’s prediction that massive objects such as the sun, warp space-time enough to throw passing rays of light off-course was both testable and falsifiable.
During a total solar eclipse on 29th May 1919, when the moon blocked out sun’s overwhelming glare as it was passing close to a bright group of background stars called the Hyades, an English astronomer Arthur Eddington verified Einstein’s prediction.
The news of this discovery made headlines worldwide, with the London Times proclaiming: “Revolution in Science / New Theory of the Universe / Newtonian Ideas Overthrown” making Einstein a household name. It is this criterion of falsifiability that demarcates Science from pseudoscience. If your hypothesis is unfalsifiable, even in principle, then it is a weakness of your idea and not a strength (more here)
What is a scientific theory?
A hypothesis which turns out to be a well-substantiated explanation of the given set of facts, trying to incorporate all available evidence, that is tested for its predictability power under conditions of falsifiability using repeatable experiments which are demonstrably consistent to a reasonable degree, is then promoted to the status of a theory.
The theory of evolution by natural selection, for example, is one such scientific theory that successfully explains the fact of evolution.
Where does a law fall into all this? Will a theory ever become a law?
A scientific law is a mathematical description of an observed phenomenon. A relationship between two or more parameters of models that serve as best approximations for mimicking natural phenomenon. It doesn’t explain why the phenomenon exists or what causes it. The explanation of the phenomenon is given by the theory.
For instance, Newton’s Law of Gravitation describes the attractive force between two bodies with masses which are some distance apart while Einstein’s General Theory of Relativity explains how gravity works.
A theory will never become a law. Laws are descriptive in nature, that is they merely describe the natural phenomenon and have no prescriptive value. A scientific law simply states what happens. A scientific theory, on the other hand, explains how and why it happens.
That we are familiar with some scientific jargon, we shall now see what the scientific method looks like in practice.
The Scientific Method
A mere five hundred years ago, nearly everyone believed the Sun, Moon, and other planets rotated around the Earth. Our story of moving away from the geocentric model of the universe and moving towards the heliocentric one begins with Nicolaus Copernicus.
Copernicus wrote out a short overview known as the ‘Commentariolus’, or ‘Brief Sketch’ describing his ideas about the heliocentric hypothesis. It was a paradigm shift, a new way of thinking about the same natural phenomenon that most people were casual observers to.
The next character in our story, Tycho Brahe made great strides forward by devising a system that allowed him to measure and catalog precise observations. Tycho and his team of astronomers compiled astronomical observations that were vastly more accurate than those made before.
Brahe’s extensive observations were then used by Johannes Kepler to make remarkable breakthroughs in astronomy. Kepler and his three laws of planetary motion, informed by Brahe’s raw data suggested that planets traveled in ellipses and that the Sun does not sit directly in the center of an orbit but at one of the foci.
Eventually, Galileo Galilei championed heliocentrism by verifying Kepler’s laws with a Galilean telescope and published ‘Dialogue Concerning the Two Chief World Systems’, a book comparing the Copernican system (the Earth and other planets orbit the Sun) with the traditional Ptolemaic system (everything in the Universe circles around the Earth).
This journey from Geocentrism to Heliocentrism captures some of the key elements of modern scientific methods, viz. observation, hypothesis, testing, verification, and publishing results. Let’s unpack this framework further for it entails a lot more than meets the eye.
Scientific discoveries usually start with a playful and curious mind at work. The first step entails methodological, precise, measured observations and data cataloging for any natural phenomenon.
The second essential step involves a process of pattern recognition. Induction or simply the process of forming generalizations about sets through a limited sampling of subsets is at the heart of all sciences.
If samplings of a particular data subset are indeed indicative of the behavior of the whole, then induction will lead to accurate probabilistic predictions. If the subset is not indicative of the whole, then sooner or later it would let us know in the form of failed empirical predictions.
The explanatory power of models thus generated when coupled with the predictive power of logical induction births robust models that are then subjected to several experimentations. These experiments help verify, falsify, and/or modify our best ideas until they reach some correspondence with reality. In other words, scientific models are by their very nature, self-correcting.
All it takes is the intellectual honesty to admit when our conclusions are wrong so that we can modify them in the face of new information. However, there is always more than one possible explanation for any given set of data and different ways of interpreting results from any given experiment.
This is one of the reasons why scientists argue, discuss, and debate so much. Findings are published and peer-reviewed. Methodologies are critiqued and conclusions challenged. Incentives are arranged such that proving an idea to be false is more rewarding than showing it to be right. Scientists are encouraged to be the harshest critics of their favorite ideas.
The scientific methods are thus designed to keep biases in check and our ideas about the world uncorrupted, which is unique from other methods of acquiring knowledge. The scientific method does not guarantee truth, but it does draw conclusions with a much higher degree of reliability than other systems of knowledge.
The Principle of Fallibilism
Scientific theories are not held because they are true in any philosophical sense. Rather, they are held because they are able to withstand a continuous barrage of attempts at falsification. The principle of fallibilism, simply states that no model which seeks to describe the external world can ever be assigned to be ‘true’ with any kind of perfect, universal certainty.
No set of positive results can ever conclusively demonstrate an absolute correspondence of a model with reality. Even if every experiment we ever run were to perfectly coincide with our expectations, there will always exist room for some philosophical doubt. Consequently, all scientific claims about objective reality are therefore provisional, i.e. they always remain open to possible revision when faced with newer and better information.
The Principle of Falsifiability
In order for an explanation to be considered scientific, there must exist some sort of test that has the potential to disprove its validity. If the reaction coincides with our expectations, then the hypothesis is merely supported. But if the reaction fails to meet our expectations, the hypothesis is immediately refuted or falsified.
Thus, in contrast to our inability of assigning any model as ‘true’ with perfect certainty, we can indeed be perfectly confident in assigning certain models a value of ‘false’. That’s because the very definition of a false propositional model is one whose empirical predictions fail to come to pass. In principle, although both ‘true’ and ‘false’ beliefs may guide us to actualize empirical predictions, only ‘false’ beliefs will consistently fail in that goal.
Principle of Parsimony
Pragmatic scientific method also helps us quantify Occam’s Razor.
If two models happen to make perfectly equivalent predictions, then the model containing fewer assumptions is automatically preferable. After all, if both models are empirically equivalent either way, then you might as well just go with the one that does not multiply entities beyond necessity.
The Null Hypothesis
A good scientific hypothesis must have the ability to predict events with measurable empirical manifestations (that which have an impact on our sensory experience). But there’s always an intrinsic epistemic disparity between a positive claim and its null.
A claim such as – I can fly – carries with it a series of predictions about my actions and their ultimate consequences. But the null carries no predictions at all, other than the continuing absence of any particular manifestations of flight. The physical expectations of the null hypothesis are therefore immediately considered to be satisfied, by default.
That’s why the burden of proof always lies with the person making the positive claim, and never with the person who rejects it. It’s important to realize that this is more than just some passing philosophical nuance, but a very real, practical principle that governs all of our daily lives, be it law, public discourse, and/or all forms of inferential statistics.
However, choosing the null hypothesis can sometimes be tricky. The difference between – evidence that something is absent, and simple absence of evidence – can be nuanced.
Consider the following claims:
- There’s a needle in the haystack.
- There’s an elephant in the room.
The probability of finding a needle in the haystack, given there, is in fact, a needle in the stack ~ 0
The probability of finding an elephant in the room, given there is, in fact, an elephant in the room ~ 1
For the first claim, it is reasonable to suggest that our inability to find the needle does not imply that the pin is not there. For the second claim, it is reasonable to suggest that our inability to find the elephant does imply that the elephant is not there. This is a basic epistemic principle called inference to the best explanation.
If evidence is lacking when we expect it to be abundant, then it very much allows us to dismiss a hypothesis, and absence of evidence then does become evidence of absence.
The ideas of the null hypothesis and the burden of proof are exactly why fallibilism and falsification are such integral aspects of the scientific method.
In our previous post (see here) we had a brief look at epistemology in philosophy of Science:
- Principle of Mental incorrigibility, or simply Empiricism.
- Axiomatic formalism, or for this discussion, Rationalism.
- and finally, we nested both of them inside Pragmatism.
In this post, we looked at the vocabulary and the methodology of Science:
- Facts, theory, hypothesis, and law
- Principles of fallibilism, falsifiability, and parsimony
- The null hypothesis and the burden of proof
In our next post, we shall uncover the nature of scientific explanations or the ontology in the philosophy of Science. Don’t forget to share this post with your friends and help increase their scientific literacy.
Follow Epistemic Worries for Philosophy, Science, and Philosophy of Science. What are your thoughts on the key elements of the scientific method?
Leave your comments below.
This story is published in Noteworthy, where 10,000+ readers come every day to learn about the people & ideas shaping the products we love.
Follow our publication to see more product & design stories featured by the Journal team.
