Lecture 3 HSI1000

3 Scientific Explanations and Models

Intended Learning Outcomes for Lecture 03
You should be able to do the following after this lecture.

(1) Define a scientific explanation and discuss the two basic ways in which a theory differs from a
hypothesis.
(2) Explain the difference between two events being correlated and being related as cause and
effect.
(3) Discuss the basic features of the following types of scientific explanations: (a) cause and effect,
(b) causal mechanism, (c) underlying processes, (d) laws, (e) function.
(4) Explain Occam’s Razor, illustrating its use with an example.
(5) Discuss what is a scientific model, the different types and their purpose, and explain the
difference between a model and a theory illustrating your explanation with an example.

3.1 Scientific Explanations – Theories and Hypotheses

In this lecture we’ll be continuing our in-depth look into the scientific inquiry by following chapter 3
of our textbook entitled “Explanation”. The five intended learning outcomes, or 5 ILOs for short, for
this lecture are listed above, and we shall be addressing each of them with its own video. In this video
we’ll be tackling the first ILO, so we’ll be discussing what a scientific explanation is and examining the
meaning of the scientific terms “theory” and “hypothesis”.

3.1.1 Scientific Explanations Must be Testable

What is a scientific explanation? It’s an account of how or why something is the case, but what makes
the explanation scientific is that it MUST be testable. We’ve already seen a number of examples of
this in our first lecture. In that lecture we saw explanations tested in several situations. If the tests
failed, then the explanation needed to be revised, or an entirely new one provided. But let’s consider
an example of an explanation that can’t be tested.
You and your friend are having a coffee together. Just as you finish your last drink of coffee, you
misplace your mug on the table, it teeters and then falls. You quickly try to catch it, but miss, and it
smashes on the floor. Your friend then explains,
“Ah, never mind, it was fate that the mug should be broken. It was meant to be.”
Here we have an explanation of why the mug broke— it was fate. But what if you had caught the
mug, wouldn’t that also be fate? In such a case, you were apparently meant to catch the mug. In fact,
no matter whether you catch or miss the mug you could say that it was fate and meant to be.
Do you see? Here we have a situation where no matter what happens, the explanation stands.
There’s no way of testing whether it really was fate that was responsible for you not catching the
mug. Fate could be the explanation for why you missed catching the mug, but it’s not a scientific

1
 Lecture 3 HSI1000

explanation because it’s untestable. Therefore, there is no way for us to know for sure whether the
explanation is correct or not.
The bottom line is that a scientific explanation for something must be subject to falsification. If there’s
no way of demonstrating that the explanation is false through experimentation, or “reality checks”,
then the explanation is not scientific.

3.1.2 Theories and Hypotheses

Now scientific explanations are frequently associated with two terms – “theory” and “hypothesis”.
Let’s talk about hypotheses first.
3.1.2.1 Hypotheses
What characterizes a hypothesis is that it is tentative or unproven, that is, it hasn’t yet been subjected
to any testing or falsification. A hypothesis could be just a simple vague hunch about what might be
going on, or it could be a finely detailed (but still speculative) account of how or why something is
the case. Regardless of the form of explanation a hypothesis might take, it is always tentative or
unproven, and it’s usually the first step to discovering something new about nature.
3.1.2.2 Theories
Now let’s talk about the term “theory”, which is a bit trickier. In regular English usage, the term
“theory” is often used to denote what is actually a hypothesis. However, from the scientific
standpoint, that usage it wrong. In fact, it’s because of this misunderstanding that many lay people
believe that a scientific theory is unproven or hasn’t been subjected to testing.
A scientific theory does NOT imply a wholly untested idea, or worse, a complete guess about the
reasons for a limited range of phenomena, a single event or fact. Such explanations are hypotheses,
as I’ve already said. A scientific theory is characterized by the breadth and depth of its explanatory
power. Scientific theories are more general structures capable of explaining a much wider variety of
phenomena than a hypothesis. A theory is a conceptual framework for providing explanations.
A scientific theory will often contain experimentally well-tested, and therefore, well confirmed rules
and principles that reveal underlying explanatory similarities between what might seem to be quite
diverse phenomena. Let’s consider some examples to understand what this means.
Well-established Theories
Consider Newton’s three laws of motion and the law of gravitation introduced in the first lecture
during the scientific revolution. These laws constitute a scientific theory – an extremely powerful one
that has been tested over and over again, and only breaks down under the most extreme conditions.
These laws can, and still do, explain and predict the motions of virtually all objects on earth and in
space. They serve as a “conceptual framework” for providing explanations for how and why a vast
number of things move about all around us.
Recall the germ theory mentioned in the first lecture. This theory, having been tested repeatedly,
now explains a vast array of diseases exhibited in not just people, but all kinds of organisms. Again,
it represents a conceptual framework for explanations of a vast array of diseases and provides insight
into their treatment.

2
 Lecture 3 HSI1000

Science contains many such theories – theories that have been repeatedly put to the test and provide
breadth and depth in their explanatory power. Theories like the theory of evolution, the big-bang
theory, atomic theory, statistical mechanics, the kinetic theory of gases, the theories of special and
general relativity, quantum mechanics. The list goes on and on.
The theories I just mentioned all represent “well-established theories”. They are “well-established”
because they have been tested numerous times in many ways and continue to pass all the tests, but
not all theories are like this.
Obsolete or Superseded Theories
Often these well-established theories have superseded previous, older theories. Theories that were
thought to be correct at the time, but eventually came undone because of new evidence that proved
them wrong – like the Aristotelian world view we discussed during the first lecture. These superseded
theories, now known to be wrong, are still, to this day, called theories. Here’s just one example.
I mentioned germ theory as
currently being the reining theory for
the cause of many diseases. But for
the longest time, the reigning theory
for the cause of disease was demonic
theory. Devils and demons were
thought to be the cause of disease.
This theory gave way to punitive
theory, which explained disease as
punishment by the gods for evils
committed.

Figure 1. Progress in theories of disease Next was something a little more

scientific. It was called Miasmatic theory – a theory that you could catch a disease from “bad air” or
even “night air”, i.e., disease was caused by a gas. In addition to this possibly being the cause of a
disease, the theory of the four humors was also employed to diagnose and treat diseases.
Incidentally, this was often applied to mothers suffering from puerperal fever as we discussed in the
first lecture. One of the treatments was to literally bleed the patient of the “bad blood”, which was
not the smartest thing to do to someone already severely weakened by the pathogens invading their
body.
So, as you can see, even superseded or obsolete theories are still called theories.
Novel or Newly Proposed Theories
But in addition to well-established and obsolete theories we also have, of course, novel or newly
proposed theories. These are theories that can not only explain phenomena that are already
explained by well-established theories, but also explain, or at least hope to, anomalies that the
current theory has a hard time supporting. These theories are currently under investigation and being
subjected to testing. Whether they survive the rigors of the scientific method and acceptance within
the wider scientific community remains to be seen.

3
 Lecture 3 HSI1000

3.1.3 In Conclusion
Of course, well-established theories also have to undergo revision when new evidence is uncovered
because science is self-correcting. This is how we end up eventually homing in on the right
explanation for things.
As you can see, the term ‘theory’ is a little more complicated than the term ‘hypothesis’. The term
‘theory’ can apply to well-established and currently accepted explanations, but it can also apply to
contested explanations or even obsolete and wrong explanations. As such it’s always important when
learning or hearing about a theory to clarify the stage of the scientific theory it is at.
Next, we’ll learn about a very common type of explanation used to account for observed phenomena:
causation and its relationship to correlation.

For those interested, here’s a well written article from the New York Times discussing W. Jason
Morgan who developed the Theory of Plate Tectonics. It uses “theory” and “hypothesis” correctly
and very easy to read.

4
 Lecture 3 HSI1000

3.2 Causation and Correlation

3.2.1 Causation
Causation is a very common explanation of phenomena, not just in science, but in everyday life. You
were late for a class because the bus you were meant to catch was very crowded and simply drove
by your stop. You were then forced to catch a second later bus. The effect is you being late for class
and the cause was the first crowded bus not stopping to pick you up. It should go without saying that
cause precedes the effect – always. Which is what makes Figure 2 funny.
Providing a causal explanation may allow us to
understand why such phenomenon exist, but causal
explanations may not be straightforward or simple.
Here’s six reasons why.
3.2.1.1 A combination of causes leading to an effect
Firstly, effects can be the result of a combination of
causes. Let’s go back to the crowded bus example again.
Say the next bus that comes soon after the crowded
one, such that you could catch it and just make it to class
on time. BUT, let’s also say that you drank quite a lot of
water before trying to catch the bus, and now you need
to go to the toilet. If the first bus wasn’t crowded, you
could have caught it, then you could go to the toilet just
before class and still get there on time.
Figure 2. Author unknown. Shared many times over
WhatsApp
HOWEVER, because you were unable to board the first
bus, you catch the second one AND are forced to rush to the toilet before class. You end up being
late. It’s the combination of both the bus being crowded and your need to, err, relieve yourself that
makes you late. In this example, only the combination of causes produces the effect of being late.
3.2.1.2 Cause and effect can refer to groups
Secondly, and on a more serious note, cause and effect can be about groups rather than individual
facts or events. For example, smoking causes lung cancer means that lung cancer occurs more
frequently among those who smoke.
3.2.1.3 More than one cause can result in a specific effect
Thirdly, the same effect can result from separate and distinct causes. Lung cancer can also be caused
by elevated levels of radon – a natural radioactive gas, asbestos exposure and elevated levels of other
chemicals like arsenic, beryllium, cadmium, coal and coke fumes, silica and nickel. It can also be
caused genetically, especially if it is present in the family history.
3.2.1.4 An effect might not result from a given cause in every case
Fourthly, effects need not always be invariably associated with a given causal factor. For example,
although smoking is the leading cause of lung cancer, not all smokers contract lung cancer.
Incidentally, you sometimes hear people say that smoking is fine because my grandfather smoked
like a chimney all his life and he didn’t get lung cancer. That’s like saying charging an enemy machine

5
 Lecture 3 HSI1000

gun picket with 100 soldiers is safe because you know the one guy that happened to make it through
the hail of bullets.
Another example is children drinking fluoridated water have fewer problems with tooth decay
compared to those that don’t. There is a causal link between fluoride and tooth decay. However, it
isn’t true that children who drink fluoridated water will be completely free from tooth decay.
3.2.1.5 Causal explanations can be negative
The previous example brings us to our fifth complication. Sometimes causal explanations can be
negative. For example, fluoridation of water helps prevent tooth decay, or wearing a mask helps
prevent COVID transmission.
3.2.1.5 Causal explanations can involve a series of linked causes and effects
Lastly, causal explanations can involve a series of linked events, like A causing B which in turn causes
C. Terms like proximate and remote are sometimes used to describe the relationship between these
causes. A is a proximate cause of B. Likewise, B is a proximate cause of C. A can be termed as a remote
cause of C. As an example, I’ve embedded a YouTube video illustrating how a butterfly causes a boat
to crash through a roof with this video. After watching the video, you should be able to state the
proximate causes and effects as well as give some examples of a remote cause and effect.

A butterfly causes a boat to crash through the roof of a house: https://youtu.be/u6dAvtbjFKQ

3.2.1 Correlation
Now that we’ve considered cause and effect and some of the subtleties that can arise when thinking
about causes, let’s discuss correlation. Causal explanations are closely related to correlation, but
what exactly is correlation?
Correlation is the degree to which two properties, traits or characteristics move in coordination, or
in sync, with one another. For example, let’s take the height of adult men and shoe size as our two
properties. There is, in fact, a tendency for taller men to have larger shoe sizes. I use ‘tendency’ here
because the relationship isn’t perfect. I can also rephrase this as a correlation between adult male
height and shoe size. While I can’t predict an exact shoe size by only knowing a male adult’s height, I
can give a range of shoes sizes that are likely to fit.
3.2.1.1 Perfect, positive, negative and no correlation
An example of perfect correlation is when there is a direct 1 to 1 correspondence between changes
in the two properties. An example of this can be found in many tree species where the age of the
tree in years is perfectly correlated with the number of rings in the tree’s trunk. The older the tree,
the greater the number of rings – no exceptions. Perfect correlation is rare though. It’s much more
common, if there is a correlation, for there to be only a tendency for the two properties to move in
coordination with one another like the height of adult males and shoe size.

6
 Lecture 3 HSI1000

Such a tendency, or a correlation, can be

seen with fertility rate and child mortality,
things discussed in the last lecture. Recall
that when mortality dropped, and this
was largely due to children not dying as
much, the population surged. This
happened during the industrial revolution
and again after the green, or third
agricultural, revolution. Initially, once the
children start surviving, the fertility rate
remains high for a short time producing a
population surge, but then the fertility
rate falls rapidly. As a result, there is a
Figure 3. From Our World in Data. Scroll to section "Increasing well being correlation between fertility rate and
and status of children".
mortality rate of children. Take a look at
the data for this correlation found in Our World in Data for the year 2019 (Figure 3).
In Figure 3 child mortality is on the horizontal axis, and, total fertility rate, measured as children per
woman, is on the vertical axis. Can you see the tendency for countries with high child mortality to
also have high fertility and vice versa? The solid line drawn through the data on the figure represents
the correlation between these two quantities. Most of the countries are randomly scattered above
and below this line, so it’s not a perfect correlation. If it were, all the countries would lie perfectly on
the line.
Figure 3 is an example of a positive correlation— when one of the quantities goes up, the other
quantity also goes up, or, when one goes down, the other also goes down. Negative correlations can
also exist — when one of the quantities goes up, the other quantity falls. For example, we might
expect that the number of hours spent gaming in a week to be negatively correlated with the number
of hours spent studying. Lots more gaming hours means less time studying – a negative correlation.

Antioxidant Activity vs Phenolic Content in Local Fruits You will also have to consider the strength of
a correlation. Figure 4 is an example of a
strong positive correlation. See how the data
Ferric Reducing Antioxidant Power

isn’t scattered very much away from the line?

This means it’s a strong correlation.
Remember, when it’s a perfect correlation all
the points are on the line. A weak correlation
is when the points are significantly scattered
away from the line. Figure 5 is an example of
a weak negative correlation. It becomes very
hard to even draw a trend line through the
Total Phenolic Content mg/g
Figure 4. Strong positive correlation found in research conducted by
data.
NUS’ Department of FST

7
 Lecture 3 HSI1000

Of course, we can also have no significant correlation

at all. In this case there is be no significant
relationship at between the two properties, and the
plot would look pretty random, like Figure 6.

Figure 5. A weak negative correlation

Some people mistake no correlation with negative
correlation, but they are not the same. No correlation
literally means no relationship, whereas negative
correlation means there is a relationship and that is Figure 6. No correlation found
where one property goes up, the other goes down. For
example, there’s no correlation between a student’s grade and how tall a student is. A student
scoring well could be short or tall or average – it does not matter.
3.2.1.2 Relationship between cause and correlation
When we identify a cause for an effect, we expect there
to be some kind of correlation – positive or negative.
HOWEVER, and this is important, just because two
things are correlated doesn’t mean there is a causal
relationship.
You might be surprised to know that it’s not that hard to
find correlations. Basically, anything that increases as
Figure 7. Correlation between per capita consumption of time goes by will be correlated with something else that
mozzarella cheese and the number of civil engineering increases, or decreases, as time goes by. An example is
PhD's awarded in the U.S.A. Data is reformatted to be
consistent with our presentations of correlations in this given in Figure 7. Look at the strong positive correlation
video. Data from Tyler Vigen’s web site. between per capita consumption of mozzarella cheese
and the number of civil engineering PhD’s awarded in the
US! This is a nonsensical correlation that certainly doesn’t represent any causal relationship. In fact,
there’s a web site dedicated to such silly correlations by Tyler Vigen, linked below. Check it out if
you’re interested.
The bottom line is, for causal
Tyler Vigen’s website: https://www.tylervigen.com/spurious- relationships, like the fertility
correlations rate linked to child mortality
rate, you expect there to be a correlation. BUT, finding a correlation between two properties certainly
doesn’t mean they are causally linked in some way. Oh, and just for the record, while child mortality
is a cause of lower fertility rates, it isn’t the only cause.
In the next section, we’ll look at other ways scientists explain observations.

8
 Lecture 3 HSI1000

3.3 Types of Scientific Explanations

In this section, we’ll be discussing different strategies scientists can adopt to explain some
phenomena. Often, these strategies enhance a cause-and-effect type explanation, leading to a
greater understanding of the phenomenon being studied. For example, explaining that carbon
dioxide emission causes global warming, or cigarette smoking causes lung cancer, or low child
mortality causes low fertility rates are all useful, but we would also really like to know how one causes
the other.

3.3.1 Causal Mechanism

Our first strategy is to provide a causal mechanism. A causal mechanism is a linked chain of causes,
taking us from the remote cause, through a series of proximate causes, eventually leading to the
effect we’re explaining. Sometimes this chain is currently under investigation, so you might hear
scientists say they have established a link, but don’t yet know the mechanism involved. For example,
we know that smoking causes lung cancer, but the exact physiological mechanism of how the
carcinogens in cigarette smoke ends up causing uncontrolled cell growth in the lungs isn’t yet
understood.
Below, I briefly interview Professor Lee who's the next lecturer in this course. In his section of the
course he'll be talking about climate change and global warming and its relationship to scientific
inquiry. I’ve asked to interview him here so he can give you an example of a causal mechanism linking
CO2 emission to global warming.
Prof Bettens: So, what I was wondering, Professor Lee, could you explain to us how we get from the
cause, which is CO2 emissions, to the remote effect of global warming?
Prof Lee: Right, so this is an incredibly complicated system that we need to describe. I'm not going to
describe it in the complexity that would probably be necessary to be accurate, but we can think about
it in a simple mechanism. Certainly, the reason why the Earth is hot is because it receives energy from
the sun. The solar radiation heats up the Earth's surface.
Prof Bettens: So, the sun causes the Earth to warm.
Prof Lee: Yes, the sun causes the Earth to warm. Because the Earth is warm, it is also radiating energy
and it radiates it upwards, back towards space and back towards the atmosphere. That radiation is
going to get absorbed by the atmosphere because there is carbon dioxide in the atmosphere.
Prof Bettens: So, the Earth causes the atmosphere to warm.
Prof Lee: Indeed.
Prof Bettens: Because there’s CO2 in it.
Prof Lee: Yes, so we have a mechanism now where the sun causes the Earth to warm, and the warm
Earth causes the atmosphere to warm. Because the atmosphere is warm it also radiates energy, and
some of that radiation is going to be directed back towards the Earth's surface, and that causes the
Earth's surface temperature to warm even more than it would otherwise be in the absence of the
atmosphere.

9
 Lecture 3 HSI1000

Prof Bettens: So, there you go, that's how CO2 emissions causes the Earth to warm up even more. I
suppose if there was even more CO2 in the atmosphere there would be even more heating up of the
atmosphere and even more heating of the Earth in turn.
Prof Lee: Indeed, yes.
Prof Bettens: OK, thanks.

Figure 8. Highly simplified causal mechanism of how CO2 emissions causes global warming

Providing a causal mechanism clearly enhances our understanding of the phenomenon and may even
provide us with a means of reducing the effect if it is bad, enhancing it if it is good, or protecting it if
the effect is important and must be maintained, etc.
Just knowing a remote cause and its effect is only the first step in any scientific inquiry. Working out
a causal mechanism provides a much clearer picture of what’s going on. But this type of explanation
can still leave us wondering why and how any particular proximate cause results in an effect.

3.3.2 Underlying Process

This brings us to the second type of explanation for a phenomenon, which is to appeal to some
underlying process. In this case we are not looking to say something causes something else to
happen. Here we go down a level and describe the observed phenomenon in terms of a more
fundamental process. It’s a reductionist approach— the description of the observed phenomenon is
reduced to a basic level and then re-described in terms of processes taking place at this more basic
level.
Let’s utilize an example from the textbook to understand this
more clearly. It’s about something called Brownian motion
(Figure 10), which was quite mysterious when first
discovered in 1828 by Robert Brown (Figure 9).
Figure 10 illustrates Brownian motion and in this figure we
have really tiny fluorescent latex spheres in water. Notice
Figure 10. Brownian
how they are all moving about apparently of their own Figure 9. Robert
motion taken from Brown from
Wikimedia: accord, almost as if they were alive. This type of haphazard Wikimedia
Jkrieger am Deutschen
Krebsforschungszentrum in der motion of particles in liquids is called Brownian motion. For
Arbeitsgruppe B040 Biophysik der
Makromoleküle almost 100 years no one had any idea how it could happen
until Albert Einstein explained it in 1905.

10
 Lecture 3 HSI1000

In water, the water molecules are crowded in on each other and

they are all moving about in the solution banging, crashing and
bouncing off of each other. Figure 11 is an example of that sort of
thing. Each sphere represents a water molecule (water isn’t a
sphere, but physically it is nearly that shape). It’s very crowded,
so I’ve had to cut down the
actual physical extent of the
Figure 11. Simulated water with the spheres so that you can see
molecules shown as spheres. Taken from the what’s happening. If I show you
video created by Prof Bettens.
how crowded it really is, it
looks like Figure 12.
If we now put a particle in
with the water, it too is Figure 12. Same as previous simulation
image, but with the spheres their "true" size
going to be crashed into by in the liquid.
all the water molecules surrounding it, making the particle
wobble and randomly “walk” about. Figure 13 shows a 2D
simulation of that happening for 5 particles (highlighted in
yellow) that collides with a set of 800 particles. The 5 yellow
particles leave a blue trail. The motion the yellow particles are
undergoing is the Brownian motion. Note that in this example
the explanation of the phenomenon literally involves the
description of the underlying process and not a series of linked
Figure 13. Simulation of Brownian motion from
Wikimedia: Lookang Author of computer model Francisco causes and effects.
Esquembre, Fu-Kwun and lookang (this remix version)

3.3.3 Laws
Some phenomena are explained through scientific laws. These are just generalized descriptions of
regularities that have been found to occur in nature. For example, what happens if the volume of a
container of gas is reduced while its temperature remains constant? The pressure increases. Why?
Robert Boyle (Figure 15), in 1662, showed,
through experiments on air, that the pressure of
a gas varied in inverse proportion to its volume,
provided the temperature remained fixed
(Figure 14). This is known as Boyle’s law. So, as
we reduce the volume, the pressure increases in
inverse proportion. The explanatory power of
laws like these comes from their ability to reveal
how particular events are instances of generally Figure 14. Robert Boyle's original
data from Wikipedia: Krishnavedala
Figure 15. Robert Boyle from understood regularities in nature.
Wikipedia.
This law, like others, tends to be thought of as universal, so we expect Boyle’s law to hold for all
gases, although we know now that gases under extreme conditions can deviate significantly from this
law. As another example, we mentioned Newton’s laws of motion and law of gravity in the first
lecture – again these are expected to hold universally.

11
 Lecture 3 HSI1000

These types of laws will be familiar to students of science. In the humanities, there are other types
of laws that exist which are statistical in nature. For example, the law of supply and demand is quite
fundamental to economics, or the law of reciprocity in psychology. There are many such laws, and
again they are rooted in how particular events are instances of regularities that are observed to occur.
So you can see what I mean by a law being “statistical in nature”, let’s consider the law of reciprocity
which says (in simple terms); when someone does something nice for you, you will have a deep-
rooted psychological urge to do something nice in return. In fact, you may even reciprocate with a
gesture far more generous than the original good deed. This type of law is statistical in nature,
meaning that you don’t expect it to be obeyed in every case, but there is a tendency for it to occur
more often than not. Such laws can also be used to explain human behavior, just like Boyle’s law can
be used to explain the behavior of gases.

3.3.4 Function
The final strategy scientists use to explain a phenomenon is function, that is, to explain something
based on the purpose it fulfils. Why am I wearing this black shirt? Because it covers my torso as
required by social norms. The black also gives a good contrast to the background in the lecture video
and doesn’t make me look so, err, large shall we say. I explained my wearing of a black shirt based
on the function it performs.
Why do we have a heart? Because it is the organ needed to pump blood around the body. Why does
a particular trait dominate in a certain species? Because that trait confers a definite evolutionary
advantage to members of the species with that trait. These are examples of function being used to
explain why something is.
It is also used to explain human behavior, like why someone might
be eating rice porridge (Figure 16) for lunch. Maybe the person is
not feeling well. Or one country went to war with another because
the first lacked a necessary resource the second had but refused to
trade. Functional explanations are often used, not just in the
physical, biological and medical sciences, but in the humanities and
social sciences as well.

3.3.5 Explanatory strategies can be interrelated

All these ways of explaining things tell us how and why something
Figure 16. Congee taken from Wikimedia: is the case, but these explanations often lead to more questions
Daiju Azuma
requiring explanations. Why is the sun hot in the first place? How
does the sun heat the Earth? Why does sunlight not significantly heat the atmosphere, but the warm
Earth does?
Consider Boyle’s law for a moment. Why do gases follow this inverse proportion law? Providing an
explanation requires us to go deeper, and we realize that our list of strategies for providing
explanations are interdependent. As an example, let’s consider the case of Boyle’s law. We can
explain further by considering the following.
Gases are made of huge numbers of molecules, or atoms, depending on the type of gas, and they are
moving about in a haphazard manner in the container they are in. They strike the walls and bounce

12
 Lecture 3 HSI1000

off. The huge number of collisions occurring with the walls each second produces a constant
outwards force on the walls— this is pressure. By reducing the volume of the container, you increase
the frequency of the collisions with the walls, thus increasing the outwards force on the walls, that
is, the pressure goes up.
This explanation for Boyle’s law is a call to the underlying process of this behavior of molecules and
atoms. It’s a very crude description of the kinetic theory of gases – a theory that explains the behavior
of all gases and their corresponding laws and can even be used to help understand the behavior of
liquids.
We can see a definite interdependence between the different ways of explaining phenomena. The
type of strategy adopted will depend on the questions we ask and the level of detail needed in the
explanation. In science the questions seem never ending, but each time we explain something we
achieve a deeper and fuller understanding of nature and the universe around us.
We’ll finish up this section on the types of scientific explanations, all of which are testable, and
consider in the next section how we might go about choosing competing explanations for the same
phenomenon.

13
 Lecture 3 HSI1000

3.4 Occam’s Razor

Often there can be more than one explanation for a phenomenon. If that’s the case, then on what
basis do we have to choose one over another? We could devise tests of each explanation, which we
know is part of our scientific method, and eliminate any explanation that fails the test. But this may
require considerable effort and expense, when it may not even be necessary. What we can first do
when sorting through possible explanations is to apply Occam’s Razor.
Occam’s Razor is named after William of Ockham (Figure 17), an English
monk, philosopher, and theologian believed to be born on Ockham in 1285.
He has been credited with the Occam’s Razor, a methodological principle,
that we still apply today. It can be used to help figure out which explanation
is most likely to be correct.
The version of Occam’s Razor we shall use can be found in our textbook in
Chapter 3, page 43. It states, “Given competing explanations, any of which
would, if true, explain a given puzzle, we should initially opt for the
explanation which itself contains the least number of puzzling notions.”. The
Figure 17. Willian of Ockham rationale behind making this choice is that if a puzzle can be explained
from Wikimedia: self-created without introducing any additional puzzling notions, then there’s no good
(Moscarlop), CC BY-SA 3.0
<https://creativecommons.org/licenses/by-
sa/3.0>, via Wikimedia Commons
reason to introduce the aforementioned additional notions.
Another way of stating Occam’s Razor might be to choose that explanation which is least complex
and/or most plausible. Again, why add additional complexity to an explanation when a simpler
explanation will suffice? The greater the complexity, we might expect, the greater than chance that
something is wrong.
We have already seen an example we
could’ve applied Occam’s razor to.
Recall in lecture 1, during the
scientific revolution, Copernicus
came up with a model of the known
solar system (Figure 18) that at the
time was considerably simpler than
the currently accepted one by
Figure 18. Copernicus' Sun-Earth- Ptolemy (Figure 19).
Mars System
Both explanations for the motions of Figure 19. Epicycles for Mars about the Earth
the planets equally explained their locations throughout the from Wikipedia
year. So which explanation is more likely to be correct? Applying Occam’s razor, we would have to
cut away Ptolemy’s model and focused our efforts on trying to test Copernicus’ model – which Galileo
did.
Of course, there’s no guarantee that the simplest explanation or the explanation with the least
number of puzzling notions, is the correct explanation. But it is wise to first consider the simplest and
least puzzling explanation until we can prove it wrong. Only then are we forced to move onto more
complicated explanations.

14
 Lecture 3 HSI1000

For example, you come home late one evening, dump your Ez-link card on the table then crash on
your bed. The next morning you wake to find your Ez-link card isn’t on the table where you’re pretty
sure you left it. Where could it be? Is it on the floor? Maybe you didn’t leave it on the table, and put
it someplace else? Maybe another family member took it? Maybe someone crept into your flat while
everyone was sleeping and stole just your Ez-link card? Perhaps it simply disappeared?
Occam’s razor would say first check under the table – it’s the simplest and least puzzling explanation,
but if it’s not there, then you have a mystery, and we’re forced to consider the more complex
explanations. Depending on how certain you were about leaving it on the table, you could either just
ask your family members if they took it or start looking for it in plausible locations. Someone creeping
into your house to steal just your card seems pretty unlikely (why steal only the card?) and there’s
no evidence of a break in. The card disappearing into thin air is just not possible. Applying the BDTK
here helps us by using the last tool "Does the claimant use flawed reasoning?" Matter can neither be
created nor destroyed. Objects simply do not disappear.
I’ll finish up this section with an example of a conversation borrowed from Carl Sagan. It helps
illustrate the use of Occam’s razor and highlights the idea of an explanation being testable, which I
discussed in the first video of this lecture. I’ll be asking Prof Lee again, the next lecturer in the course,
to help me out.

Conversation with Prof Lee

Prof Bettens: OK I have here with me Professor Lee who's the next lecturer in this course after my
part of the course, so I'm going to ask him a few questions because he has a rather unusual claim,
and that claim is that he has a dragon in his flat. Now that's a pretty wild claim because I don't think
any dragons been seen at all, ever, but anyway let's just accept that that might be true. [Prof Bettens
turns to Prof lee and asks] So can I come over Adrian and see this dragon of yours?
Prof Lee: Yeah, you can definitely come over. Of course, you can come over.
Prof Bettens: And can I see it?
Prof Lee: Ahh, no. No you can't see it; it's an invisible dragon. I told you it was an invisible dragon.
Prof Bettens: OK, right, right. Yeah, I forgot. OK, so it's an invisible dragon. Yes, it's invisible so I can't
see this dragon?
Prof Lee: That’s right. You can’t see it.
Prof Bettens: But I surely can feel the dragon, right?
Prof Lee: If only. It's an ethereal, invisible dragon.
Prof Bettens: Ethereal, what does that mean actually?
Prof Lee: You can't touch. It's not corporeal.
Prof Bettens: I see, so my hand would pass right through this dragon, right. So, it's invisible and it's
ethereal. But it's a dragon though, it'll probably make quite a lot of noise I expect.
Prof Lee: No, no, no. This is specially trained ethereal, invisible dragon. This is a ninja, ethereal
invisible, dragon.
15
 Lecture 3 HSI1000

Prof Bettens: Ha! A ninja dragon. Really?

Prof Lee: Completely silent, really, yeah, yeah.
Prof Bettens: OK, so we've got an invisible, ethereal, ninja dragon. OK let me think what else I can do.
OK, it surely walks on the floor, so I could put some flour on the floor, right, you should be able to
see the footprints of the dragon.
Prof Lee: Well, apart from the fact that it is ethereal, it's also flying. It always flies, it never lands on
the ground.
Prof Bettens: OK this is starting to sound a bit funny, but what about the heat? The thing gives off
heat, so if I had an infrared camera I should surely be able to see the thing, right? It should be able
to be detected with an infrared camera.
Prof Lee: No, no. The IR camera won’t detect it because its breath is heatless.
Prof Bettens: It's heatless?
Prof Lee: Yeah, Yeah. Now it's an ethereal, invisible, ninja, always flying, heatless dragon – it's very
special.
Prof Bettens: Yes, yes. It is very special, and how do you know it's there actually?
Prof Lee: You trust me? Why don't you believe me?
Prof Bettens: Uh, because I can't think of any test that could possibly… Well maybe I should just simply
believe you.
Prof Lee: Yeah. I think so. We've had a friendship for a long time. Why wouldn't you believe me?
Prof Bettens: Yeah, yeah, well at least that's true.

Summary of Conversation with Prof Lee

Do you see the issue here? The claim that there's a dragon in his flat is clearly untestable. Every
experimental test that I come up with, he has a reason for why it won't work. I'm sure if I continued,
he would have an excuse as to why I can't test his claim any further.
So, we have a situation where we can either simply believe Prof Lee that he has an invisible, ethereal,
ninja, flying, heatless dragon in his flat, or we can apply Occam's razor and say it's all in his head. I
admit that's one puzzling notion because apart from this one thing I know Prof Lee to be quite a
reasonable and logical person, but at least it's only one puzzling notion. To accept his claim, I would
have to require a whole bunch of puzzling notions to be true. Many of which simply don't match with
our idea of the way the world works.
That's it for this section on deciding how we can choose between competing explanations for puzzling
phenomena. In the next section we'll be discussing scientific models. This topic isn't covered in our
textbook, but we actually think it's important enough that it warrants its own section.

16
 Lecture 3 HSI1000

3.5 Scientific Models

Models are under the topic of “explanations” because very often this is exactly what models are
about – and we’ll see shortly where models fit into the various explanatory strategies we mentioned
earlier this lecture. However, models aren’t just about their ability to explain phenomena. They can
also be used to test those explanations, make predictions and projections, and enhance our
understanding of nature and even aid in the research of figuring out what is going on. That is, they
can be a tool used to better analyze what’s occurring in the real world.
So, what is a scientific model? A scientific model is a cut-down and simplified representation of real-
world objects, systems or events. They are idealizations of reality, with the extraneous and hopefully
irrelevant parts of reality ignored, or treated very crudely or simply. Scientific models can be broadly
classified into three categories: Physical models, mathematical and/or computer models, and
conceptual models. To get a clear idea of what these are, I’ll talk about each of them now with
examples.

3.5.1 Physical Models

These are actual physical objects representing some aspect of nature. For
example, a globe – a model of planet Earth (Figure 20), or some other planet
or star, is a physical model. It can be used for all sorts of things, like helping
you understand why it’s a different time of the day at different places on
Earth, or why in Summer, north of the Arctic circle, the sun doesn’t set, and
in winter it never rises. It is a tool that assists in understanding reality.
To understand why it can be daytime in Singapore, and nighttime in New York
at the same time, you could refer to a globe and describe the underlying
process of how the sun and the rotating earth system leads to different times
of day for the same moment in time.
Figure 20. A globe, from
Wikimedia: John Phelan, CC BY-SA
4.0
But physical models, like all models,
<https://creativecommons.org/licenses/by-
can also be used to help analyze and
sa/4.0>, via Wikimedia Commons

study the real system they are meant to represent. There are
many such examples, like architectural models of buildings can
help with visualization of internal relationships within the
structure or external relationships of the structure to the
environment. Figure 21 is a picture of a scale model of
Singapore’s Marina Bay area from 2006 exhibited at the URA
Gallery Museum. Figure 21. Physical model of the Marina Bay
area taken from Wikimedia: No machine-readable author
To assist in better understanding reality, the physical model provided. Calvin Teo assumed (based on copyright claims)., CC BY-SA 3.0
<http://creativecommons.org/licenses/by-sa/3.0/>, via Wikimedia
can be augmented with instruments to make measurements Commons

of what’s happening in and around the model. Such measurements can assist in optimization and the
design of equipment or processes. For example, the instruments may measure the external flow of
air or water around model buildings, vehicles, people, or hydraulic structures. The physical models
can be placed inside a wind tunnel (Figure 22), like the one on campus, or water tunnel (Figure 23)
or a wave tank (Figure 24) for the testing and making of measurements. Internal flows can also be

17
 Lecture 3 HSI1000

studied using physical models with instruments. For example, in the design of ductwork systems,
pollution control equipment, food processing machines, and mixing vessels.

Figure 22. Wind tunnel from Wikimedia: Figure 23. Water tunnel from Figure 24. Wave tank from Wikimedia:
NASA Ames Research Center, Public domain, via Wikimedia Wikimedia: U.S. Navy, Public domain, via Wikimedia Jplourde umaine, CC BY-SA 4.0
Commons <https://creativecommons.org/licenses/by-sa/4.0>, via
Commons
Wikimedia Commons

Of key importance in all these models is getting the scale

right, because the forces and physical laws that operate
in the larger real-world environment don’t simply
reduce and apply equally on a smaller scale.
Nevertheless, when done properly (Figure 25), physical
models can simulate complex air and water flow to a
degree of accuracy that is not possible with other types
of models.
Physical models need
not only be smaller than
Figure 25. Towing tank from Wikipedia: Xtrememachineuk at
English Wikipedia, CC BY-SA 3.0 <http://creativecommons.org/licenses/by-sa/3.0/>, the thing they
via Wikimedia Commons
represent. They can
also be larger. For example, chemists often use different types of
physical models to better understand the relationships between
atoms, or groups of atoms in a molecule, or to assist in
understanding chemical reactions. The discovery of the structure of
DNA by Watson and Crick was greatly faciltated by physical models
of the molecules involved, and in Figure 26 we see them pictured Figure 26. Watson and Crick with DNA
next to a physical model of DNA. physical model. Picture in the public
domain.
So, physical models are actual physical objects that assist scientists
and engineers to better understand the real world in several ways.

3.5.2 Conceptual Models

Conceptual models are cut down versions of
reality with only the parts of interest included.
A map is one example. It’s a conceptual
representation of real-world physical locations
and relevant objects. If someone asks you for
directions from, say, the faculty of science to
Clementi MRT, you might sketch them out a
Figure 27. Crude map for getting from Frontier to Clementi MRT map (Figure 27). You wouldn’t include all the

18
 Lecture 3 HSI1000

things on the map between FOS and the MRT, just the relevant
features needed for someone to find their way there. It conveys
the necessary concepts and ideas to navigate in the real world.
Diagrams and figures representing ideas and concepts in science
can be thought of as conceptual models. A diagram used to help
explain and even analyze Archimedes’ Principle (Figure 28) is an
Figure 28. Archimedes principle from Wikimedia:
example of a conceptual model. It doesn’t necessarily directly
MikeRun, CC BY-SA 4.0 <https://creativecommons.org/licenses/by-
sa/4.0>, via Wikimedia Commons
represent what might be happening in a lab – some
extraneous details are left out. But it does contain all the
important information needed to understand what’s going on.
A diagram of a block sitting on an inclined plane, with all the forces indicated (Figure 29), is a
conceptual model of a real block (Figure 30) sitting on a slope. Again, there’s some things missing
from the diagram, and the objects and surfaces might not even be the same as the diagram. Yet, it
contains what’s needed to describe what is happening, even if the block is not moving.

Figure 30 Real-world block on inclined plane.

Figure 29 Block on inclined plane from Wikimedia: Jer S.

Conceptual models can be quite abstract, and not physically look

at all like the real-world objects they may represent. For example,
Lewis’s chemical structures of molecules are conceptual models
(Figure 31). Chemists use them to explain all kids of phenomena
including why some chemical reactions can occur, and others not,
Figure 31. Lewis structure of dopamine from
Wikimedia
or even use them to make predictions of what chemical
transformations could take place.

19
 Lecture 3 HSI1000

A chemist knows there’s no lines

between atoms in the real-world
molecules (Figure 32), and that the
structures themselves bear no
Figure 32. Lewis structure of
water from Wikimedia: Dan Craggs resemblance at all to the
macroscopic material they
represent (Figure 33). Yet for their intended purpose, these
chemical structures are critical for understanding, interpreting Figure 33. Real drop of water from Wikimedia:
José Manuel Suárez, CC BY 2.0
and predicting a lot of chemistry. <https://creativecommons.org/licenses/by/2.0>, via Wikimedia
Commons

Conceptual models of reality might appear quite similar to real world objects, but they could also be
diagrams illustrating processes, even one as abstract as a circuit diagram. All conceptual models
depict ideas and concepts within science and can be used to explain phenomena and make
predictions in the real world, just as physical and mathematical/computer models can.

3.5.3 Mathematical/Computer Models

We group math and computer models together because very often the math involved in the models
is coded up on a computer. This is done so that the computer can do all the tedious calculations
necessary to apply the mathematics. Indeed, sometimes the mathematics is so complex that only a
computer can solve it.
So, what are these types of models anyway? It’s when math can be used to describe nature in some
way. We’ve actually already met some mathematical models. When we discussed Boyle’s law earlier
in this lecture, we could have expressed this law in a math formula. Pressure of a gas varies in inverse
1
proportion to its volume when the temperature is kept constant, i.e., 𝑝 ∝ 𝑉, or put another way,
𝐴
pressure equals a constant divided by volume, i.e., 𝑝 = . These math formulae models how the
𝑉
real-world pressure and volume of all gas varies when the temperature is fixed.
Recall in lecture two, we saw the population of
the world plotted against time. The curves drawn
to connect the population estimates given by
different researchers on the Our World in Data
site were drawn using simple straight lines, which
are math functions. These math functions
(𝑦 = 𝑚𝑥 + 𝑐) were used to model how the world
population varied between the estimated
population points given by the researchers. Here
math was used to model how the world
Figure 34. Plot of world population from 8000 BCE to 1 CE from population varied over time by filling in the gaps
Our World in Data between population estimates.
When we looked at the correlation between two properties (Figure 5, for example), and we drew a
trend line through the data, we were mathematically modelling the ideal relationship between the
two properties. We learnt only when the correlation is perfect will the points lie perfectly on the line,

20
 Lecture 3 HSI1000

but the correlation may not be perfect in the real-world, so our line, which is a math formula, only
represents a trend.
Many laws in science can be expressed using mathematical formula. These formulae can then be
used to derive even more expressions via the rules of mathematics. Taken together the equations
represent how nature behaves, from the motion of large objects, to the tiniest atom and even sub-
atomic particles, to how fluids flow around objects, or current passes through circuits, or heat is
exchanged between parts of machinery or the ocean and atmosphere, or how waves propagate
through space and time, or diffract around objects, and even how fast chemical reactions are
occurring in a flame or even the biochemistry taking place in organisms. There are just far too many
applications of mathematics to represent or list here.
Nevertheless, when a scientist wants to model something, say the climate, he or she will need to
consider all the necessary laws relevant to the phenomenon in question. This can involve a lot of data
and equations, including chemical reactions. Each of the equations can all be interrelated, making
these models inherently quite complex. When all these equations and data are taken together, they
then constitute a mathematical model. To solve such a complicated model the equations will often
need to be programmed up on a computer for them to “number crunch” the data with the equations
and produce meaningful output.

3.5.4 Predictions versus Projections

Now this output could be a prediction of the future, given current data, or it could be a projection –
a “what if” type scenario. Projections differ from pure predictions in that they utilize “what if”
scenarios. For example, we could use a COVID-19 forecast model to project the number of deaths
that is likely to occur if no safe management measures are taken, or use a climate model to project
what will likely happen if no action is taken regarding carbon emissions.

A write up on the difference between predictions and projects and “The Truth about Scientific
Models” was given in Scientific American recently. Here’s the link if you’re interested:
https://www.scientificamerican.com/article/the-truth-about-scientific-models/
By now you probably can appreciate that each of these model types, physical, conceptual and
math/computer models can be interrelated. For example, the physical model with instruments
attached could have the data read from the instruments and fed into a computer model. It would
then evaluate the performance of the object being modeled and predict how it could behave in the
real world.
A conceptual model could help to establish what equations should be used in a computer model
meant to mimic the real-world thing the conceptual model represented. You’ll get a better idea of
this in the second workshop.

21
 Lecture 3 HSI1000

Of course, we should never lose sight of the fact that models, being cut-down and simplified
representations of reality, can actually leave too much of the real-world out and end up producing
the wrong results. However, because science is self-correcting, once such an error is discovered, the
ignored effect can then be included in the model to make an improved version.

3.5.5 Models versus Theories

Models can be created from the concepts and principles provided in a theory or a hypothesis. As
such, these models are concrete applications of the principles or concepts outlined by the theory or
hypothesis. Because of this, the results of these models can actually be used to test a hypothesis.
That is, the model can make a prediction of what should happen if a hypothesis is correct. We can
then check and see if it really does happen in the real-world, and if it doesn’t, and the model is
accurate, then the hypothesis is wrong.
Because models are constructed from the concepts and principles provided in a theory, and therefore
represent a concrete and specific application of a theory, we can think of them as being subordinate
to the higher-level theory which just expresses the concepts and principles. For example, Newton’s
statement of the three laws of motion and the law of gravitation can be used to construct a computer
model of our entire solar system, i.e., the Sun, planets and all their moons, to predict where the Earth
and all of these heavenly bodies will be moving into the future. The model is a specific concrete
application of the higher-level theory of Newton for the motion of objects with mass and how they
attract each other.
Despite this type of subordination of a model to a theory, you still often hear scientists
interchangeably use the terms “theory” and “model”. Like big-bang theory or big-bang model.
Nevertheless, from the context of their discussion and if you have understood this section, you should
be able to decide for yourself what they mean.
With that, we conclude this lecture on scientific explanations and chapter 3 of our textbook. We shall
be moving onto chapter 4 and a little of chapter 5 in the next lecture where we will consider the
testing step of the scientific method.

22