How Scientific Thinking Works: From Observation to Explanation

“Science does not begin when we know the answer. It begins when we learn to distinguish what we observed from what we think explains it.”

— Tymur Levitin

A glass falls from a table.

A plant grows differently near a window.

A metal object becomes warmer.

A population changes.

A medicine appears to help.

A student notices a pattern.

Something happened.

That is where scientific thinking can begin.

But noticing something is not the same as explaining it.

And an explanation that sounds reasonable is not automatically a scientific explanation.

Between observation and knowledge lies a system of questions:

What actually happened?

What do we know?

What are we assuming?

What could explain the observation?

What else could explain it?

What evidence would distinguish between those explanations?

What does our model predict?

What would make us revise it?

Science is therefore much more than a collection of facts about physics, chemistry or biology.

It is also a way of managing uncertainty.

Knowing Science and Thinking Scientifically Are Different

A student may know:

the planets;

the parts of a cell;

Newton's laws;

chemical symbols;

the definition of photosynthesis.

All of this matters.

Scientific thinking requires knowledge.

But knowledge alone does not tell us whether the learner can investigate a new situation.

Suppose we ask:

Why did this happen?

Now another level begins.

The learner needs to connect evidence, mechanisms, models and possible explanations.

This is similar to the distinction developed in our Four-Level Learning Model:

Knowledge → Understanding → Ability → Independence

Scientific knowledge is essential.

Scientific reasoning asks what the learner can do with that knowledge when the answer is not already supplied.

Observation Is Not Explanation

Imagine two plants.

One grows faster than the other.

We observe:

Plant A grew 4 centimetres more than Plant B during the same period.

That is an observation.

Now someone says:

Plant A grew faster because it received more sunlight.

That is an explanation.

Perhaps it is correct.

But notice what happened.

We moved from:

what we observed

to:

what we believe caused it.

That movement needs evidence.

Maybe Plant A also received more water.

Maybe the soil was different.

Maybe the plants were genetically different.

Maybe temperature differed.

Maybe the measurement was inaccurate.

Scientific thinking requires us to keep observation and explanation conceptually separate long enough to test the connection between them.

The Scientific Reasoning Cycle

A useful architecture is:

Observe → Question → Hypothesize → Model → Predict → Test → Evaluate → Revise → Explain

I call this the Scientific Reasoning Cycle.

It is not a rigid recipe for every scientific discipline.

Real research is considerably more complicated.

But as an educational model, it shows something fundamental:

scientific knowledge does not emerge from one intellectual operation.

It develops through a sequence of connected decisions.

1. OBSERVE

What actually happened?

Observation sounds passive.

It is not.

What should we measure?

At what scale?

With what instrument?

Under which conditions?

Which details matter?

Which can be ignored?

Two people can look at the same event and notice different things because observation is already influenced by questions and prior knowledge.

Good observation therefore tries to separate:

what the evidence shows

from

what we expect it to show.

2. QUESTION

What exactly are we trying to understand?

“Why does this happen?” may be a useful beginning.

But scientific investigation often requires a more precise question.

Instead of:

Why do plants grow differently?

we might ask:

How does the amount of light affect the growth rate of this plant species under otherwise similar conditions?

The question defines the problem.

A poorly defined question can produce a great deal of data without producing much understanding.

This principle is not unique to science.

In How to Solve a Problem You've Never Seen Before, the first stage of independent problem solving is Orient.

Before solving a problem, we need to know what problem we are actually solving.

3. HYPOTHESIZE

What could explain the observation?

A hypothesis is not simply a guess.

It is a proposed explanation or relationship that can guide further investigation.

For example:

Increasing light exposure will increase growth rate within a certain range.

Now we have something that can interact with evidence.

But one hypothesis is rarely enough.

Strong reasoning also asks:

What alternative explanations exist?

This prevents the first plausible story from becoming the automatic conclusion.

4. MODEL

What part of reality are we representing?

Science uses models constantly.

An atom is represented through models.

Climate is modelled.

Populations are modelled.

Forces are represented through diagrams.

Biological systems are represented through networks and processes.

A model is not reality itself.

It is a structured representation that makes some features easier to reason about.

Every model therefore has limits.

This is one of the most important scientific ideas a learner can understand:

A useful model does not have to contain everything. It has to represent the relevant relationships well enough for the task.

5. PREDICT

If our explanation is correct, what should happen?

Prediction connects an idea to possible evidence.

If greater light exposure causes increased growth under the conditions we are studying, then changing light exposure should produce a measurable consequence.

Prediction forces an explanation to take a risk.

A vague explanation can survive almost anything.

A useful scientific model should tell us what we would expect to observe under particular conditions.

6. TEST

What evidence can challenge the prediction?

Testing is often associated with laboratory experiments.

Experiments are extremely important.

But scientific testing can also involve:

observational data;

field studies;

historical evidence;

computer simulations;

comparative analysis;

measurements;

existing datasets.

The method depends on the question.

The central principle remains:

we expose an explanation to evidence that could potentially show us that something is wrong or incomplete.

7. EVALUATE

What does the evidence actually support?

Suppose the result matches the prediction.

Have we proven the explanation absolutely?

Usually not.

Perhaps another explanation predicts the same result.

Perhaps the sample was too small.

Perhaps a variable was uncontrolled.

Perhaps measurement error matters.

Perhaps the effect is real but the proposed mechanism is wrong.

Scientific evaluation therefore asks not simply:

Was I right?

but:

How strongly does this evidence support this explanation compared with the alternatives?

8. REVISE

What should change in our explanation?

Revision is not scientific embarrassment.

It is part of scientific work.

A model may need:

a new variable;

a narrower claim;

a different mechanism;

better measurements;

another experiment.

Sometimes an explanation survives.

Sometimes it changes.

Sometimes it is abandoned.

The ability to revise in response to evidence is one of the differences between defending an opinion and investigating a question.

9. EXPLAIN

Can we build the best current account of what happened?

Scientific explanation integrates:

observations;

relationships;

evidence;

mechanisms;

models;

limitations.

A strong explanation should make clear not only what we think, but also why the available evidence gives us reason to think it.

And ideally:

what remains uncertain.

The Scientific Reasoning Cycle

The complete model is:

OBSERVE → QUESTION → HYPOTHESIZE → MODEL → PREDICT → TEST → EVALUATE → REVISE → EXPLAIN

Notice something important.

The cycle does not end with:

PROVE.

It ends with an explanation that is supported to some degree by evidence and remains open to refinement when better evidence appears.

That intellectual discipline matters far beyond school science.

Evidence Is Not the Same as Proof

In everyday speech, people often say:

This proves it.

Science frequently operates with more careful language.

Evidence may:

support;

contradict;

increase confidence;

decrease confidence;

rule out some explanations;

leave several possibilities open.

Different fields also have different standards of evidence.

A mathematical proof and experimental evidence in biology are not the same epistemic object.

This distinction matters because the word science covers disciplines that investigate reality in different ways.

Correlation Is an Observation About Relationship

Suppose two variables change together.

That can be important evidence.

But correlation does not automatically identify the causal mechanism.

Perhaps A influences B.

Perhaps B influences A.

Perhaps C influences both.

Perhaps the relationship is coincidental.

The slogan correlation is not causation is useful, but incomplete.

Correlation can be evidence.

What it cannot do by itself is settle every causal question.

Scientific thinking asks what additional evidence would help distinguish between competing explanations.

A Scientific Fact Is Not “Just a Theory”

Everyday language can create confusion here.

People sometimes use theory to mean:

an unsupported idea.

Scientific usage can be very different.

A scientific theory may be a highly developed explanatory framework supported by substantial evidence.

Similarly, a model, hypothesis, law, observation and theory do not simply form a ladder where one eventually turns into another.

They perform different roles.

Learning science therefore includes learning the conceptual language through which scientific knowledge is organized.

Models Are Powerful Because They Are Incomplete

Consider a map.

A useful map does not contain every tree, stone and window.

If it did, it might become useless.

It selects.

Scientific models do the same.

A model of motion may ignore air resistance.

An economic model may hold some variables constant.

A biological model may simplify a complex system.

The question is not:

Is the model identical to reality?

It cannot be.

The useful questions are:

What does it represent?

What does it omit?

Under what conditions is it useful?

Where does it stop working well?

Understanding models protects learners from two opposite mistakes:

treating models as literal reality;

or rejecting them because they simplify reality.

Physics Makes Models Visible

Physics is a particularly clear example.

A student sees a moving object.

To solve a problem, they may need to represent:

forces;

velocity;

acceleration;

energy;

momentum.

The physical situation becomes a model.

Then mathematics operates on the model.

A formula is not magic.

It belongs to a representation of the system.

This is why learning physics cannot be reduced to memorizing equations.

The learner needs to understand which model applies and why.

Our existing Online Physics Tutoring for School, University, and International Education provides the practical learning route; this scientific reasoning framework supplies the conceptual parent above it.

Biology Requires Thinking in Systems

Biological systems contain interactions.

Cells.

Organs.

Organisms.

Populations.

Ecosystems.

Changing one component may affect several others.

Simple linear explanations can therefore be insufficient.

Scientific thinking in biology often requires questions such as:

What level of organization are we examining?

Which mechanism could produce this effect?

What variables interact?

How does the system regulate itself?

What evidence distinguishes adaptation from coincidence?

Facts remain essential.

But systems thinking connects them.

Chemistry Connects the Invisible With the Observable

Chemistry creates another reasoning challenge.

We observe:

colour changes;

temperature changes;

precipitates;

gas formation;

measurable concentrations.

Then we use models involving particles, bonds, structures and reactions to explain those observations.

Learners therefore move constantly between:

macroscopic observation

and

microscopic model.

Confusing the two can create major conceptual problems.

Again, representation matters.

Science and Mathematics Are Connected but Not Identical

Mathematics provides powerful tools for science.

Measurement.

Modelling.

Statistics.

Equations.

Probability.

But mathematical validity alone does not tell us whether a model describes the physical world accurately.

A mathematical structure may be internally consistent while its scientific application depends on empirical evidence.

This distinction helps learners understand why mathematics and science work together so closely without being the same discipline.

For the mathematical side of the system, see Understanding Mathematics: How Mathematical Thinking Develops.

Scientific Thinking Is Also Language-Dependent

Science is conducted through concepts, symbols and language.

A learner may understand a physical phenomenon but struggle to explain it in English.

Another may know the English terminology without understanding the phenomenon.

These are different educational problems.

Consider:

force

energy

power

work

In everyday English, these words have broad meanings.

In physics, they participate in more precise conceptual systems.

Learning scientific English therefore cannot be reduced to translating terminology.

The learner needs to connect:

word → scientific concept → relationship → application.

Learning Science Through Another Language

This becomes especially important for students studying abroad or in international education.

A Ukrainian student may understand physics but now need to study it in German.

A Spanish-speaking student may enter an English-language biology program.

A learner may use English documentation to study programming or scientific subjects.

The question becomes:

Is the difficulty scientific, linguistic, or both?

Our framework Learn a Subject Through Another Language: When Language Becomes a Tool for Knowledge addresses precisely this interaction.

Existing knowledge can become a bridge into the new language.

But only if we diagnose correctly what the learner already understands.

Why Explaining Science Develops Understanding

Ask a student to explain:

Why does this happen?

Not merely:

What is the definition?

Explanation exposes the learner's model.

They may know every required term and still connect them incorrectly.

Or they may understand the mechanism but lack the terminology needed to express it precisely.

Both become visible when the learner must construct an explanation.

This connects scientific education directly with our broader distinction between knowledge and understanding.

What Makes a Scientific Question Good?

Not every question is equally investigable.

A useful scientific question usually needs enough precision that evidence can meaningfully interact with it.

Compare:

Why is nature beautiful?

with:

How does temperature affect the rate of this reaction under these conditions?

The first may be philosophically valuable.

The second allows a specific empirical investigation.

Science does not answer every meaningful human question.

Its strength comes partly from being disciplined about which kinds of questions its methods can address.

Scientific Thinking Does Not Mean Distrusting Everything

Critical thinking is sometimes misunderstood as permanent skepticism.

Reject every claim.

Trust nothing.

Demand impossible certainty.

That is not useful scientific reasoning.

Evidence has different strengths.

Sources have different reliability.

Claims deserve different levels of confidence.

Scientific thinking means calibrating belief to evidence.

Sometimes the rational conclusion is:

We know this extremely well.

Sometimes:

This is likely.

Sometimes:

The evidence is mixed.

Sometimes:

We do not know yet.

Those are not weaknesses.

They are different states of knowledge.

“I Don't Know” Can Be a Scientific Answer

There is enormous intellectual value in being able to say:

I don't know.

But scientific thinking adds:

What would we need to know?

That transforms ignorance into a research program.

What should we measure?

What evidence is missing?

Which explanations remain possible?

Which experiment could distinguish them?

This is the same transition we developed in our independent problem-solving framework:

from

I don't know, therefore I stop

I don't know yet, therefore I investigate.

Scientific Thinking and Independent Thinking Meet Here

Our Independent Problem-Solving Cycle asks learners to:

orient;

represent;

connect;

generate;

test;

evaluate;

revise;

transfer.

Scientific reasoning adds a specialized architecture for questions about evidence and explanation:

**observe;

question;

hypothesize;

model;

predict;

test;

evaluate;

revise;

explain.**

The two frameworks overlap because science is a form of structured problem solving.

But scientific reasoning places particular emphasis on the relationship between models and evidence.

What Should Science Education Produce?

Not merely students who can remember scientific conclusions.

It should increasingly produce learners who can ask:

What was observed?

What is inferred?

What model explains it?

What alternatives exist?

What does the model predict?

What evidence supports the claim?

What evidence would weaken it?

What assumptions are being made?

Where are the limits of the explanation?

That is a much stronger form of scientific literacy.

From Learning Science to Using Science

Eventually the learner should be able to approach a phenomenon that was not explained in the previous lesson.

They may not know the answer.

That is expected.

But they possess a structure for inquiry.

Observe.

Ask.

Propose.

Represent.

Predict.

Test.

Evaluate.

Revise.

Explain.

The teacher no longer needs to supply every intellectual move.

Scientific thinking has begun to become independent.

“The goal of science education is not to make the world look as if every question already has an answer. It is to teach us what to do when it does not.”

— Tymur Levitin

Continue Learning

For the general architecture of unfamiliar problem solving, read How to Solve a Problem You've Never Seen Before: The Architecture of Independent Thinking.

For the distinction between knowledge, understanding, ability and independence, continue with Knowing vs Understanding: The Four Levels of Real Learning.

For mathematical reasoning, see Understanding Mathematics: How Mathematical Thinking Develops.

For academic learning through another language, explore Learn a Subject Through Another Language: When Language Becomes a Tool for Knowledge.

Learn Science and Academic Subjects Online

Levitin Language School provides individual online education for children, teenagers, university students and adults internationally.

Our educational ecosystem works across:

Languages · Academic Subjects · Language + Subject

Academic learning can include mathematics, physics, biology, chemistry, economics, programming and other disciplines according to the learner's goals and teacher availability.

Students can also study academic subjects through another language when their educational situation requires it.

For a current practical physics route, see Online Physics Tutoring for School, University, and International Education.

International and U.S.-focused resources are available through Language Learnings.

Contact Levitin Language School

Email: notification@levitintymur.com
Telegram: https://t.me/TimurLevitin717
WhatsApp: https://wa.me/380972456826
Website: https://levitintymur.com/

About the Author

Tymur Levitin
Founder & Director, Levitin Language School

Educator and author working across language learning, academic education, scientific reasoning, problem solving and integrated Language + Subject learning.

His educational framework focuses on understanding systems, distinguishing knowledge from performance, developing independent reasoning and connecting academic knowledge with language when the learner's real educational goals require it.

Levitin Language School: https://levitintymur.com/
Language Learnings — USA: https://languagelearnings.com/
Language Thinking Laboratory: https://languagethinkinglab.blogspot.com/

Author contact: tymurlevitin@levitintymur.com