How Can Knowledge Be Reliable If It Is Built on Assumptions?
By Michael Thompson · Education Specialist; 10 years teaching the IB at Bromsgrove School · Published 20 November 2026
How can knowledge be reliable if it is built on assumptions? For TOK Title 3 in May 2027, a workable answer is that reliability depends less on having no assumptions than on whether a knowledge community states them, tests them and limits its claims to where they hold. That is why the title opens with 'in the production of knowledge': it points you at method, not at a list of things that might be wrong. You must write about the human sciences and choose one other area of knowledge from history, the natural sciences, mathematics or the arts. The essay is capped at 1,600 words and receives a single mark out of 10 on a global impression basis. This guide gives you the key terms, real examples for the human sciences and three possible second areas, a word-budgeted plan and the traps to avoid before your school's internal deadline.
Key Takeaways
- Use the title exactly as published: The full wording is 'In the production of knowledge, how can knowledge be reliable if it is built on assumptions? Discuss with reference to the human sciences and one other area of knowledge.'
- Reliable is not the same as true: Treat reliable as 'dependable for a purpose', so knowledge can be reliable within limits even when its assumptions are known to be false, as Newtonian mechanics is at everyday speeds.
- Hidden assumptions carry most of the argument: Stated assumptions can be tested and bounded, while hidden ones, like the Western-sample assumption Henrich, Heine and Norenzayan exposed in 2010, are where reliability tends to break down.
- The human sciences are fixed; the second area is your choice: The natural sciences, mathematics and history each give a different contrast, so choose the one where you can show assumptions being stated, tested or limited in a different way.
- The essay is worth two-thirds of your TOK grade: It is marked out of 10 by an IB examiner, the exhibition makes up the other third, and the TOK grade combines with the Extended Essay for up to 3 diploma points.
- Your teacher comments on one draft only: Hand in a complete draft with both areas of knowledge, defined terms and a judged conclusion, because schools upload final essays by 15 March 2027 and set earlier internal deadlines.
In This Article
- What TOK Title 3 is really asking
- Key terms: reliable, assumptions and production of knowledge
- Justified true belief, foundationalism vs coherentism and the problem of induction
- Human sciences examples for knowledge built on assumptions
- Choosing your second AOK: paradigm shifts in science, axioms and historical sources
- A 1,600-word essay plan for knowledge built on assumptions
- Common traps and how to use your one teacher-commented draft
- What to do next on TOK Title 3
1. What TOK Title 3 is really asking
> "In the production of knowledge, how can knowledge be reliable if it is built on assumptions? Discuss with reference to the human sciences and one other area of knowledge."
How can knowledge be reliable if it is built on assumptions? That is the question at the centre of Title 3 in the May 2027 TOK essay set. You must use the title exactly as published, with no paraphrasing or shortening. An essay that answers a reworded title is answering a different question, so a "tidied up" version of the wording carries risk and no benefit.
The title asks "how can", not "whether". It takes as given that knowledge rests on assumptions and asks how reliability is still possible. The opening phrase, "in the production of knowledge", points the essay at method: how knowledge communities state, test and limit their assumptions. An introduction that spends hundreds of words proving that assumptions exist is answering a question the title has already settled.
The essay facts:
- TOK is mandatory in the IB Diploma Programme core and asks students to reflect on "how we know what we claim to know" (IB).
- You answer one of six prescribed titles in no more than 1,600 words. The essay is externally marked and weighted 2/3, with the exhibition at 1/3 (IB subject brief).
- The examiner gives a single mark out of 10 on a global impression basis.
- Your TOK grade (A to E) combines with your Extended Essay grade for up to 3 diploma points.
- Schools upload final essays by 15 March 2027 and set earlier internal deadlines.
The current course has five areas of knowledge: history, the human sciences, the natural sciences, mathematics and the arts. The human sciences are fixed for this title. The second area is your choice.
2. Key terms: reliable, assumptions and production of knowledge
Everything in this section is interpretation written for this guide, not IB wording. Each term has more than one defensible reading. Pick one, state it in your introduction, and keep to it.
| Term | Reading 1 | Reading 2 | Reading 3 |
|---|---|---|---|
| Reliable | Dependable for a stated purpose and range | Consistently reproducible | Likely to be true |
| Assumptions | Stated: axioms, model simplifications, declared sampling frames | Hidden: unexamined beliefs about who is typical, what counts as rational, or that the future resembles the past | - |
| Built on | Foundation: remove the assumption and the structure falls | Scaffolding: the assumption can be revised and the knowledge survives | - |
| In the production of knowledge | The methods knowledge communities use: experiment, replication, peer review, proof, source criticism | - | - |
Reliable is not the same as true. A tool can be reliable within limits while resting on assumptions known to be false. Newtonian mechanics still works at everyday speeds, even though relativity shows its assumptions are wrong.
Hidden assumptions carry most of the argument. Stated ones are easy to list and test. The interesting cases are the beliefs nobody wrote down.
The foundation/scaffolding split is useful. Ric Sims describes language and social norms as "scaffolds" that make individual knowledge possible, and notes that "social norms tend to police the knowledge-making practices of individuals" (Pearson).
Define terms from examples, not dictionaries
Sims recommends defining each keyword from two concrete examples in two different areas of knowledge. He warns that keywords often carry "slightly different meanings in different areas of knowledge" (Pearson). If an example will not fit your definition of "reliable", Sims says you may need a better example, or the difficulty may point to "a systematic issue with the keyword", which is itself a knowledge question.
Knowledge questions you can adapt (written for this guide)
- Does making an assumption explicit make knowledge more reliable?
- Can knowledge be reliable for a purpose even if its assumptions are false?
- Who decides when an assumption has failed?
Takeaway: write one sentence for each term that names the reading you have chosen. Then test that sentence against one example from the human sciences and one from your second area of knowledge.
3. Justified true belief, foundationalism vs coherentism and the problem of induction
You do not need this theory to answer Title 3. A little of it can give your argument a clear structure, so treat it as optional support, not a philosophy lecture.
Justified true belief and the Gettier problem. The traditional analysis says knowledge is justified true belief. Edmund Gettier's 1963 paper described cases where someone holds a justified, true belief that still looks like luck rather than knowledge. Noah Lemos notes that "a satisfactory and widely accepted solution to the Gettier problem remains elusive" (Lemos, Cambridge). For your essay, the useful point is that justification can fail, so "reliable" cannot simply mean "justified."
The regress problem. Every justification rests on something else, and that also needs justifying. There are three classic responses:
- Foundationalism: basic beliefs stop the regress.
- Coherentism: beliefs support each other, like a web.
- Infinitism: the chain of justification never ends.
Foundationalism vs coherentism is a central debate in epistemology (Sosa, Knowledge in Perspective; Lemos).
The problem of induction (Hume). David Hume made this argument in A Treatise of Human Nature (1739-40) and An Enquiry Concerning Human Understanding (1748). He said that inferring the future from past observation assumes nature is uniform. That assumption cannot be justified without circularity, because "induction has worked before" is itself an inductive argument. On Hume's argument, every law in the natural sciences and every prediction in the human sciences relies on this assumption. One question to test in your human sciences section: how safe is that assumption when the subjects are people?
Takeaway: A foundationalist reading makes Title 3 a question about which assumptions are basic. A coherentist reading makes reliability depend on how well assumptions fit together and get cross-checked. Pick one lens and use it consistently.
4. Human sciences examples for knowledge built on assumptions
The human sciences are the required area of knowledge for this title, so your examples here carry the most weight. This site's TOK titles guide already uses two. Henrich, Heine and Norenzayan (2010) found that about 96% of participants in leading psychology journals came from Western industrialised countries holding about 12% of the world's population. That is a hidden sampling assumption: findings from one group were treated as findings about humans. Kahneman and Tversky's prospect theory (1979) tested the rational-choice assumption in economics and showed that people weigh losses and gains differently.
Four further real examples show how assumptions are made, exposed and argued over:
- *Margaret Mead, Coming of Age in Samoa (1928), and Derek Freeman's 1983 critique. Freeman argued that Mead's belief in cultural determinism led her to find what she expected to find, and that her adolescent informants may have misled her. The dispute shows that assumptions about fieldwork and informants* are argued out in public. Freeman's own work was also challenged: he worked mainly in a different part of Samoa and years after Mead, which brings its own assumptions.
- The replication crisis and the Open Science Collaboration (2015). This large-scale attempt to repeat published psychology studies, reported in Science, found that many original effects came out weaker or did not reappear. Replication is a method for testing assumptions, and here it exposed some that had gone unchecked.
- The Stanford Prison Experiment (Zimbardo, 1971). Later critiques questioned whether the guards behaved "naturally" or responded to cues from the researchers. This raises hidden assumptions about participant behaviour and researcher influence.
- Milgram's obedience studies (1963). The design assumed that obedience in a laboratory would tell us about obedience in real life. Critics have asked whether participants believed the set-up, which tests the assumption that lab behaviour generalises.
Claim: the human sciences make assumptions explicit and test them through replication, sampling and peer review. That is one answer to how knowledge can be reliable if it is built on assumptions. Counterclaim: hidden assumptions can persist for decades, and they are hardest to detect when the researchers checking the work share them. The WEIRD sampling problem fits this pattern: Henrich, Heine and Norenzayan argued in 2010 that researchers had routinely treated a narrow sample as typical of humans.
Bias is a recognised TOK question in its own right. The IB lists "Is bias inevitable in the production of knowledge?" as an exhibition prompt (IB subject brief).
Takeaway: pick one example where the method caught an assumption, such as the Open Science Collaboration. Pair it with one where the assumption went unnoticed for a long time, such as the WEIRD sampling problem. The pairing gives you a claim and counterclaim with real evidence on each side.
5. Choosing your second AOK: paradigm shifts in science, axioms and historical sources
The human sciences are fixed, so your second area of knowledge has one job: to show assumptions being stated, tested or limited in a different way.
| Area of knowledge | Main type of assumption | How it is tested or limited | Contrast with the human sciences | Risk for this title |
|---|---|---|---|---|
| Natural sciences | Paradigm assumptions about how nature works | Experiment, anomalies, paradigm change | Similar empirical testing, so the contrast is weaker | Repeating the human sciences argument with different examples |
| Mathematics | Explicit axioms | Logical consistency within the system | Assumptions are stated openly, not hidden in samples or models | Drifting into maths history with no knowledge claim |
| History | Interpretive and source assumptions | Source criticism, corroboration | Assumptions sit in the historian's selection and interpretation of sources | Turning the essay into a narrative of events |
Natural sciences: paradigm shifts in science examples
Start with the familiar case. Newtonian mechanics stays reliable at everyday speeds even though relativity shows its assumptions are false, and GPS satellite clocks must be corrected by about 38 microseconds a day. Then go further:
- *Thomas Kuhn, The Structure of Scientific Revolutions (1962):* scientists work inside a paradigm whose assumptions go unquestioned until anomalies build up. Reliability depends on the community, not just the data.
- Michelson-Morley experiment (1887): it failed to detect the luminiferous ether, an assumption later dropped. A null result challenged an assumption the whole field shared.
- Alfred Wegener's continental drift (1912): rejected until plate tectonics gained acceptance in the 1960s. Shared assumptions can make a community resist a correct idea.
Mathematics: assumptions stated as axioms
- Euclid's parallel postulate: Lobachevsky (1829) and Bolyai (1832) changed it and got non-Euclidean geometries that are different but consistent. Here a changed assumption produces a new system, not an error.
- Russell's paradox (Russell's 1902 letter to Frege): it showed a contradiction hidden in the foundations of set theory. Even explicit assumptions can carry hidden flaws.
- Gödel's incompleteness theorems (1931): a consistent system strong enough for arithmetic cannot prove its own consistency from within. Reliability has limits that the system cannot check itself.
History: assumptions about sources and objectivity
- Leopold von Ranke (1824) aimed to show the past "as it actually was", which assumes objectivity is possible.
- *E. H. Carr, What Is History? (1961)*, argues that the historian's selection shapes which facts count.
- The Hitler Diaries (1983), published by Stern and first authenticated by Hugh Trevor-Roper, were forgeries. An assumption of authenticity failed, and forensic source criticism exposed it.
Takeaway: pick the area that gives the sharpest contrast in how assumptions are stated and tested. Mathematics (explicit axioms) and history (interpretive assumptions) give a clear contrast with the human sciences. The natural sciences work too if you build the comparison around something the human sciences lack, such as a theory known to rest on false assumptions that stays reliable within limits, or paradigm change.
6. A 1,600-word essay plan for knowledge built on assumptions
This plan was written for this guide as a scaffold. It is not a model essay and not student work. The examples come from sections 4 and 5, with the natural sciences as the second area of knowledge.
| Section | What it does | Example to use | Word budget |
|---|---|---|---|
| Introduction | Quotes the title exactly: "In the production of knowledge, how can knowledge be reliable if it is built on assumptions? Discuss with reference to the human sciences and one other area of knowledge." Defines "reliable" and "assumptions", states the knowledge question and the line the essay will take | None needed | 200 |
| Human sciences claim | Shows a stated assumption being tested and revised | Kahneman and Tversky's prospect theory (1979) testing the rational-choice assumption in economics | 300 |
| Human sciences counterclaim | Shows a hidden assumption weakening reliability | Henrich, Heine and Norenzayan (2010): about 96% of participants in leading psychology journals came from Western industrialised countries holding about 12% of the world's population | 250 |
| Second AOK claim | Shows knowledge staying reliable for a purpose despite false assumptions | Newtonian mechanics at everyday speeds | 250 |
| Second AOK counterclaim | Shows where those assumptions break | GPS satellite clocks corrected by about 38 microseconds a day because of relativity | 200 |
| Comparison | Asks why one area can test its assumptions more directly than the other | Reuse the four examples, no new ones | 200 |
| Conclusion | States a judged answer and its implications | None needed | 200 |
| Total | 1,600 |
The line you take is your own. These are directions to test, not thesis sentences:
- Does reliability depend on whether assumptions are stated and tested, rather than on whether they are absent?
- If some assumptions can never be tested from inside a discipline, is reliability always provisional?
- Is "reliable" always reliable for a purpose, so the same assumption can be safe in one use and misleading in another?
1,600 words is a maximum, not a target. The plan uses four main examples, one for each claim and counterclaim. If the comparison section tempts you to add a fifth example, cut it. That section works best when it reuses the evidence you already have.
7. Common traps and how to use your one teacher-commented draft
Here are six common traps on this title and how to fix each:
| Trap | One-line fix |
|---|---|
| Treating "reliable" as meaning "true" | Define reliable as dependable for a stated purpose, then test whether the knowledge meets that purpose. |
| Listing assumptions without method | Show how a knowledge community states, tests or limits each assumption. That is what "in the production of knowledge" points to. |
| Arguing whether knowledge is built on assumptions | The title asks "how can", so it takes the assumptions as given. Spend your words on reliability. |
| Treating the second area of knowledge as an afterthought | Give it its own claim, counterclaim and example, with the same depth as the human sciences. |
| Name-dropping Kuhn, Hume or Gettier | Tie each thinker to a real, dated example, or cut the name. |
| Paraphrasing or shortening the title | Copy the prescribed title word for word from your school's official sheet. |
Getting the most from your one written draft
Your teacher can give written comments on one draft only. Your conversations about the essay are recorded on a planning and progress form, which is sent to the IB with the essay. Use those conversations to test ideas. Save the written draft for a finished argument.
Before you hand the draft in, check you have:
- a complete essay within 1,600 words
- the title quoted exactly
- both areas of knowledge developed, each with a claim and a counterclaim
- your key terms defined
- every example real and dated
- a conclusion that states a judged answer
A partial draft wastes the only written feedback you get. If your second area of knowledge or your conclusion is missing, your teacher can only comment on half the argument, and you won't get a second set of written comments.
8. What to do next on TOK Title 3
Start with the calendar. Schools upload final essays to the IB by 15 March 2027, and your school's internal draft deadline will be earlier. Your teacher can give written comments on one draft only, so a rushed first draft uses up your only round of written feedback.
Steps for this week:
- Confirm your internal draft deadline with your TOK teacher, and work backwards from it.
- Pick your second area of knowledge from history, the natural sciences, mathematics or the arts.
- List one stated and one hidden assumption for the human sciences and for your second area, each with a real, dated example.
This week, copy the title word for word into your planning document and write two sentences defining "reliable" before you choose any examples. Then check the official IB Theory of Knowledge page, which frames the course around "how we know what we claim to know."
FAQ
How can knowledge be reliable if it is built on assumptions?
One defensible answer is that knowledge can be reliable for a purpose when its assumptions are stated, tested and limited, as Newtonian mechanics remains reliable at everyday speeds even though relativity shows its assumptions are false.
Can assumptions be factual?
An assumption is accepted without being proven within the inquiry, but it can still turn out to be true, false or approximately true, which is why distinguishing reliable from true matters for this title.
Which second area of knowledge works best for TOK Title 3?
No area is officially better; the human sciences are fixed and you choose from history, the natural sciences, mathematics or the arts, so pick the area where you can show assumptions being stated or tested differently from the human sciences.
How many examples fit in a 1,600-word TOK essay?
The essay has a 1,600-word maximum, so a plan with four main examples, one per claim and counterclaim across two areas of knowledge, leaves room to analyse each rather than list them.
Can I reword or shorten the TOK Title 3 prescribed title?
No, the prescribed title must be used exactly as published by the IB, including 'Discuss with reference to the human sciences and one other area of knowledge.'
When is the May 2027 TOK essay deadline?
Schools upload final TOK essays to the IB by 15 March 2027, and each school sets an earlier internal deadline that students must follow.
References
- Theory of knowledge - International Baccalaureate - https://www.ibo.org/programmes/diploma-programme/curriculum/dp-core/theory-of-knowledge/
- Diploma Programme Subject Brief - Theory of knowledge, first assessment 2022 (International Baccalaureate Organization) - https://www.ibo.org/globalassets/new-structure/programmes/dp/pdfs/tok-sb-english.pdf
- Top 10 TOK tips for IB teachers: Tips 1 to 5 - https://www.pearson.com/international-schools/international-schools-blog/2020/09/top-10-tok-tips-for-ib-teachers-tips-1-to-5.html
- Foundationalism (Chapter 3) - An Introduction to the Theory of Knowledge - https://www.cambridge.org/core/books/an-introduction-to-the-theory-of-knowledge/foundationalism/099B42A9AEF0D577DA4637686D19310B
- Knowledge in Perspective - https://www.cambridge.org/core/books/knowledge-in-perspective/68B3AB7E152E5FC131BEB39A41CAF6D3
- An Introduction to the Theory of Knowledge - https://www.cambridge.org/core/books/an-introduction-to-the-theory-of-knowledge/39A91031D4D486C453868B393F01E60F