How to work with this page

This page is the workshop for Lesson 7 of the Core unit, How does shared knowledge differ from personal knowledge? You work through it alone, with your Lesson 7 handout and the two readings beside you: Popper, Conjectures and Refutations, ch. 1, §§I–II, and Fricker, Epistemic Injustice, §1.3. It has one job: to turn the two readings into two tools you can use on a case you have not seen before.

Each part is built the same way:

Explanation — short, with the readings’ own examples.
Exercise — you type or choose, then press Check. Nobody sees your mistakes. The page counts your first try; that number is for you.

Two rules for this page

Retrieve before you look. Where the page says “from memory”, scroll the explanation away first. A wrong guess you then correct sticks better than a right answer you copied.

Every exercise says what goes in the field: a word from a closed list, yes or no, or a choice. Spelling counts as written; capitals do not.

The page has three stages. Stage 1 puts the lesson’s case on trial and asks you to take a side before you have the tools. Stage 2 gives you the tools: Popper’s falsifiability, Fricker’s testimonial injustice, and the error-spotting. Stage 3 sends you back to your verdict with the tools in hand. Your sheet tells you when each stage is due.

Recall — Lessons 5 and 6 from memory

Before anything new: eight questions on the last two lessons, without your notes. Guess first: how many of the eight will you get? …… of 8.

Exercise 1 — Lessons 5 and 6

Write one word or a short phrase in each field, in English, without looking anything up.

Knowledge we get from what others tell us is .
Hume: believe testimony only as far as experience shows it reliable — the view called . Reid: believe others by default — the principle of .
The Post Office prosecuted more than 900 subpostmasters on the word of a computer system called .
Kahneman’s fast, automatic, intuitive thinking is ; slow, effortful thinking is .
Knowledge independent of experience is ; knowledge from experience is .
Adjusting principles and judgements about cases until they fit together (Rawls) is .

Fewer than 5? Read the Lesson 5 and 6 key terms on your handouts once more, then come back and do not redo the exercise.

Stage 1 — The case first

Why the case comes first

A tool you have never needed is a tool you will not remember. Take your side now; Stage 2 will give you the words for what you did.

In 2012, five behavioural scientists — among them Dan Ariely and Francesca Gino — reported in PNAS that people who sign an honesty declaration at the top of a form report more honestly than those who sign at the bottom. The finding was cited several hundred times, written into insurance and tax paperwork, and trialled by government “nudge” units. In 2021 three unpaid bloggers at Data Colada, with anonymous researchers, showed that the insurance data were fabricated; PNAS retracted the paper. In 2023 they documented fabricated data in several of Gino’s own studies. Harvard found research misconduct and in 2025 revoked her tenure. Gino denies wrongdoing; she sued Harvard and the bloggers for some $25 million, and the claims against the bloggers were dismissed in 2024.

Exercise 2 — Read the case

Choose one answer per question.

Was the honesty finding falsifiable?

Yes — it forbids a result: people signing at the top reporting no more honestly than those signing at the bottom No — honesty cannot be measured No — it was fabricated, so it was never a scientific claim

Who caught the fabrication?

The journal’s peer reviewers, before publication A replication study commissioned by PNAS Unpaid outsiders, about ten years after publication

Now take your side, in your notebook, before you go on. Two or three sentences: did the self-correcting machinery of shared knowledge work here (Position A: the claim was public and checkable, outsiders refuted it, and the record was revised), or did it break down (Position B: peer review missed it for a decade, volunteers caught it off the books, and the reward was a lawsuit)? Name your side and one reason. You will come back to this at Stage 3.

Stage 2 · Part A — Popper: falsifiability

A1. Popper’s question

Popper’s problem is not “When is a theory true?” It is “When is a theory scientific?” — the demarcation problem. Science often errs, and pseudo-science may stumble on the truth, so truth cannot be the mark. Nor can “it rests on observation”: astrology has “a stupendous mass of empirical evidence”.

The trap

Unfalsifiable does not mean false or worthless. Popper thinks Freud and Adler saw “certain things correctly”. Their theories are simply not testable — yet.

A2. Why explaining everything is a weakness

Popper’s friends admired Marx, Freud and Adler because their theories explained everything. Popper describes a child’s case to Adler, who explains it at once by inferiority feelings without having seen the child. “Because of my thousandfold experience,” says Adler. Popper: “And with this new case, I suppose, your experience has become thousand-and-one-fold.”

Two men: one pushes a child into the water to drown it, the other dies trying to save it. Freud explains both (repression, sublimation), and so does Adler (both men need to prove something to themselves). A theory that fits the drowning and the rescue forbids nothing — and a theory that forbids nothing cannot be tested.

Einstein is the opposite. His theory predicted that starlight passing the sun is bent, so stars near the sun appear shifted. Before Einstein, everyone expected no shift. If the eclipse photographs of 1919 had shown none, “the theory is simply refuted”. That is a risky prediction.

A3. The tools from the reading

Term	What it means
Falsifiability	Some possible observation would refute the theory. “Testability is falsifiability.”
Prohibition	A good theory forbids certain things to happen. “The more a theory forbids, the better it is.”
Risky prediction	Without the theory, we would have expected the opposite. Only these confirmations count.
Corroboration	Support from a serious but unsuccessful attempt to refute.
Conventionalist twist	Rescuing a refuted theory by reinterpreting it or the evidence ad hoc. Always possible — but it lowers the theory’s scientific status.

The test, in one question: what does this claim forbid? If the answer is “nothing that could happen”, the claim is not falsifiable.

Exercise 4 — What does it forbid?

For each claim, choose the observation that would refute it.

Signing at the top makes people more honest.

Some people sign at the top and still lie In a large trial, people who sign at the top report no more honestly than people who sign at the bottom The study’s authors turn out to have fabricated data

Light is bent by the sun’s gravity.

Eclipse photographs show stars near the sun exactly where they stand at night Some astronomers doubt the theory The effect is too small to see on an ordinary day

Row 1: one person lying does not refute a claim about more honesty on average. And fabricated data refute the study, not the claim — the claim could still be true. Data Colada showed the evidence was worthless; that is not the same as showing the effect is absent.

Exercise 5 — Name the move

Write exactly one of these five names in each field: risky prediction, corroboration, conventionalist twist, confirmation hunting (collecting instances that fit, without any test), unfalsifiable. The first row is done for you.

What happens	The move
A Marxist finds evidence for his theory on every page of the newspaper, and in what it leaves out.	confirmation hunting
Einstein’s theory says stars near the sun will appear shifted; nobody before him expected that.
The 1919 eclipse photographs show the shift the theory predicted.
The predicted revolution does not come; Marx’s followers reinterpret both the theory and the evidence until they agree again.
“Because of my thousandfold experience” — and each new case adds one more.
A theory explains the man who drowns the child and the man who saves it, equally well.
The honesty effect fails to replicate; a defender says the declaration only works “when people are paying attention”, which no study had measured.

Part B — Fricker: testimonial injustice

B1. The definition

Every time we listen, we judge how far to believe the speaker. That judgement can go wrong in two directions: a credibility excess (more credibility than the evidence warrants) or a credibility deficit (less). Fricker’s definition of the central case:

A speaker suffers testimonial injustice “if and only if she receives a credibility deficit owing to identity prejudice in the hearer”.

Two parts, and both are needed: the deficit, and its cause in prejudice. The hearer’s duty is plain: to “match the level of credibility she attributes to her interlocutor to the evidence that he is offering the truth”.

The double wrong

The hearer fails epistemically (misses out on knowledge) and ethically (wrongs the speaker “in her capacity as a knower”). The second is what makes it an injustice and not just an error.

B2. The test, in four steps

Step	Question	If no
1	Did the speaker get less credibility than the evidence warrants?	No testimonial injustice. Excess is not it, even when it harms.
2	Was the cause a prejudice — an ethical fault in the hearer?	No injustice: an innocent error, or a merely careless one.
3	Is the prejudice about the speaker’s social identity?	Not the central case.
4	Does the prejudice follow the speaker through every area of life — a tracker prejudice?	Incidental testimonial injustice: real, but confined to one setting. If yes: systematic.

The reading’s cases, by step: the GP and the professor get too much credibility (step 1: no). The philosopher who takes an ethicist for a medic after a web search makes an innocent error; after a careless search, still no injustice — “a stupid mistake” (step 2: no). A journal panel prejudiced against a research method, and scientists who dismiss philosophers of science at a conference: incidental. Tom Robinson, before an all-white jury in 1935 Alabama: systematic — racism follows him into the courtroom, the job market, the street.

Exercise 6 — Run the test

Two fields per case. Testimonial injustice? yes or no. Kind: systematic, incidental, or none. The first row is done for you.

Case	Testimonial injustice?	Kind
The jury does not believe Tom Robinson, though Atticus Finch has proved he could not have struck Mayella.	yes	systematic
A philosopher looks a colleague up online, takes him for a medic, and gives his view on ethics less weight. He is a medical ethicist.
Patients trust their overworked GP on questions far beyond his training.
Journal referees discount every paper that uses a research method they dislike.
At a scientists’ conference, whatever the philosophers of science say is dismissed as out of touch.
A landlord, a police officer and a bank clerk each disbelieve a speaker for the same reason: her ethnicity.
A teacher doubts a pupil’s excuse because the same pupil invented one last week.

Last row: the teacher matches credibility to evidence — the pupil’s own record. That is what the hearer’s duty asks. Change the reason to “pupils from that village always lie”, and the answer changes.

Exercise 7 — Which step fails?

Each case is not the central case of testimonial injustice. At which step does it fail?

The admiring junior colleague gives the professor’s draft too little criticism, and the professor is let down at the conference.

Step 1 — it is a credibility excess, not a deficit Step 2 — there is no prejudice It passes every step

A juror misremembers which witness said what, and so gives one witness’s testimony less weight.

Step 1 — there is no deficit Step 2 — the cause is an error, not a prejudice Step 4 — it is incidental

The boy from the elite school is believed on everything all his life, and grows up arrogant and closed to criticism. One single conversation in which he is over-believed:

Step 1 — a single moment of excess is not testimonial injustice (Fricker allows only a cumulative case) Step 3 — class is not a social identity It is systematic testimonial injustice

Part C — Spot the error

C1. How answers go wrong in this lesson

Six things students write about Popper and Fricker, and the one mistake in each. Naming the mistake is what lets you find it in your own essay.

Exercise 8 — Name the mistake

Write exactly one of these four labels in each field: unfalsifiable for false (treats “not falsifiable” as “false” or “worthless”), confirmations for tests (counts fitting instances as support), error for injustice (calls a deficit without prejudice an injustice), excess for injustice (calls being over-believed testimonial injustice). The first row is done for you.

The student writes	The mistake
“Psychoanalysis is not falsifiable, so Freud was simply wrong about everything.”	unfalsifiable for false
“Adler’s theory is very well supported: he applied it to thousands of patients.”
“The philosopher did the ethicist a testimonial injustice when she misread his web page.”
“Famous researchers like Ariely suffer testimonial injustice too: everyone believed them far too readily.”
“Astrology has been confirmed by millions of horoscopes that came true.”
“‘Everything happens for a reason’ is unfalsifiable, so it is meaningless.”

Row 4 is the one to watch. Ariely and Gino were over-believed, and that over-belief is part of why the fabrication lasted a decade. Fricker has a word for it — credibility excess — but it is not testimonial injustice. Saying exactly what went wrong is the work.

Exercise 9 — Where does the tool need a judgment call?

Not every application is mechanical. Decide for each:

“Peer review catches fraud.” Is this claim falsifiable as it stands?

Yes — the honesty study refutes it No — it is a claim about institutions, not nature Cannot say yet — it depends on what “catches” means: eventually? within a year? before publication? Each reading forbids something different

Anita Hill, 1991: senators suggested she was a “scorned woman” or lying for political gain. Was this testimonial injustice?

Yes — she was disbelieved, and that is enough No — the senators had political reasons, not prejudice Cannot say yet — it depends on step 2: was the discounting driven by a prejudice about women, or by political disagreement? The definition tells you what to look for, not what you will find

This is the philosophy

The answer key to Exercise 9 says cannot say yet both times. That is not a dodge. Pinning down what a claim forbids, or what caused a deficit, is the work — the drill can only show you where it is.

Stage 3 — Back to the case

Exercise 10 — The retrieval checks

Two questions from your handout, from memory. Write your answer in the box in full sentences; the bar fills as your words match the model answer’s key words. Then compare.

State Popper’s falsifiability criterion: what must a claim specify to count as scientific, what made Einstein’s light-bending prediction the model case, and why was Adler’s “thousandfold experience” a weakness?

State Fricker’s definition of testimonial injustice: what must happen to a speaker’s credibility, and from what cause? When is a case systematic?

Your verdict, with the tools

Take out what you wrote at Stage 1. Now rewrite it in three sentences:

Your side: did the machinery work (A) or break down (B)?
Your reason — and say which tool it uses: falsifiability (was the claim refutable, and who did the refuting?) or credibility (whose word counted — the famous authors’, the reviewers’, the volunteers’?).
The price your side pays. A: a decade in which a false claim functioned as knowledge, and correction that came at personal risk. B: the fraud was caught, retracted and punished — what would the machinery working have looked like, if not this?

Then check your own paragraph:

Did you keep the fraud was eventually corrected apart from the machinery worked? The retraction alone settles nothing.
If you used Fricker: did you call the authors’ over-belief a credibility excess, not testimonial injustice?
Did you say the price in your own words, not the sheet’s?

Only now: compare with the three model paragraphs — two good ones, one bad one.

Done?

You can now:

say what Popper’s demarcation problem is, and why truth cannot settle it;
ask of any claim what it forbids, and recognise a risky prediction;
name confirmation hunting and the conventionalist twist when you see them;
state Fricker’s two-part definition and run the four-step test;
tell systematic from incidental, and both from innocent error and credibility excess;
see that unfalsifiable is not false — in your own essays first.

Go back to your sheet.