Memory Makes Testimony Necessary and Knowledge Fragile

A justified true belief can still miss knowledge by an accident.

Gettier’s counterexamples made the point unforgettable, and the point outlives the cases. A belief can pass every stated examination and reach the truth by a route that has nothing to do with the evidence supporting it.

Truth is not enough when luck connects it to justification.

That familiar trouble becomes oddly intimate when the source is one’s own past.

Most testimony has an obvious absence at its centre.

A report arrives from a parent, a textbook, a witness, an expert. The event is elsewhere. The speaker’s original contact with it cannot be opened on the desk like a filing cabinet.

Hume’s reductionism asks for independent reasons to trust the speaker. Reid’s anti-reductionism gives testimony a defeasible presumption instead: trust it unless something defeats the entitlement.

Either way, the hearer receives a present belief through an unavailable earlier source.

Memory is the same arrangement, except the witness is wearing one’s face.

The trace is not the event. Retrieval is not inspection. Whatever survives has been through encoding, consolidation, and a reconstruction that nobody attended.

Before looking back, try to recall the exact wording of the first sentence of this article.

The confidence of the reconstruction is not the wording itself.

Bartlett made the general point in Remembering.

In serial retellings of “The War of the Ghosts,” English participants made unfamiliar details familiar. Canoes became boats. Gaps acquired sensible connective tissue. A strange story developed the manners of a story already known.

Retrieval does not unspool a record. It assembles one from a trace, a cue, and a schema with strong opinions about narrative decor.

Which suggests a short and very depressing conclusion.

If memory is reconstructive, perhaps memorial knowledge is impossible. Every recollection is a little testimony from an unreliable narrator. No earlier scene remains present to check its minutes. Epistemology should close the office and take up gardening.

It should not. Fallibility is not defeat.

A belief may be justified without being infallible, as modest foundationalism insists. A process may be relevantly reliable without providing an introspective certificate of its reliability, as reliabilism proposes.

The question is not whether a memory feels preserved in amber. It is whether its truth arrives through a competence rather than luck.

That is Sosa’s refinement: accuracy, adroitness, aptness. A belief is apt when it is true because competent.

Memory is thus not rescued by a private glow of certainty. It is assessed by the relation between a trace, the conditions of retrieval, available defeaters, and the truth that results.

The inner witness can be heard without being sworn in as a saint.

The laboratory provides good reasons for the caution.

In Loftus and Palmer’s traffic-film study, participants asked how fast the cars were going when they smashed into each other gave higher speed estimates than participants asked when the cars contacted each other.

A week later, the smashed group was more than twice as likely to remember broken glass that was never in the film.

Post-event information can be credited to the event itself. That is source misattribution, not dishonesty and not a mind giving up on reality.

The Deese-Roediger-McDermott paradigm makes the confidence problem less courteous.

Present a list of words all associated with an unpresented one, and a large share of participants will recall the word that was never there.

Worse, most of those false recognitions are judged as remembered rather than merely known. A feeling of recollection can be as vivid at the wrong door as at the right one.

But no one should smuggle a word-list rate into a particular life and call it a verdict.

Laboratory lists, staged films, and suggestive interviews isolate mechanisms. Autobiographical events are emotionally charged, unevenly rehearsed, and embedded in an actual history.

Williams’s prospective study found that a substantial minority of adults with documented childhood sexual abuse reported no memory of the recorded abuse.

That a specific remembered event is inaccurate without independent corroboration is speculation, not a finding licensed by a word-list rate or a headline about false memories.

This is why source monitoring is the useful skill rather than confidence.

Where did this come from? Was it seen, told, inferred, or read afterward? Has it been rehearsed into smoothness? Is there anything outside the head that could check it?

None of those questions asks the memory how sure it feels, which is the one question it always answers.

What holds across the two channels is not that memory is merely testimony turned inward, or that testimony is a defective form of memory.

Each can fail in its own theatrical fashion.

Testimony can be prejudiced, deceptive, careless, or simply wrong. Memory can lose a cue, absorb later information, or reconstruct a coherent stranger from a partial trace.

It is that knowledge commonly begins after direct inspection has ended.

The speaker has left the room. The moment has passed.

What remains is a trace, a report, a practice of checking, and a belief that may be true because the channel that carried it was competent.

The past does not stay to testify. It leaves deposits.

Knowing is the work of reading them without mistaking the ledger for the day.