Radar Perene / Archive / science

The error of thinking \"we have seen this before\" guarantees the next step

◦ Index methodology v2.2 (working papers with DOI). See the methodology.

Science

There is a phrase that signals experience and delivers the opposite: "I've seen this movie before". Whoever says it feels they are playing a trump card — the lived years, the repertoire, the cycles crossed. But notice what the phrase does underneath: it converts the recognition of a resemblance into the certainty of an outcome, with no step between one and the other. The recognition is genuine; the certainty is contraband.

Recognizing a familiar pattern in the present licenses no conclusion about the next step — because familiarity is a state of the observer's memory, not a property of the process being observed. Psychology calls the mechanism representativeness bias: judging the probability of an outcome by the situation's resemblance to a mental prototype, instead of by the actual frequency of outcomes. The house's historical analog engine was designed, in good measure, as an institutional counterweight to that mechanism — and this article explains why a counterweight is needed.

What "I've seen this movie" actually consults

When intuition declares it has seen this movie, it has consulted an archive — just a terrible one. Human memory does not store episodes with weights proportional to their frequency: it stores them with weights proportional to emotion, recency and narrative convenience. The crash that hurt is kept in high definition; the several similar scares that ended in nothing vanish without a record, because outcomes without drama generate no remembrance. The result is a catalog in which spectacular endings are over-represented — and it is that catalog which "we have seen this before" consults in a voice of authority.

The consequence is measurable in public behavior: at every market tension, recognitions of different movies multiply. One veteran recognizes the script of a crisis; another, with equal sincerity, the script of a recovery. Both have "seen this movie" — different movies, kept by different memories, edited by different biases. If familiarity guaranteed the next step, it would at least have to guarantee the same next step for everyone who feels it.

From resemblance to outcome, the road has two bridges

Between "the present resembles X" and "what happened after X is about to happen" stand two logical bridges, and representativeness bias jumps both without looking.

The first bridge is verifying the resemblance: similar by what criterion, measured on which variables? The feeling of familiarity does not answer — it is global, diffuse, impressionistic, and for that very reason unverifiable. The second bridge is the sufficiency of the resemblance: even a real, measured similarity does not imply a single outcome, because similar starting points have historically opened into diverging endings — the argument that opened this trail. The statistical cousin of that leap is well known: striking correlations between series that have nothing to do with each other, the territory of spurious regression, where the resemblance of trajectories also seduces and also cannot bear the weight placed on it.

The counterweight: replacing remembrance with the archive

The defense against a memory bias is not better memory — it is taking memory out of the circuit. That is the function of the analog engine under study at the house: when the feeling of "we have seen this" appears, the question is reformulated into a shape memory cannot manipulate. Similar in what? — declared variables. Similar how much? — proximity measured, not felt. And what came after the lookalikes? — all the archive's outcomes, with their frequencies, including the mediocre ones remembrance discarded for lack of drama.

No concrete episode from the house's data bank needs to appear here for the contrast to be clear, because the contrast is architectural: intuition consults a catalog edited by emotion and returns a single outcome with a feeling of certainty; the method consults a complete catalog and returns a fan of outcomes with the proportions in view. The two processes may even agree at times — but only one of them shows its work, and only one of them is auditable when it errs.

Experience still counts — in a different office

None of this retires experience; it reassigns it. The trained eye is irreplaceable for generating the hypothesis of resemblance — noticing that something in the current environment rhymes with something once lived is a real talent, and the best queries to the archive are born from it. The error begins when the generator of hypotheses also wants to be their judge. In the mature division of labor, lived experience proposes and the archive disposes: the déjà-vu feeling opens the investigation that the method closes. "We have seen this before" is an excellent first sentence — provided the second is "let us check what happened all the times".

Frequently asked questions

Are representativeness bias and overconfidence the same thing?

Neighbors, not twins. Representativeness is the specific shortcut of judging probability by resemblance to a prototype; overconfidence is the inflated calibration of one's own certainty. The first frequently feeds the second.

Do experienced investors suffer less from this bias?

The decision literature suggests caution: experience enlarges the catalog of prototypes, which refines hypotheses — and simultaneously strengthens the feeling of familiarity, which is the bias's fuel. More movies watched, more movies to recognize with false assurance.

How does one tell a useful recognition from a misleading one?

By what the person does next. The useful recognition becomes a verifiable question — criteria, variables, a query against the full record. The misleading one becomes an immediate conclusion. The content of the feeling is identical; its destination is the test.

Does the house's engine eliminate the bias?

It eliminates the bias of the query — the search and the split of outcomes pass through no one's memory. The engine's design choices, which are indeed human, are fixed in advance and documented, which is the most any method can promise.

---

Continue the trail: From the analog engine to the signal registry: how the two converse →

House reading: the complete catalog — including the drama-free outcomes memory erases — is in the Atlas; today's reading, in the Diário.

Putting an analogy that feels obvious through the sieve of the full archive is a second opinion the house's bench offers to those who seek it.