"Nothing is true, everything is permitted." This phrase, often attributed to Nietzsche or to the assassins of the Hashashin sect, captures a contemporary temptation: throwing out the baby of reality with the bathwater of social construction. Faced with deepfakes, filter bubbles and conspiracy theories, a vertigo grips us: how can we know what is real? Must we give up on truth altogether?
This investigation proposes a narrow but workable path: fallibilism. Neither naive dogmatism (reality is transparent) nor nihilistic relativism (nothing is real). Our access to reality is mediated, partial, revisable — but improvable. And it is precisely because it collides with a reality that resists that we can make progress. Here are six bearings for surveying this boundary. Throughout, one principle: separating what science has established from what belongs to philosophical speculation, which we will flag as such.
The first boundary of the real is not a transparent pane but a translator. Our brain does not merely record stimuli passively: it interprets them, completes them, anticipates them. Optical illusions are the most spectacular proof: two squares of identical grey look different depending on their context. This is not a defect, it is an adaptive mechanism.
Cognitive neuroscience describes perception as a predictive process: the brain constantly generates hypotheses about the world and tests them against sensory data, with prediction error correcting the model. This is the theoretical frame of the "free-energy principle" proposed by the neuroscientist Karl Friston (2010), which remains a debated unifying theory rather than settled fact — but rests on robust observations: our access to reality is built from the inside, and reality resists and corrects our errors.
Two solid experimental results illustrate this mediation. The "invisible gorilla" experiment by Daniel Simons and Christopher Chabris (1999) shows that nearly half of viewers absorbed in a counting task fail to see a gorilla cross the scene: attention sculpts what we perceive. And the affair of "the dress" (2015), analysed by Lafer-Sousa, Hermann and Conway in Current Biology, shows that the same photograph is seen as blue/black by some and white/gold by others, depending on the assumptions each brain makes about the lighting. Knowing that one infers is the surveyor's first skill. This does not mean "nothing is real": the dress has physically measurable pixels, the gorilla really is in the video. It means our access is mediated, and that knowing this mediator is essential so as not to mistake the map for the territory.
One of the most frequent confusions is to believe that what is socially constructed is illusory. Yet in The Construction of Social Reality (1995), the philosopher John Searle proposed a decisive distinction between brute facts (the mountain, the mass of a stone) and institutional facts (money, borders, marriage). The latter exist only because a collective assigns them a function and believes in it — Searle's formula is "X counts as Y in context C" — but they are real in their effects.
Money really is money: it buys bread, it creates debt, it makes people live or die. Saying it is constructed does not make it any less real. It is another mode of existence of the real, dependent on human agreement. Likewise, a border between two countries is a line on a map, but it has very concrete consequences: customs, visas, the separation of families.
This distinction is crucial for not falling into the relativist trap: everything is constructed, so everything is equally valid? No. Constructed does not mean arbitrary. Some constructions are more robust, more coherent, more useful than others. And above all, brute facts go on existing: you cannot build a bridge that defies gravity.
Philosophy does not prove that reality exists — no demonstration definitively refutes the solipsist who claims to dream the world. But it sheds light on a decisive practical point: the position "nothing is real" is sterile and self-refuting. If nothing is real, the assertion itself is not real either; and above all, one can no longer denounce any lie, any injustice, any forgery — because there would be no truth left to serve as a yardstick.
Radical scepticism, pushed to its limit, disarms precisely those it claims to awaken. Authoritarian regimes love relativism: if truth does not exist, then power can impose its version without contradiction. This is why the surveyor refuses the comfort of "anything goes" and accepts the discomfort of the quest for truth.
A remark is needed here about a philosophical argument often invoked to claim that "nothing is real": the simulation hypothesis. In 2003, the philosopher Nick Bostrom published in The Philosophical Quarterly a rigorous argument showing that at least one of three propositions must be true, including the one according to which we would be living in a computer simulation. Caution: Bostrom does not claim to prove that we are in a simulation; he demonstrates a trilemma. This is speculative philosophy, not an established scientific result. Even if the hypothesis were true, it would not dissolve the distinction between true and false within our world: a lie would remain a lie, a miscalculated bridge would collapse. Metaphysical speculation about the ultimate nature of the real never exempts us from the work of discrimination here and now.
Around 375 BC, in The Republic (Book VII), Plato offers the allegory of the cave: prisoners chained since childhood take the shadows projected on a wall for the totality of the real; one of them, freed, painfully discovers the world of objects and of the sun, then returns — misunderstood — to announce to the others that what they saw were only shadows.
Let us read it with rigour: Plato does not say "everything is illusion." On the contrary, he affirms that there is a reality truer than the shadows, and that knowledge is an ascent towards it. It is an anti-relativist parable. The freed prisoner does not say "your reality is worth as much as mine"; he says "you see shadows, I see real objects." The cave is not an argument for relativism, but for the possibility of a truer access to the real.
In 1981, the philosopher Jean Baudrillard published with Galilée Simulacra and Simulation. In it he describes a hyperreality: a regime where signs no longer refer to a real but refer to one another — the map that precedes the territory, the copy without an original. Here again, interpretive caution: Baudrillard is doing provocative cultural theory, not physics; his work is a philosophical essay, to be read as such.
Reading "hyperreality" as a literal description of the world — "nothing is real" — is to commit exactly the error this investigation fights. Its value is heuristic: it names an observable tendency (the proliferation of self-referential representations), not an abolition of the world. And this tendency now has a technical face: deepfakes. In March 2022, a fake AI-generated video showing Ukrainian president Volodymyr Zelensky calling on his soldiers to surrender was circulated after a news website was hacked, before being debunked within hours — the first documented use of a deepfake in an armed conflict. The episode illustrates Baudrillard's point precisely without proving him right on the substance: the simulacrum did not replace the real, it was confronted with the real (Zelensky filmed live from Kyiv) and unmasked. Filters, synthetic images and deepfakes mask, distort, compete with the real; they do not abolish it. The surveyor must learn to spot these simulacra without believing they have replaced the world.
Fallibilism — a term coined by Charles Sanders Peirce and central to Karl Popper — affirms something quite different: none of our knowledge is certain or definitive, yet not all of it is of equal worth. Some claims withstand attempts at refutation better, overlap more, predict more accurately. Our access to reality is partial and revisable, and it is precisely for this reason that it is improvable: we can eliminate errors, so there is a reality to collide with.
This bearing also applies to our collective fears about the media. The idea that algorithms mechanically lock us into homogeneous "filter bubbles" is intuitive — but the literature review by the Reuters Institute (Arguedas, Robertson, Fletcher and Nielsen, 2022) concludes that echo chambers are less widespread than commonly assumed, finds no empirical support for the filter-bubble hypothesis, and paints a nuanced picture of polarisation. This is fallibilism at work: an appealing claim, confronted with the data, turns out to be more fragile than advertised. The danger remains real but must be measured, not fantasised.
Fallibilism keeps doubt and progress; relativism abandons the second along with the first. It is this criterion — our access to reality is mediated and constructed, but reality exists and our access is improvable — that allows us to hold both ends without falling on either side. It avoids naivety (believing that we see the world "as it is") and nihilism (believing there is no world to see).
Between the dogmatist who thinks he knows everything and the sceptic who thinks nothing can be known, the surveyor holds the uncomfortable but fruitful position of one who knows that we can know better. The six bearings presented here turn a vertigo — "how can we know what is real?" — into a method. They do not abolish mediation, which is impossible, but they teach us to take it into account without drowning in it. Partial and lucid access to the real is infinitely better than a fantasised reality or a denied one.
Dossier : Égrégore & souveraineté cognitive
Accueil · Collection · Journal · Dossiers · Sources