OBSERVE REALITY
E

Philosophy · A DIFFERENT PERSPECTIVE

Are You Living in a Simulation?

When reality is convincing enough, could we ever know the difference?

By Observe Reality Editorial ·

10 min read
An ordinary sunlit kitchen whose far corner dissolves into plain rendered geometry

There is no way to know for certain, and that is precisely what makes the question philosophically interesting rather than merely science fictional. No experiment has ever detected evidence that reality is simulated, and many philosophers argue the question may be untestable in principle. What can be said is that the simulation hypothesis is a modern, formally structured descendant of a much older philosophical worry about whether our experience reliably tracks an external reality at all.

An old doubt in a new form

Worries about whether our senses can be trusted are not new. In the early seventeenth century, the philosopher Rene Descartes used a thought experiment, often called the evil demon or evil genius scenario, to ask whether an immensely powerful deceiver could feed a person a completely convincing stream of false sensations, making an entirely fictional world indistinguishable from a real one. Descartes was not claiming this was likely. He was using the scenario as a tool to identify which of his beliefs could survive radical doubt, eventually arriving at the famous conclusion that the very act of thinking proves that a thinking subject exists, even if everything that subject perceives is false.

The twentieth century version of this thought experiment is the brain in a vat. In this scenario, a disembodied brain is kept alive in a vat of nutrients and wired to a supercomputer that feeds it electrical signals indistinguishable from the signals a real body would generate while walking through an ordinary day. From the inside, the brain's experience would be identical to that of an embodied person living a normal life. The philosopher Hilary Putnam used a version of this scenario in the 1980s to explore questions about meaning and reference, arguing, in a much debated argument, that a brain in a vat could not even coherently think the words I am a brain in a vat and have them refer correctly, because its words would be causally and semantically tied only to the simulated environment it had always known.

What unites Descartes's demon, the brain in a vat, and the modern simulation hypothesis is a shared structure: each imagines a scenario that would be subjectively indistinguishable from ordinary reality from the inside, and each uses that scenario to probe how much we can really know about the external world based on experience alone. The simulation hypothesis updates this old philosophical architecture with the language and assumptions of computing, information theory, and the trajectory of technological progress.

Bostrom's trilemma

The contemporary version of this debate was given its most influential formal statement by the philosopher Nick Bostrom in a 2003 paper titled Are We Living in a Computer Simulation?, published in the journal Philosophical Quarterly. Bostrom did not simply assert that we are probably in a simulation. Instead he constructed an argument in the form of a trilemma, a claim that at least one of three propositions must be true, even though we cannot currently know which one.

The first proposition is that almost all civilizations at our current level of technological development go extinct before ever reaching the technological capability to run high fidelity, consciousness-containing simulations of entire civilizations. The second is that advanced civilizations which do reach that capability are, for some reason, overwhelmingly unlikely to choose to run many such simulations, perhaps for ethical, legal, resource, or cultural reasons. The third is that if a meaningful number of advanced civilizations do both survive and choose to run many such simulations, then the number of simulated conscious beings would vastly outnumber the number of original, non-simulated beings, which would mean that any individual observer, reasoning from the available statistical evidence alone, should assign a high probability to being one of the simulated many rather than one of the rare originals.

Bostrom's argument is carefully structured as a conditional one. He does not claim we are in a simulation. He claims that if we reject the first two propositions as unlikely, which many technologically optimistic people are inclined to do, then we are logically committed to taking the third proposition, and therefore the possibility that we are simulated, quite seriously on probabilistic grounds. The strength of the argument depends heavily on assumptions that are themselves uncertain, including whether consciousness can in principle be substrate independent, meaning realizable on computational hardware rather than only biological neurons, and whether running civilization-scale simulations populated by conscious beings is physically and computationally feasible at all, even for a very advanced civilization.

What would have to be true for a simulated universe to work

Taking the simulation hypothesis seriously as a physical proposition, rather than only a logical one, requires grappling with some demanding technical assumptions. A simulation detailed enough to produce convincing conscious experience for its inhabitants would need either to simulate the relevant physics at an extremely fine grained level across an enormous spatial and temporal extent, or to use some form of selective rendering, in which only the parts of the simulation actually being observed in detail are computed at high resolution, a strategy familiar from video game engines but wildly more complex at the scale required for a convincing universe.

This raises genuine physical questions rather than purely philosophical ones. Some physicists have explored whether any observable features of physical law could hint at an underlying computational substrate, for instance whether space and energy show any sign of being fundamentally discrete, pixel-like, rather than smoothly continuous, since digital computation generally requires discretization. No confirmed evidence of such discretization has been found, and several proposed tests, including searches for specific signatures in cosmic ray physics, have returned null results or remain highly speculative and contested within the physics community.

A different kind of assumption concerns consciousness itself. The simulation hypothesis generally presumes that consciousness is substrate independent, meaning that a sufficiently accurate simulation of a brain's informational processes would give rise to genuine subjective experience, regardless of whether that processing happens in biological neurons or in silicon based computation. This assumption is contested in philosophy of mind. The philosopher David Chalmers has written extensively about consciousness and computation, including direct engagement with the simulation hypothesis in his book Reality Plus, arguing that if a simulation reproduces the relevant causal and informational structure of a brain in sufficient detail, there is no principled reason to deny that it would produce real conscious experience, a position that remains one perspective among several competing views on the so called hard problem of consciousness.

The testability problem

A central and persistent criticism of the simulation hypothesis, as commonly presented, is that it is difficult or impossible to test. If a simulation were built competently enough, it could in principle be engineered so that any potential internal evidence of its own artificiality is either absent or indistinguishable from ordinary physical law. Philosophers of science have long emphasized that a claim which can accommodate any possible observation, by design, offers little genuine explanatory traction and sits uneasily with standard accounts of what makes a hypothesis scientific, echoing older debates in the philosophy of science about falsifiability associated with thinkers such as Karl Popper.

Some researchers have nonetheless proposed specific empirical tests, reasoning that a simulation running on finite computational resources might need to take detectable shortcuts, for example a kind of built in spatial granularity or statistical regularities in high energy physics data that would not be expected from a fully continuous, non-simulated universe. Physicists including Silas Beane and collaborators have explored whether lattice-based simulation methods used in nuclear physics would leave observable fingerprints if applied to the fabric of reality itself, and have proposed looking for such fingerprints in the distribution of the highest energy cosmic rays. These remain speculative research directions rather than confirmed detection methods, and no result so far has provided evidence either for or against the hypothesis.

Because of this testability problem, many philosophers treat the simulation hypothesis less as a scientific claim awaiting confirmation and more as a modern philosophical tool, useful for sharpening questions about the nature of reality, the reliability of perception, and the conditions under which a belief can be said to be justified, much as Descartes's demon was useful without ever being taken as a literal historical claim about an actual demon.

Does it matter whether we could ever know

One response to the entire debate, associated loosely with a broader philosophical tradition sometimes called pragmatism, is that if a simulated reality were truly indistinguishable from a non-simulated one in every observable and experiential respect, the practical difference between the two scenarios might be negligible for how a person should live, reason, or make decisions. Pain would still hurt, relationships would still matter to the people in them, and moral reasoning about how to treat others would arguably proceed in much the same way regardless of the deeper metaphysical substrate underneath experience.

Others resist this move, arguing that the question of what is ultimately real retains importance even if it has no immediate practical consequence, because human beings have historically cared about truth for its own sake, independent of whether that truth changes behavior. Philosophers working on skepticism, including those engaging with Bostrom's trilemma directly within academic philosophy journals, have noted that the simulation hypothesis forces a renewed and more technically precise confrontation with very old epistemological questions about what justifies confidence in the external world, questions that predate computers by centuries but that computing technology has made newly vivid and concrete.

There is also a psychological dimension worth separating from the strictly philosophical one. Public fascination with the simulation hypothesis, amplified by its uptake in popular culture and by occasional remarks from prominent technologists, has sometimes outpaced the comparatively narrower and more cautious claims actually made in Bostrom's original academic paper. The original argument is explicitly probabilistic and conditional, not a confident assertion that we are simulated, and treating it as a settled or even a likely truth goes well beyond what its author or the surrounding philosophical literature claims to establish.

SOURCES AND FURTHER READING