In a very informal survey on the campus of a STEM oriented university, students were asked, “What is the probability we are living in a simulation?” Of those who were willing to respond, a little over half thought there was some probability that we are. When asked to explain their thinking on this, most related the possibility of living in a simulation to the recent advances in technology, particularly the recent gains in artificial intelligence.
This hypothesis regarding our existence has gotten some purchase in American culture due to its promotion by science and technology gurus such as Elon Musk and Neil DeGrasse Tyson. Even popular writer and cognitive scientist/philosopher David Chalmers is willing to give it a 33% chance. While it seems to be a tech-savvy idea, this notion is ironically anti-science.
One of the first problems with this hypothesis is there is no way to have access to the world of the simulator, so there is no empirical way to test and verify its existence. This by itself does not make it untrue – there are many things in this world which we hold to be true that evade scientific testing. For these we need different types of evidence – observations from which we can infer a conclusion. For a rational person to place one’s trust in a simulation as a way to explain our universe still requires some evidence, but there seems to be little more than conjecture being offered.
In philosophical circles, the plausibility for our simulated existence hinges on the ability of our descendants to create simulations of our world. This would require computational resources many factors greater than what we have access to now. Theoretical physicists at Oxford (Ringel and Kovrizhin) have recognized in their attempts to model quantum effects that the computational demands grow exponentially with increases in the complexity of simulations. There simply would not be enough resources in our universe to model our universe.
Anyone who supposes we could faithfully simulate this world, and therefore be in a simulation ourselves grossly under appreciates the layers of complexity which make up the experiences we have as objects in this supposed simulation – what all would have to be simulated. The “life-like” environments of current video games gives us an unreasonable expectation for developing a simulation of the world we actually live in. The data feeds from our real world experiences are exponentially greater than the visual and audio feeds of data which make up the simulated world of a video game. In a video game, a high proportion of your sense data is related to the task at hand, and is only sufficient to accomplish the task. We can play along in the game even if we have a very limited amount of sense data.
In the experience of our present reality, our sense data extends way beyond audio and visual stimuli and also includes smell, taste, six different mechanoreceptor’s (your sense of touch) and proprioception (your ability to know where all the parts of your body are located in space without you looking). We receive so much sensory data from our environment that in order to function our brains must filter out a tremendous amount of it. This requires some higher executive function which must decide which data to pay attention to and which data to ignore.
Another layer of complexity is realized in our human intelligence – how we deal with the sense data we do admit in for processing. Cognitive scientists have identified nine different types of intelligences[1] which make up what we collectively refer to as human intelligence. In any given moment, we must select which of these intelligences is most appropriate for the given situation – it may be one or various combinations of them. In a sense, there is an intelligence above our intelligences – an even higher executive function which manages the various ways we are capable of interacting with our environment.
To accurately simulate this universe, the simulator would not only need to simulate what we perceive but also make it possible to detect and analyze the unseen. All the objects in the simulation would have to be susceptible to scientific analysis (e.g., microscopic inspection, chemical analysis, bombardment in particle accelerators). It is even more problematic when we consider that we currently have no way of reconciling general relativity (physics of the very large) and the quantum realm (physics of the very small) – they do not seem to play with each other well. If you cannot do the physics, you cannot write the algorithms to do the simulation.
In addition to all the physical or material objects which would require coding to be part of your experience of the world, there are a number of immaterial aspects to our human experience which would also have to be part our coding: emotions, memories, beliefs, creative thinking, and consciousness. Beyond our human selves there are also the immaterial realities of numbers, logic and language which would also require an algorithm.
To me, what destroys the plausibility of the simulation hypothesis is how it is disruptive to epistemology – how we know what we know. Take for example your vision. We think we see because light waves reflect off objects and enter our eyes, land on the retina which in turn converts that light energy into electrical impulses which are sent for processing in the optic centers of our brain. As an object in a simulation, however, what you “see” would be the result of receiving a streaming file of digital code which is processed by your part of the overall program. Yet, you as a simulated object you would somehow be coded to perceive sight as the former, not the latter. If I cannot accurately perceive the nature of myself, then I cannot accurately perceive what nature is.
Fundamentally, the simulation hypothesis is very anti-science. It undermines the ability to freely test and question the nature of the world we live in. Were we to be objects of a simulation, any experiment we might devise would simply be the result of coding by the simulator, and the data we obtain would also be fed to us by the simulator. We could only perceive the world in a way the simulator desires us to see the world.
In a simulated world, the activity of a scientist is an aimless and fruitless pursuit. All the science we think we have been doing is completely unproductive. But science is the very thing required to instantiate a simulated world, but in a simulated world it would be incapable of doing so. If we in our simulation were successful in producing a simulation of our world, it would only be because the one who simulated us coded for that as well – a simulation within a simulation. If we tend to prefer the simpler explanation for our existence, then the simulated world is a poor explanation.
For me, the more simple and plausible explanation for my existence is one which has been around for thousands of years: that I am a created being made of flesh and blood who has been placed into a created world which is independent of my own mind. I can have confidence in this reality because I have been informed about this by the creator of this world. Furthermore, this creator has given us plenty of indications of His existence. I think the fascination with the simulation hypothesis has less to do with its plausibility, and more with its potential to replace the traditional theistic worldview for those who do not like the implications of that worldview. The ways in which the simulation hypothesis is able to answer life’s important questions is decidedly worse than its plausibility, but that is fodder for another blog post (which you may read here).
[1] Multiple intelligences as identified by Howard Gardner include: verbal-linguistic, logical-mathematical, spatial-visual, bodily-kinesthetic, musical, interpersonal, interpersonal, naturalist, and existential.
Cicada Song 
1 thought on “Failure of the Simulation Hypothesis”