A few years ago, I picked up a book because it’s title referred to a subject I have always been interested, and some skims of various chapters contained many unfamiliar concepts. Unfortunately, upon sitting down to read the book, I was stunned at the shoddy thinking evident in the first few pages. Later on, I was flabbergasted that the person who wrote the book was not some pulp nonfiction hack writer but is actually a well-regarded philosopher in the field of the philosophy of mind.
The point of the introductory passage was to explain why a popular thought experiment known as a “
Brain in a vat” is not just technically difficult, that it is impossible in fact.
For those unfamiliar with the “Brain in a Vat” thought experiment, it is exactly the same problem presented by the Matrix’s "body in a vat" idea. The only variation is minor- the body is subtracted leaving just the brain. The key point of the problem is that the person does not know what is real because their access to their senses is being manipulated so that the person would believe that an artificially constructed world was real.
The book “Consciousness Explained” states that this scenario is impossible in fact because the number of computations necessary for such a simulation is high. Further, the problem is “insoluble” because there is a “combinatorial explosion”. However, rather than demonstrate via proof this supposed
computational complexity, or site an expert in computer science that does the same, he makes inferences based on his observation of computer games to jump from “lots of computation” to make the unsupported and surprising assertion that such tedious complexity makes the problem “insoluable”/ “impossible”.
Impossible on current computers? Yes, yes, of course. But this is not what the author Daniel Dennet is asserting. He is asserting that regardless how powerful computers of the future get, the thought experiment is based on an impossible scenario.
Mr. Dennet would have done better by starting his book by consulting authorities in the field of computer science rather than making frail conjectures based on naive notions of how computer simulations work.
Anyway: Some key passages of Dennet’s proposition:
Suppose evil scientists removed your brain from your body while you slept, and set it up in a life-support system in a vat. Suppose they then set out to trick you into believing that your were not just a brain in a vat, but still up and about, engaging in a normally embodied round of activities in the real world….
But now suppose the scientists, having accomplished all this, tackle the more difficult problem of convincing you that you are not a mere beach potato, but an agent capable of engaging in some form of activity in the world. Starting with little steps, they decide to lift part of the 'paralysis' of your phantom body and let you wiggle your right index finger in the sand. They permit the sensory experience of moving your finger to occur, which is accomplished by giving you the kinesthetic feedback associated with the relevant volitional or motor signals..., but they must also arrange to remove the numbness from your phantom finger, and provide the stimulation for the feeling that the motion of the imaginary sand around your finger would provoke.
Suddenly, they are faced with a problem that will quickly get out of hand, for just how the sand will feel depends on just how you decide to move your finger. The problem of calculating the proper feedback, generating or omposing it, and then presenting it to you in real time is going to be computationally intractable of even the fastest computer, and if the evil scientists decide to solve the real-time problem by pre-calculating and “canning” all the possible responses for playback, they will just trade one insoluble problem for another: there are too many possibilities to store. In short, our evil scientists will be swamped by combinatorial explosion [sic] as soon as they give you any genuine exploration powers in this imaginary world…
…Throw a skeptic a dubious coin, and in a second or two of hefting, scratching, ringing, tasting, and just plain looking at how the sun glints on its surface, the skeptic will consume more bits of information than a Cray supercomputer can organize in a year. Making a real but counterfeit coin is child’s play; making as simulated coin out of nothing but organized nerve stimulations is beyond human technology now and probably* forever.
…One conclusion we can draw from this is that we are not brains in vats- in case you were worried. Another conclusion we can draw from this is that strong hallucinations are simply impossible!”
Does this author honestly expect that anyone with a university degree ought to draw this conclusion from this gibberish? The encapsulation of what Dennet has said is basically this: “Hey look- this is complex stuff here- it involves whatchacallits and thingadangles! What’s more there are really really lots of them to build! And it has to be really really fast! Our biggest computers can’t even do it today! Look how bad computer games are! So take it from me, this is a really really big problem.” His inferences based on Donkey Kong would be perhaps forgivable, being ignorant of advanced concepts in computer science. Apart from his poor information, what was exceptionally disappointing was the frailty of the logic being used to form his inferences and conclusions. The proposition was laced together with a pattern of thinking so tenuous that I felt embarrassed for the writer, which quickly gave way to anger for the arrogant pomposity in asking someone to pay money to read such idle thinking.
Dennet bases his conclusion on a startlingly clumsy leap which is in essence: “Since this task is so hard, it is insoluble- take it from me- a philosopher, not a computer scientist.”
To begin with, Dennet does not grasp how
computer modeling works, falling into the neophyte's confusion between simulations and virtual reality. He seems to think that simulations used for predicting nuclear blast yields are programmed in a canned fashion as are virtual reality computer games.
Such simulations cannot be canned because such short cuts would defeat the very purpose of computer models. Instead, such simulations are dynamically generated. The level of confidence that scientists must have in these simulations is very high- the lives thousands of people depend on correct and exceptionally complex fluid dynamics modeling of wind flows around objects and how the complex of structure of a skyscraper interacts with such flows. The lives of millions rely on correct prediction of whether a nuclear device will function as designed or not.
Technically, there are many approaches for implementing such simulations but to illustrate how elementary an error it is to assume a computational model where there is a combinatorial explosions of predicted cases for event outcomes, consider for example
constraint based approaches. These models generate simulated phenomena based on large numbers of primitives that reflect particular physical laws. Naturally it would be a monumental task to build such a model that included interactions between simulated objects at fine levels of granularity obeying all laws that typical individuals understand. Admittedly, the computer processing required for the calculations might necessitate a massively parallel design for the simulation to be able to run in real time.
Bayesian evaluation might be used in place of boolean rule logic.
Clearer exposition of practical approaches would be possible except that Dennet leaves his crucial point undescribed- what exactly is it that Dennet believes presents the insoluble problem? Is he expecting the reader to so flummoxed by these vague hand waves about complexity that the reader will simply give up and take the professor's word for it?
The engineering requirement is that the simuluation encode all known physical laws and allow multiple agents interact with that simulated world.
Some Computer Science engineers might object to this exposure of Dennett's vacuousness in this way: “Well sure, the set of physical laws is closed and therefore as a finite domain it can in principle be encoded, but there may be hard limits to future computation capacity. There may be laws of physics which limit future ability to compute such massive models in real time.” But this notion doesn’t even pass muster as a frail objection.
Firstly, the simulation is self limiting. We know the laws of the physical world from what we learn of it. If there were “real” laws too complicated for the evil scientists to compute, then how would we know about these “real” laws? Things behave as “our science” observes them in the simulation to behave. Within the simulation, the person has no external point of reference, so the evil scientists are free to leave out any laws that are too hard to simulate.
Secondly, even if such a hard limit on computational power were known, it is based on a feeble understanding of how we perceive time. Each microsecond of simulated phenomena could take a millisecond, and the human mind would not know because we measures time in terms of the pace that events take place. If the events were to take place at a slower pace, by what clock would we be able to detect that the pace of events “was too slow”?
Sorry, but the model that science has of the world reduces every phenomena to a small number of fundamental laws. The evil scientists do not have the hard problem of simulating entities with free will and consciousness- (such simulations may well be computationally intractable). But correct modeling of consciousness is outside the problem domain- conscious actors can exist by rigging up other brains in other vats.
The task of representing a physical laws in a global sized simulation may be large since it would have to encode our immense knowledge of the rules of the mechanisms of nature. But because these rules represent a finite domain, and because we have the option of punting any computationally expensive rule from the model, the problem is without doubt a tractable one. Our knowledge though comprehensive and though endowed with great predictive power is in fact a finite domain, and the rules though intricate and highly interconnected form a closed system.
On a closing historical note, the idea of the Matrix was predated by the brain in a vat thought experiment which was itself just a modernized version of Descartes experiment which led to his famous conclusion, “I think, therefore I am”. It is unclear to me what purpose Dennet has in attacking the validity of this thought experiment. Though I do not believe we are Brains in Vats- that neither God nor some other entity controls our perceptions in this way, Dennet’s argument has no force and we are forced to admit that the scenario is factually possible. Dennet’s follow on to his false conclusion is that “Strong hallucinations are impossible”. If he bases this belief on this line of reasoning, he is equally mistaken. The shoddy work this writer has done convincing himself of this argument was an exceptional disappointment to me. I would have thought that modern day philosophers had better training in logic and skills in research than this.
* [my italics]