Am I a brain-in-a-vat?

The brain in a vat hypothesis is an example of radical scepticism, and is a contemporary reworking of Descartes’ evil demon argument. It may have no practical significance for me (even if true), but it does challenge us to consider and justify what we think we know about the external world, and so is a useful exercise.

The sceptic’s challenge is that, for all I know, I might exist only as a disembodied brain floating in a vat of nutrients. The brain is connected to a supercomputer that is producing electrical impulses that are stimulating it in an identical way to those that would arise from normal perception. So, when I currently think that I am sitting in my office typing on my computer, I actually exist only as a brain that is sitting in a vat and being fed the inputs for all of my perceptions of the world around me. Is there any way that I can answer the sceptic’s challenge?

Firstly, we should note that arguing that I cannot be a brain in a vat, as I can experience the external world in all of its rich texture (and I have detailed memories of it) gets me nowhere, as the hypothesis states that the inputs to my brain are identical to those that it would receive from my senses if I was really experiencing the external world, and therefore the two scenarios are indistinguishable to me.

One possible solution is to say that the hypothesis is an empty one. If I am actually a brain in a vat, but have no way to ever verify this, then it makes no difference to me, so it is as if I really do exist in the world. In this sense, it is a bit like having an invisible and intangible elephant in my garage – it is no different to having no elephant at all. However, there is a crucial difference here that makes the analogy false. At the moment the brain in a vat hypothesis makes no difference to my existence, but if it is actually true then there is the possibility that it might make a difference in the future (the inputs might be changed radically, or stopped completely, for example). So, in that sense, the brain in a vat hypothesis is not an empty one.

Another solution might be attempted by appeal to Occam’s razor. I might venture that the hypothesis that I am a brain in a vat is less parsimonious than the hypothesis that I really do experience the external world as I believe it to be. However, my idea of which hypothesis is the more parsimonious is based upon my experience of the world. If this experience is illusory, and the external world doesn’t exist as I seem to experience it, then my real world hypothesis may be the less parsimonious one. Having said that, there is an intuitive sense in which the brain in a vat hypothesis contains more ad-hoc elements. In this scenario, my experiences need to be based upon something – either upon some reality within the world in which my brain is en-vatted, or else pure fabrications (or some combination of the two). If the former, then some version of the world that I experience does exist somewhere but, for some unknown reason, I exist in it purely as a brain in a vat. If the latter, then it would seem to require a great deal of effort in order to fabricate an internally consistent world of the complexity that I experience, with no obvious reason as to why this should be done at all (note though that the simulation needn't be of an entire universe, as I don't actually experience the entire universe - only a very small subset of it). In both cases, intuition suggests that these options are less parsimonious than the hypothesis that the real world just exists with me in it.

A number of semantic responses have been attempted (by Putnam and others) along the following lines:

P1: If I am a brain in a vat, then my word ‘tree’ does not refer to trees.
P2: My word ‘tree’ refers to trees. So,
C: I am not a brain in a vat

However, all such responses appear to beg the question. The whole point of the sceptic’s argument is that I don’t know that my perception refers to anything in the real world, so P1 and P2 above are not both true (and might be both false). My word ‘tree’ may or may not refer to actual trees, if they exist, but I cannot deduce anything significant from that. For example, imagine that the brain in a vat world is identical to our world, with the exception that the expertise and technology exists to put brains into vats and feed them inputs that are indistinguishable from external perception in the BIV world, and that ‘I’ am one of those brains. In this case, there is a sense in which my word ‘tree’ does refer to trees (the stimulation of my brain is identical to the stimulation that would come from perceiving a real tree, as real trees do exist in the BIV world). In that case, P1 is false, and P2 is true, so the conclusion does not follow. There is also a sense in which my word tree does not refer to trees, since I have never actually seen a ‘real’ tree. In that case, P1 is true, but P2 is false, so again the conclusion does not follow.

In the end, I think that I am warranted in believing that I am not a brain in a vat - even though I cannot prove it. The most parsimonious explanation for all of the evidence that I observe is that I actually do exist in the real world. Any other 'fake world' hypothesis, including the brain in a vat hypothesis, requires me to posit some inscrutible other world in which some agency or agencies exist that have the means and motivation to create a fake world for me. This has to be less parsimonious than the real world hypothesis.

Furthermore, based upon my experience of my world (whether real or fake), the most rational course of action is to attempt to achieve my goals (happiness, for example). Whether my world is real or fake, this course of action is the same. There may be no way, even in principle, for me to determine whether I am a brain in a vat or not, but since the course of action is the same in both cases, it makes no difference to me now (although that might change in the future).

Do some people know the moral rules by which God wants us to live?

Some theists believe that God has laid down a set of moral rules detailing how we should live our lives and, further, that they know what these rules are (I will restrict this discussion to the Judeo-Christian god). This idea that morality is dependent upon God is known as the Divine Command Theory of Morality (as opposed to Divine Essence Theory, which I will not be discussing). But do these people really know what they claim to know? They might believe what they claim, but do these beliefs count as (propositional) knowledge? This is an important question for, if we have good reason to think that the beliefs are true then we should all wish to know this, as the belief system entails that eternal suffering (or, at least, the failure to achieve eternal happiness) may await those who don’t accept these beliefs and live by these rules. Apart from that, if such a wise moral system exists, then we should all wish to live by it. Conversely, if we have good reason to think that the beliefs are false, then the believers should wish to know this, as they otherwise risk spending their lives striving towards bad goals, and not striving towards good goals (by good goals, I mean the goals that they would strive for if they had the correct factual knowledge about the universe, and acted rationally based upon this knowledge; by bad goals, I mean goals that they would not strive for if they had the correct factual knowledge about the universe, and acted rationally based upon this knowledge).

How might we determine whether these beliefs can be described as knowledge? We need to determine the necessary and sufficient conditions for something to be classified as knowledge, and then establish whether the beliefs in question meet these conditions. Most epistemologists would agree that knowledge requires at least true belief. So, for example, for me to ‘know’ that the Earth orbits the Sun, then I must believe this proposition and, further, it must also be true. However, whilst belief and truth are necessary conditions for knowledge, they are not usually judged to be sufficient conditions, as this would otherwise allow lucky guesses to be included as knowledge. Intuitively, it makes sense that we should need some reasonable justification for our true beliefs in order for them to qualify as knowledge. There is a sense in which acquiring knowledge, as opposed to mere true belief, requires some work or achievement on our part. For example, if I believed in advance that a tossed coin will come up heads, and it did, then I could not legitimately claim this as knowledge, unless I could somehow justify my true belief (for example, if I knew that the coin was a biased one that was almost guaranteed to come up heads).

However, this tripartite definition of knowledge as justified true belief has a serious flaw, as pointed out in the notorious counterexamples of Edmund Gettier. I might believe some true proposition, and I might be able to justify my belief, and yet it might still not be real knowledge. For example, in the coin toss case I mentioned above, if I think that a biased coin is being used, and therefore have a justified true belief that the coin will come up heads, then I would seem to have knowledge. However, it could be that unbeknownst to me somebody switched the coin for an unbiased one, which came up heads by chance alone. In this case, I could not really be said to ‘know’ that it would come up heads.

The definition of knowledge as justified true belief also has an additional challenge, in that it needs to answer a regress problem in the series of justifications for a true belief. If I believe some true proposition P, and I justify it by means of justification X, then what belief justifies X, and so on? The three possibilities are that we have an infinite series of justifications; that some of our justifications are circular (i.e. they are used to justify each other); or that we finally reach a point in our series of justifications where we have basic beliefs that are in need of no further justification. Each of these possibilities presents its own particular difficulties. It is hard to see how an infinite or circular set of justifications can be used as a solid foundation for knowledge, and it is not at all clear that we can have any basic beliefs that are at once indisputably self-evident (e.g. 2+2 =4) and yet can act as a foundation for all other knowledge. For example, Ayer postulated in his book Language, Truth, and Logic that from tautologies we can deduce only other tautologies, and that there is no such thing as a true a priori synthetic statement (we can only attempt to verify it by observation).

I believe that the way around these objections is to accept that there are properly basic synthetic beliefs that act as a foundation for all of our empirical knowledge. These beliefs are those of our direct experience - since it is undeniable at those moments that we are experiencing some cognitive sensation, even if we might be wrong about what that sensation is. From these properly basic beliefs, we can create our chain of justifications for the rest of our empirical knowledge. The other required element in our theory of knowledge is that each justification in our chain should be the result of a reliable process. That is, knowledge is justified true belief where each justification is the result of some reliable (i.e. truth-conducive) mechanism (and our chain of justifications ultimately rests upon foundational properly basic beliefs). So, for example, I can be said to know that there is a computer in front of me if this belief is true and if I can justify this by some reliable mechanism (e.g. if I can see it, and I can justify the general reliability of my visual perception through a series of justifications that ultimately rest upon my properly basic beliefs). I thus avoid the need to produce an infinite or circular series of justifications for this belief. Now, reliabilism, as this concept of using reliable mechanisms is called, is open a range of possible objections, including Gettier ones, but it still seems to be the most promising of the possible alternatives.

So, armed with our definition of knowledge, we can now look at the Divine Command Theory beliefs of the theists (note that there are a number of objections to this theory, including Euthyphro’s dilemma, which I will not examine here). If we unpack the Divine Command Theory of morality, then we have the following propositions:

1. God exists (and is usually defined as being omnipotent, omniscient, omnibenevolent, immaterial, the uncreated creator of the universe, and having a particular interest in humans).
2. God has a set of moral rules that humans are required to obey.
3. God has communicated these rules to humanity.

Note that propositions 2 and 3 are dependent upon the truth of proposition 1, as they each assume God’s existence (or, at least, that God has previously existed and communicated his moral rules to humanity) – if this is false, then propositions 2 and 3 are necessarily false. What are we to make of these? Ayer would suggest that these propositions are literally meaningless, as they fail his Verification Principle. He would maintain that the proposition that God exists is not true by definition and, further, its truth or falsehood cannot be determined (even in principle) by any observation. However, if we ignore this objection for the moment, and grant that the proposition that God exists is meaningful, then what can we say about these three propositions?

Making use of the analysis of knowledge given earlier, can the belief of these three propositions be considered to be knowledge? In this case, and unlike my example of the coin toss (where I might have a lucky true belief), the actual truth of the beliefs is not known, so we haven’t necessarily satisfied the truth requirement of knowledge. All we have so far is belief. However, since the reliability criterion itself will tend to maximise true beliefs whilst minimising false ones (assuming the validity of inductive reasoning), can we at least determine whether these beliefs have been reached by some reliable mechanism? If so, then this will point towards their truth or, if not, then towards their falsity. Furthermore, if the mechanism can be shown to be an unreliable one, then these beliefs will not count as knowledge, even if they might be true. In that case, they would at most be mere true belief, and would thus lack the stability of real knowledge.

Attempts have been made to prove Proposition 1 by analytic means alone – for example, the Ontological Argument. However, this argument is open to a number of objections, and few philosophers are convinced of its soundness (even if it is difficult to pin down exactly where it goes wrong). Theists might also seek to establish the truth of proposition 1 by means of one or other of the cosmological, teleological, or moral arguments for God’s existence. However, all of these arguments have been convincingly rebutted, most can anyway only argue for some inscrutable universe-creating entity as opposed to the specific Judeo-Christian God, and none successfully supports the additional elements of propositions 2 and 3 . In particular, since we are interested in determining the reliability of the theists’ belief-creating mechanism, it should be noted that few theists form their beliefs based upon these arguments anyway (although they may trust that some of these arguments are sound). Rather, in order to establish propositions 2 and 3, theists will typically resort to the evidence in the Bible, to what their religious leaders (or other people) tell them, or to their own perception of religious experience. I will examine each of these in turn to determine if they constitute reliable, truth-conducive mechanisms.

Firstly, the Bible. In general, I would contend that forming empirical beliefs about the world based upon statements contained in ancient historical documents is liable to be unreliable. Due to the many known examples of credulity, exaggeration, and fabrication in historical documents, and to the lack of knowledge of the world possessed by people in these ancient times, we must be very careful about granting too much credence to the contents of these documents. Descriptions of events are always questionable unless they can be corroborated by relevantly similar descriptions in a other sources – preferably those already known to be reliable. Additionally, in the case of contentious elements, we need to consider whether the events or things described seem plausible. In the particular case of the Bible, we have a number of serious problems:

• We have accounts that were written many years after the events themselves were supposed to have happened


• We have inconsistent and contradictory descriptions of the same events


• There are other supposedly revealed religious texts that differ in significant ways from the Bible, and yet God's message is held to be universal and revealed the same to everyone


• We have little or no corroboration in any other sources


• We have methods that are known to introduce unreliability - word of mouth, geographically dispersed multiple authors, editing and compilation that is arbitrary or that has an agenda, elements that were revealed in dreams etc.


• On top of all of this, we have a number of extraordinary metaphysical claims about supernatural entities (including God) and events. Some of these claims are in opposition to our current reliably formed knowledge about how the universe works.

In light of this, to take the word of the Bible when it comes to extraordinary metaphysical claims about God and his supposed moral rules for humanity would clearly seem to be an unreliable mechanism for generating true beliefs.

Appeal to religious leaders or other people fares no better than appeal to the Bible as a reliable mechanism for generating true beliefs. The first problem when it comes to relying on such testimony is that the set of statements made by such people is partially or wholly contradictory, so choosing to believe any particular statement or set of statements on this basis alone would seem to be purely arbitrary. Generally, we can accord some measure of reliability to testimony if the statements come from someone who has demonstrated that they are an expert on the topic concerned. However, in the case of religious leaders, the statements that they make generally have no means by which their truth or falsity may be determined. We have no independent yardstick against which to judge the reliability or otherwise of their metaphysical statements, so no determination can be made. Hence, we have no good reason to judge this as a reliable truth generating mechanism, and plenty of other reasons to judge it otherwise (some statements contradict reliably formed knowledge, and many statements are in contradiction with each other etc.)

How about people’s religious experience – is this reliable? Does Calvin and Plantinga’s hypothesised sensus divinitatis constitute a reliable mechanism through which to acquire true beliefs? I would suggest that the simple answer is no. Firstly, any divine messages and revelations that people claim to have received by such means are culture-specific, inconsistent and contradictory. Secondly, any such messages seem to consist of nothing more than banalities or vague and unverifiable metaphysical notions. Thirdly, we are aware of many alternative explanations for such feelings of the transcendental –dreams, hallucinations, power of suggestion, psychotic episodes etc. Plantinga would have us take such experiences at face value (as basic beliefs), in the same way that I would do when I perceive an object in front of my eyes. However, the difference between the two cases is that we have no evidence of the reliability of the sensus divinitatis, as it is unverifiable, whereas we have lots of mutually-reinforcing evidence of the reliability (most of the time) of our senses. Furthermore, as the beliefs acquired through the supposed sensus divinitatis are sometimes mutually contradictory, we know that some or all of them must be wrong. Hence, it would clearly seem to be an unreliable mechanism for generating true belief. This also confutes the idea that such a hypothesised sensus divinitatis could produce properly basic beliefs.

Based upon the foregoing, I would conclude that the theistic belief that God has communicated to us a set of moral rules by which we must live has been generated by unreliable mechanisms. Unreliable mechanisms tend to produce false beliefs, so this belief is likely to be false. However, even if it is true, this belief does not constitute real knowledge, but mere true belief. As such, it lacks the stability and robustness of real knowledge.