|The Holodeck - Star Trek|
There is an apple in front of me. I can see it, but I can’t touch it. The reason is that the apple is actually a 3-D rendered model of an apple. It looks like an apple, but exists only within a virtual environment — one that is projected onto the computer screen in front of me. I can interact with the apple. I have an avatar that I can control on the screen. That avatar is a virtual projection of my self. It can pick up the apple, throw it around the virtual room, or eat it. But I can’t touch it and interact with it using my own physical hands.
Is the apple real? Of course not: it’s virtual. But are virtual objects (or events or states of affairs) ever real? This question is of considerable importance. We already live a considerable amount of our lives online (in ‘virtual’ worlds). We interact with people virtually. We deal in virtual goods and services. And if the prognostications of technological enthusiasts are anything to go by, we will probably live more and more of our lives in virtual worlds in the future. With the emergence of immersive virtual reality (VR) headsets such as the Occulus Rift, the Samsung Gear, and Sony Playstation VR, we can now participate in highly realistic and engaging virtual activities. It would be nice to know whether any of these qualify as being ‘real’, particularly given that the technology is marketed to us as virtual reality.
The reality (or unreality) of the virtual is, fundamentally, a philosophical question. And, fortunately, philosophers have already begun to answer it. The philosopher Philip Brey, in particular, has developed a sophisticated framework for thinking about the reality of virtual reality. He argues that some virtual objects and events are obviously not real (they are merely representations or simulacra), but others are every bit as real as their real world analogues. He suggests that we use John Searle’s theory of social reality to tell the difference.
I want to analyse and evaluate Brey’s proposed framework in the remainder of this post.
1. The Physical and the Functional
To warm up, let’s think a little bit more about the opening example of the virtual apple. As Brey points out, this ‘apple’ clearly exists in some form. It is not a mirage or hallucination. It really exists within the virtual environment. But its existence has a distinctive metaphysical quality to it. It does not exist qua real apple. You cannot bite into it or taste its flesh. But it does exist qua representation or simulation. In this sense it is somewhat like a fictional character. Sherlock Holmes is not real: there was no one by that name living at 221b Baker Street in London in the late 1800s, nor did anyone answering his description solve the crimes that befuddled the hapless inspectors from Scotland Yard. But Sherlock Holmes clearly does exist qua fictional character. There are agreed upon facts about his appearance, habits, and intellect, as well as what he did and did not do qua fictional character.
So Sherlock Holmes and the apple have a simulative reality, but nothing more. They do not and cannot exist qua real apple or real person. Why not? The answer seems to lie in the essentially physical nature of apples and detectives. An apple does not exist qua real apple unless it has certain physical properties and attributes. It has to have mass, occupy space, consist in a certain mix of proteins, sugars and fats, and so on. A virtual apple cannot have those properties and hence cannot be the same thing as a real apple.
The same goes for detectives like Sherlock Holmes. Although there are some complexities there. Human detectives have to have mass, occupy space, and consist in a certain mix of proteins and metabolic processes. But do all detectives have to have these properties? Here we get into one of the great debates in philosophy. It seems to be at least conceivable that there could be a virtual detective that could solve real world crimes in the same manner as Sherlock Holmes. Imagine a really advanced artificial intelligence (AI) that is constantly fed data about crimes and criminal behaviour. It spots patterns and learns how to solve crimes based on this data. You could then feed information about new crimes into the AI and it could spit out a solution. This AI program would then be a ‘real’ detective, not a mere simulation or representation of a detective. In fact, you don’t really have to imagine such a detective. Companies like PredPol are already creating them.
We can draw some lessons from these examples. First, we can see that there are at least some kinds of entities — like apples and human detectives — that are essentially physical in nature. We can call them essentially physical kinds. These are objects, events and states of affairs that must have some specific physical properties in order qualify as an instance of the relevant kind. Virtual versions of these kinds can never be real; they can only be simulations or representations. But then there are other kinds that are not essentially physical in nature. A ‘detective’ would seem to be an example. A detective is a non-physical functional kind: an entity qualifies for membership of the class of detectives in virtue of the function it performs — attempting to investigate and solve crimes — not in virtue of any physical properties it might have. Virtual versions of these kinds can be every bit as real as their real-world equivalents.
Some functional kinds are essentially physical in nature. A lever is a functional kind. A wooden stick can be counted as a ‘real’ instance of a lever in virtue of the function it performs, but it can only perform that function because it has certain physical characteristics. Just try lifting a heavy object with a virtual lever — one simulated on the screen of your smartphone. You won’t be able to do it. On the other hand, a spirit level does not require any particular physical shape or constitution. You can quite happily assess the levelness of your bookshelf with a spirit level that has been simulated on the screen of your smartphone.
Furthermore, the term ‘non-physical functional kind’ is something of a misnomer. Objects and entities that belong to that class will have some physical instantiation (after all virtual objects are physically instantiated, in some symbolic form, in computer hardware); it’s just that they don’t require any particular or specific physical characteristics in order to perform the relevant function.
2. Social Kinds and Social Reality
So there are some essentially physical kinds: virtual instances of these kinds can only be simulacra. There also some non-physical functional kinds: virtual instances of these kinds can be as real as their real world equivalents. Are there any other kinds whose virtual instances can be every bit as real as their real world equivalents? Yes, there are: social kinds. These are a sub-category of non-physical functional kinds, which are particularly interesting because of their practical importance and their ontological origins.
In terms of their importance, it goes without saying that large chunks of the reality with which we engage on a daily basis is social in nature. Our relationships, jobs, financial assets, property, legal obligations, credentials, social status, and so on, are all socially constructed and sustained. Brey argues that much of this social reality can be recreated in virtual form. He argues that we can use John Searle’s theory of social reality as a guide to when and whether social kinds can be ‘ontologically reproduced’ (as he puts it) in virtual form.
To understand his proposal, we need first to understand Searle’s theory. Searle distinguishes physical kinds and social kinds along two dimensions:* their ontology (what they are) and their epistemology (how we come to know of their existence). He argues that physical kinds are distinctive in virtue of the fact that they are ontologically objective and epistemically objective. An apple does not depend on the presence of a human mind for its existence — it is thus ontologically objective. Furthermore, we can come to know of its existence through intersubjectively agreed upon methods of inquiry — it is thus epistemically objective.
Social kinds are distinctive because they are ontologically subjective and epistemically objective. Money depends on human minds for its existence. Gold, silver, paper and other physical tokens do not count as money in virtue of their physical properties or characteristics (contrary to what people often believe). They count as money because human minds have conferred the functional status of money on them through an exercise of collective imagination. In other words, particular physical tokens only count as money because most of us agree that they count as money. In theory, we can confer the functional status of money on any token, be it an exquisitely sculpted metal coin or a digital register of bank balances. In practice, certain tokens are better suited to the functional task than others. This is due to their durability and incorruptibility. Nevertheless, this hasn’t stopped us from conferring the functional status of money on virtual tokens. Indeed, most money that is in existence today is virtual in nature: it only exists in digital bank balances; it does not, and never will, exist in the form of notes or coins. We happily pay for goods and services with this ‘virtual’ money, even though it lacks physical tangibility. This virtual money is still epistemically objective in nature. I cannot unilaterally imagine more money into my bank account. My current financial status is a matter of intersubjectively agreed upon fact.
Searle argues that many social kinds share these twin properties of ontological subjectivity and epistemic objectivity. Examples include marriages, property, legal rights and duties generally, corporations, political offices and so on. He calls these ‘institutional facts’. These are social kinds that come into existence through the collective agreement upon a constitutive rule. The constitutive rule takes the form ‘X counts as Y in context C’. In the case of money, the constitutive rule might read something like ‘Precious metal coins of with features a, b, c, count as money for the purposes of purchasing goods and services’. Searle doesn’t think that we explicitly formulate constitutive rules for all social objects and events. Some constitutive rules are implicit in how we behave and act; others are more explicit.
What’s interesting about Searle’s theory is that it means that much of our everyday social reality is, in a sense, already ‘virtual’ in nature. It doesn’t depend on any physical, real world properties or characteristics for its existence. Money, marriages, property, rights, duties, political offices and the like do not exist ‘out there’ in the physical world; they exist inside our (collective) minds. They are fictional projections of our minds over the physical reality we inhabit. In principle, we can project the same social reality over anything, including the representations and simulations that exist within virtual reality. Thus, according to Brey, we can ontologically recreate things like money, marriage, rights, duties, political offices, and so forth in virtual worlds. All it takes is some collective imagination and will.
3. Conclusion: What's real and what's not?
Brey’s view on the reality of virtual reality can be summarised as follows:
Essentially physical kinds: i.e. entities that have some specific physical properties or characteristics can never be ontologically reproduced in a virtual environment; their virtual form can only ever be a simulation or representation (e.g. apples, chairs, cars etc.).
Non-physical functional kinds: i.e. entities that perform functions that do not depend on any particular physical properties or characteristics can be ontologically reproduced in a virtual environment; their virtual form can be every bit as real as their real world equivalents.
Social kinds: i.e. a sub-set of non-physical functional kinds whose existence depends on the collective coordination and agreement upon a constitutive rule (of the form ‘X counts as Y in C’) can be ontologically reproduced in a virtual environment; their virtual form can be every bit as real as their real world equivalents.
Note that this theory covers virtual objects, events and states of affairs. It does not include virtual actions. As Brey points out in his paper on the topic, virtual actions have to be treated differently for the simple reason that virtual actions are typically performed by human controllers of characters operating in virtual worlds. As such, virtual actions can have ‘extravirtual’ origins and effects, and this means that they share a much more fluid relationship with reality than do virtual objects and events. Virtual actions are constantly spilling over into the real world. It would require another post to clarify the exact ontological status and classification of these acts, but suffice to say virtual actions are often every bit as real as real world actions.
For what it is worth, I think Brey’s theory is pretty much spot on. There clearly are some objects and events that require a particular physical instantiation and this can never be recreated in virtual form; and there are also clearly other objects and events that do not depend on a particular physical instantiation (that are ‘multiply realisable’ - to use the philosophical parlance). I also agree that much of our everyday social reality can be recreated in virtual form because it depends for its existence on collective agreement. I think this is an important observation because its consequences could be far reaching. We can certainly quibble about the utility of Searle’s specific theory of how social kinds come into existence, but there is general agreement that much of the social world is constructed by the minds of human actors. (If you are interested in a slightly different theory of social kinds, see my previous post on the philosophy of social construction).
That said, I think that there might an alternative approach to differentiating between the virtual and the real that is overlooked by the theory. I’m not sure that defining everything that is represented or created on a computer as ‘virtual’ captures what we really mean by the term. Indeed, I tend to favour something closer to an exclusionary definition of the virtual. In other words, I would prefer a definition of the virtual that necessarily excludes reality: that holds that the virtual can never be real.
Furthermore, even though I agree with the theory in its current form, I think there will be much disagreement over specific cases. For any particular object or event, people might disagree about whether it requires some essential physical property or characteristic or not. Consider the debate about the human mind. There are some philosophers, called functionalists, who think that the human mind can be realised in multiple different physical forms. It is, consequently, not an essentially physical kind. There are others who think that only a human brain could instantiate a mind. This means that it is an essentially physical kind. We can expect disagreements of this sort to arise over allegedly real objects and events that are instantiated in virtual worlds, even if we agree on the general principles that apply to distinguishing that which is real from that which is merely a simulation. To be fair, Brey recognises this point. One of his main observations is that virtual objects and events tend to exist in an ontologically uncertain/contested state.
* Searle uses slightly different terminology in his work. He distinguishes between brute facts and institutional facts.