(This is work in progress, and updates are likely. Comments, criticisms and suggestions welcome.)
Philosophers regularly use complex virtual machines (not virtual realities) composed of enduring interacting non-physical subsystems (e.g. operating systems, word-processors, email systems, web browsers, and many more). Yet, almost all ignore (or misdescribe) these VMs when discussing functionalism, supervenience, multiple realisation, reductionism, emergence, and causation. Such VMs depend on many hardware and software designs that interact in very complex ways to maintain a network of causal relationships between physical and virtual entities and processes. I'll try to explain this, and show how VMs are important for philosophy, in part because evolution long ago developed far more sophisticated systems of virtual machinery (e.g. running on brains and their surroundings) than human engineers so far. Most are still not understood. This partly accounts for the apparent intractability of several philosophical problems. E.g. running VM subsystems can be disconnected from input-output interactions for extended periods, and self-monitoring can be unreliable.
For details see: http://www.cs.bham.ac.uk/research/projects/cogaff/09.html#vms
INTRODUCTION: MORE OR LESS ABSTRACT ENTITIES AND PROCESSES
This paper is about the importance, for philosophy of science, philosophy of mind and metaphysics, of the notion of virtual machine (not to be confused with virtual reality systems) that has been developed over the last 60 years or so, but has been almost completely ignored or misunderstood by philosophers.
A rare exception is John Pollock, though I think his analysis deals with only a subset of types of virtual machine:
J. Pollock What Am I? Virtual machines and the mind/body problem, In Philosophy and Phenomenological Research., 76, 2, pp. 237--309, 2008,
downloadable here. (PDF)
I apologise to all the philosophers who are exceptions, but not mentioned below.[*]
For centuries, philosophers have pondered questions about different kinds of entity and their relationships. At one extreme, there are abstract, unperceivable, non-spatial, non-temporal entities such as numbers, sets of numbers, and at another extreme there are perceivable and manipulable types of physical entity such as mud, trees, rocks, mountains, rivers, billiard cues, and human bodies.
Between the two extremes are relatively abstract things that somehow seem to be related to physical objects, and to particular places and times, without being physical, including abstract entities such as minds and their contents, and social phenomena, illustrated by poverty, fashions, the moral standards in a community, legal systems and economic processes.
Moving in a different "direction" from perceivable and manipulable physical entities are the unperceivable, though sometimes indirectly detectable and manipulable entities referred to in scientific theories that seem able to explain some or all of the observed physical phenomena, for example, sub-atomic particles, electric charges and currents, magnetic fields, electromagnetic radiation, other sub-atomic entities, states and processes, and more recently genes, which some people regard as patterns in chemical structures.
In short there appear to be different aspects or components of reality that vary in kind of abstraction, in how large a space or time period they occupy, in how amenable they are to being perceived, physically manipulated, and measured using physical devices, and in what they depend on for their existence and what things depend on them for their existence, with timeless non-spatial entities (e.g. numbers, theorems, geometrical shapes, propositions) at one extreme, and the fundamental physical constituents of the universe (relative to the best available explanatory theories) at the other extreme.
There are several related philosophical questions that arise about most of these perceived or postulated entities including:
(a) Which sorts of entity can, or cannot, exist independently of which other sorts?
Is there an ontological dependency hierarchy?
(b) What exactly are the relationships between those that depend on something else for their existence and whatever they depend on?
(c) Which of the types of entity are capable of being causes or being involved in causing other things to happen?
(d) What sorts of causal interactions can occur between the entities of different sorts?
In particular, are there some entities, events, or processes that can be caused but cannot themselves be causes of anything (epiphenomenal entities)?
(e) In what ways can the various sorts of entities be combined to form more complex entities?For example, sentences can be parts of larger sentences or parts of arguments, stories, or explanations. A physical machine can be assembled spatially from parts that are juxtaposed or connected by intervening parts. The running of the machine over an extended time period can form a process that is made of many spatially and temporally combined sub-processes.(f) Which of the types of entity can usefully be referred to in explanatory scientific theories?
Some of the questions do not arise for some types of entity. For example question (c) does not arise for the timeless unchanging entities such as numbers, shapes, and perhaps also concepts and propositions, because they are timeless cannot change or be involved in processes of change, so they cannot be directly involved in causation. Neither the number seven nor the mathematical function addition, is capable of causing anything to happen, though we may wish to mention both in explaining how some human mental processes cause things to happen: a person can go through a thought process of multiplying the number seven by the number thirty, and that might cause the thought of another number to occur, or something to be written down.
Other abstract entities such as poverty, or a change in fashion, or discovery of evidence supporting a new theory about what causes stomach cancer can cause abstract events and processes to occur, such as a desire to earn a higher wage, a decision to go shopping for new clothes or a government passing laws that make the sale of foods containing some substance illegal. There can also be physical effects such as humans going to college, money changing hands, products being taken off shelves, and the behaviours of neurons being modified.
RUNNING VIRTUAL MACHINES: A NEW KIND OF ABSTRACT ENTITY
In the last half century a new unobservable, yet scientifically respectable, type of entity has come into existence, namely virtual machines running on computers and the entities, events, and processes within virtual machines. Although most philosophers nowadays interact with many such virtual machines every day (e.g. running word processors, mail programs, internet browsers, and many more) they mostly regard them as not being worthy of any philosophical study -- though there are some exceptions, philosophers who acknowledge that virtual machines are important. However, I am not sure that they have noticed all the ways in which VMs are important.
As for the other philosophers, apparently the vast majority, it is as if the fact that you can buy something in a shop, and it is a commonplace object, implies that nothing philosophically interesting can occur when it is used.
There are facts about running virtual machines that I conjecture are not only philosophically interesting, but also capable of providing new insights regarding some of the oldest philosophical problems, including problems about the relations between mind and body, and problems about what can cause what, as well as suggesting new research questions in psychology, neuroscience, biology and studies of the evolution of brains and minds.
These seem to be special cases of something more general, for example other examples of causal interactions between abstract processes, or between abstract and concrete physical processes. Examples of such causal relations include: the relationship between socio-economic phenomena and psychological phenomena, or between ecosystem phenomena and the behaviours of individual organisms, or between meteorological phenomena (e.g. tornadoes, large scale weather processes) and the behaviours of individual molecules. It could even be said that the relationship between chemical processes, e.g. autocatalytic cycles and physical processes are of that kind.
DIFFERENT SORTS OF LEVELS
Although the "levels" terminology is used in several different contexts, it seems clear that there are importantly different cases, of which a few can be expressed in a table of types of supervenience.
However, from a mathematical viewpoint, both Euclidean geometry and the real (continuous) number system can be considered a separate type of structure which can be studied in its own right. So this is not an example of a kind of ontological dependence, as the others seem to be.
Some features of this notion can be roughly indicated by the differences between the first two images below, and the third:
These two figures indicate, crudely, (a) simple property or event
supervenience, and (b) multiple property or event supervenience.
On these models all causation goes in one direction, although some philosophers would treat some of the examples as cases of identity, not causation: the supervening properties or events are sometimes thought of as identical with their physical bases -- merely viewed or thought about in a different way that omits many details. These conceptions of supervenience tend to assume a simple mapping between portions of physical reality and the things that supervene on them.
This roughly represents what I mean by virtual machine supervenience: what supervenes is a complex collection of causally interacting entities, and some of the events can also cause physical changes to happen, e.g. in computer memories, computer displays, or other devices connected to a computer. The existence of such virtual machines depends on tangled networks of causation (with multiple true counterfactual conditionals at any time) linking entities, events and processes at different levels of abstraction in ways that depend on a host of developments in digital electronics and also computer science and software engineering. The entities, events and processes in the virtual machines have much coarser granularity than the entities, events and processes in physical machines and there need not be well-defined portions of physical machines corresponding to the virtual machine components. Moreover, for reasons related to the theory of reference expounded in P.F.Strawson's 1959 book Individuals: An essay in descriptive metaphysics, and for other reasons, the implementation of some aspects of the virtual machinery can depend on things in the environment, including other intelligent systems (also discussed in The Computer Revolution in Philosophy (1978) e.g. in section 6.4 and Chapter 7 e.g. section 7.7
The conjecture is that similar things (virtual machines supervening on physical mechanisms) were produced by biological evolution long ago, some of which are far more complex than anything so far produced by human engineers.
Why did this happen? Because evolution "discovered" the same problems relating to complexity of control and monitoring, and the need to reduce complexity (especially of search spaces) by using abstraction.
For more details see
What Cognitive Scientists Need to Know about Virtual Machines
Virtual Machines in Philosophy, Engineering & Biology (presentation at WPE 2008)
MODELLING THE MIND-BRAIN RELATIONSHIP
It is now fairly common to draw an analogy between two relations, namely
HAVE MANY PHILOSOPHERS NOTICED?
As a result of informally interrogating a diverse, though not very large, sample of philosophers I have come to the conclusion that very few of them are aware of the existence of virtual machines even though they interact with many of them regularly, when they use computers, to compose papers, reader send email, browse the internet, run spelling checkers, use spreadsheets, or play computer games. This even applies to philosophers who worry about the mind-body relationships, write and talk about problems related to supervenience, realisation, or emergence, or who study causality.
My claim is that the concept of a running virtual machine (RVM) developed in computer science and software engineering over the last half century has many interesting features that are directly relevant to a number of old philosophical problems, many of which are easier to deal with in connection with RVMs than with naturally occurring systems, such as minds and brains, because RVMs are things we have designed and which, though increasingly complex, we (collectively, though not individually) understand well, unlike naturally occurring systems such as minds and brains.
At present different parts of the story are understood by different theorists, engineers of various kinds and users. The parts need to be brought together, in something like the way Newton unified disparate phenomena, except that this case is much more complex and difficult. There is no existing branch of mathematics that can play the role that differential and integral calculus played for Newton.
MACHINES CAN SUPERVENE, NOT JUST PROPERTIES AND EVENTS
A key idea is that whereas most discussions of supervenience have talked about property supervenience, state supervenience or process supervenience, there is a deeper, richer, notion of 'machine supervenience', where what supervenes is a machine, i.e. a complex entity made of enduring, asynchronously causally interacting parts, that supervenes as a whole on a very different kind of machine, the implementation machine (e.g. a digital electronic machine, or network of such machines, or a brain) plus environment usually, where the implementation machine is also made of enduring, causally interacting parts, but parts of quite different kinds, with quite different properties, obeying quite different laws, and usually with a very much lower level of granularity, and where the mappings between parts of the two machines can be dynamically changing, and may depend on intermediate virtual machine layers.
I argue that the problems discovered by engineers and (to some extent) solved by creating a succession of increasingly sophisticated types of virtual machine over the last six decades were probably 'discovered' long ago by biological evolution, which has produced far more complex biological virtual machines solving design problems that overlap with the engineering problems, including problems of control, self-monitoring and self-modification.
All of this has to be connected with analysis of the concept 'cause', of course. A virtual machine and its physical implementation machine can both include causes, and can interact causally with the environment in which they are embedded. This leads to apparent paradoxes (about redundancy of causes) that have to be defused by an appropriate analysis of 'cause' (based on truth of counterfactual conditionals).
There are also very deep difficulties about how to test scientific theories about virtual machines that we have not designed ourselves. But that is a topic for another occasion.
I find it very strange that most modern philosophers ignore all this, despite making regular use of sophisticated virtual machines every day, e.g. operating systems, spelling checkers, spreadsheets, word processors, internet browsers, email systems, online shopping services, etc. It is particularly curious that the notion of a virtual machine is not mentioned in the online Stanford Encyclopedia of Philosophy entry on "Supervenience", which depends for its existence on the operation of a host of interconnected virtual machines.
A related paper on the importance of virtual machines for cognitive science:
What Cognitive Scientists Need to Know about Virtual Machines
To be presented at the CogSci'09 Conference in Amsterdam 29th july-1st Aug 2009.
This paper will overlap with that one, but is aimed at a different audience, philosophers rather than cognitive scientists.
Related online documents:
What are virtual machines? Are they real?
Virtual Machines in Philosophy, Engineering & Biology (at WPE 2008)
(also on slideshare using 'flash'.)
Kenneth Craik's little book
The Nature of Explanation, CUP, 1943, anticipated some of
these ideas I think.
Reading it long ago influenced my own thinking.
[*] I think their rarity is due to a phenomenon that is characteristic of much philosophical teaching, research, and discussion: it refers mainly to the writings of other philosophers, mostly ignoring the rest of the world. And again I must emphasize that there are many exceptions to this generalisation (e.g. see the contents of The British Journal for the Philosophy of Science).
There may be many more exceptions than I have encountered in my attempts to sample philosophical discussions, both published and in real life. I hope there are, and that I am wrong about their rarity. And if they are rare, I hope improved philosophical education practices will change things.