Thursday 13 October 2016

The Grammar of Information




Wittgenstein said he preferred to reserve the expression “I know” for the cases in which it is used in normal linguistic exchange. […] If we are to build a theory of information, if there is ever to be a science of information, that, after all, is what we want a theory, a science, of — whatever we, in normal conversation, are talking about when we talk about information. (Dretske 2003)
According to the Stanford Encyclopaedia of Philosophy: "There seems to be no pragmatic pressure in everyday communication to converge to a more exact definition of the notion of information." The following discussion is not intended to offer a more exact definition, but it is intended to present what I think are some important observations about the concept of information that are all too commonly overlooked, ignored or misunderstood.

Communication Dependence

The first of these is that information is dependent upon communication and the skills that mediate between intelligent creatures. To fail to account for this connection is to risk undermining the important distinction we typically assume between communication and causation. To communicate is to share information, usually between two or more parties. Inanimate objects cannot share information. Their influence is causal, not communicative. Nonetheless, the effects of causal influence can be used as information, but only by creatures with some understanding or awareness of cause and effect.


Transmissibility and Communication
In a paper entitled “The Metaphysics of Information” (2003), Fred Dretske argues that information has three “essential properties”, one of which is its transmissibility. On this point I agree, but I think we could probably be a little more specific. Communication always involves transmission but transmission does not always involve communication. Conductive materials transmit heat or electricity, but they do not communicate heat or electricity. Communication is the narrower conceptual category, but it still encompasses all forms of information. It might be rightly argued however, that information about the origins of the Universe is transmitted from distant stars without being communicated by them. The important point to bear in mind though, is that without our culturally acquired communication skills and associated technologies we would have no capacity to treat the light from distant stars as information.


The of/about Distinction

The next “essential property” of information, according to Dretske, is that it is about something and is therefore necessarily semantic: “If it isn’t about anything, it isn’t information.” If this means that nonverbal (i.e. non-semantic) representations do not qualify as information, then we would have good reason to object. A simple but vitally important distinction will help to exemplify the issue. We commonly distinguish between what nonverbal representations are of and what they are about. I can answer the question of what an image (for example) is of by offering another depicted view (by providing different information that is). But I cannot answer the question of what a representation is about in the same way. “Aboutness,” as it is sometimes called, is necessarily semantic, whereas of-ness is not. A more detailed image of what a representational sculpture is of will certainly provide more information, but it will not necessarily provide more information concerning the semantic content of the sculpture, i.e. what it is about. We could document Rodin’s sculpture “The Kiss” in minute detail without providing the slightest information regarding what it is about.


False Information

The last essential property of information according to Dretske, is that it is true. He acknowledges that we sometimes talk of misinformation and false information, but he regards it as “heavy-handed” to conclude that information can be false. He states: “False information is fake information and fake information is not a species of information any more than fake diamonds are a kind of diamond.” If a theory of information needs to take account of “whatever we, in normal conversation, are talking about when we talk about information,” then it makes little sense to reject our ordinary talk of false information, misinformation and misleading information etc. It is also mistaken to suggest that “fake information” will serve as a superior substitute. Fakes are always intended to be fakes, whereas false arguments, for example, are almost never intended to be false.

Unreliable information needn’t always be false information. Information may be adequate in some circumstances and not in others. And even if such information were always reliable in most circumstances, this would only make it contingently true and not universally true.


Effects as Information

According to Steven Pinker (1997) “Information itself is nothing special; it is found wherever causes leave effects.” There is potential to be led astray here. Causation is not in the business of leaving representations in its wake. We can measure and evaluate many of the effects of causal processes, but these measurements and evaluations—not the effects measured and evaluated—are invariably instantiated in the form of representational tokens of one sort or another. It should be clear then, that information concerning the precursors of a certain effect is not to be found like apples lying around an orchard. It can only be found through the application of various skills, tools and techniques, many of which have taken many centuries of experimentation and discovery to reach their current levels of sophistication.


Carrying Information

Pinker continues: “We can regard a piece of matter that carries information about some state of affairs as a symbol; it can ‘stand for’ that state of affairs.” It may not be immediately obvious, but this is a tautology. It is like saying that we can regard a symbol as a symbol. To say that something “carries information” is already to imply that it can be regarded in symbolic terms. When information is transmitted, the process can be described in either of two ways: causal or symbolic. Despite being a tautology, Pinker’s point is correct—a piece of matter can be regarded as a symbolic stand-in for a state of affairs of which it is the effect. That’s what it means to “carry information about some state of affairs”.


The Burden of Information


It is also important to distinguish the carrying involved in information-transfer (or communication more generally) from the carrying of a physical burden. Carrying in the sense that we use when discussing information, is not at all like the carrying of a sack of potatoes or the carrying of a virus. It is more like the carrying of value or responsibility. When we say that a coin “carries” value, it would be absurd to assume that the value inheres, and is therefore detectable, in the coin. Value is conferred upon the coin through a system of exchange that enables the coin to be traded for various goods or services. The carrying of responsibility is likewise, something that finds its expression in actions. To carry responsibility for looking after a friend’s dog, is to be trusted to treat the dog with care and consideration. Burdens of responsibility are not burdens that can be measured by the kilo; they are burdens of expectation.

So when we say that something “carries information,” we mean that it can be treated in representational terms and, by virtue of this treatment, we acquire information either in the form of actual representations or through the ability to produce them.


Representational Utility

In a paper entitled “The Informational Turn in Philosophy” (2003), Fred Adams writes:
Waves of radiation traveling through space may contain information about the Big Bang before anyone detects it. Fingerprints on a gun may contain information about who pulled the trigger before anyone lifts the prints. Thus, information appears to be mind-independent (and, thereby, language independent too).
Just as the utility of objects precedes any use that might be made of them, so too does the capacity for something to be used as information precede any use we might make of it. So, whilst the ways of exploiting the world did not exist before they were discovered, the regularities—that enable exploitation—almost certainly did. Fingerprints on a gun are not representations, but when used as representations about the identity of of a murderer say, they can have very significant utility indeed. It should be noted though, that without the necessary skills and techniques, this utility of fingerprints—and thus any information they might contain—would remain wholly inaccessible. A latent image on a sheet of unprocessed film is not yet a representation, even though it contains information. It clearly follows then, that it is the techniques and processes that we apply to things that draw out their representational utility; their information.


Information Reified

According to the Cambridge Dictionary of Philosophy (Audi, 1999), information is: “an objective (mind independent) entity.” If information is an entity, we could be forgiven for trying to put a finger on it. Sometimes philosophers use the term “entity” to cover concepts (as does Dretske in fact) and this is usually unproblematic. In Audi’s case though, the claim that information is an objective mind-independent entity, would surely make it more substantial than a concept. Representations are entities but a capacity to represent something is not. Skills exist of course, but it is both silly and misleading to suggest that they are entities. If I commit some information to memory, I have not taken an entity on board, I have developed an ability. Abilities are not entities, they are actions that we can perform.


The Principle of Information Proliferation

When we produce information, the way that we do so can also be treated as information. When we make a phone call, the time, duration, origin and destination of the call can all be measured and recorded. This kind of information has recently become known as "metadata" and there is no reason in principle why the measurement and recording of metadata could not itself be measured and recorded. It will be evident then, that, in principle at least, this process could proliferate infinitely. I can think of no better or more persuasive evidence that information is a consequence of the way that we treat things as representational tokens.


The Media of Information

Without its representational utility, information would have no capacity to inform. So when we say that something "carries," "contains" or "conveys" information, this utility is implicit. Information is something we use. To use an object as information is to treat it or respond to it in a particular way or ways that have to be learned. Using information is thus a skill and is reliant upon techniques of communication in which representations function as the fundamental medium of exchange.

I would like to thank the members of the British Wittgenstein Society Facebook Group for valuable input during the preparation of this text.

Saturday 1 October 2016

A Brief Introduction To My Research




The following is an introductory text produced for an exhibition of research staff currently teaching at Gray’s School of Art.

Throughout my career as an artist and academic, I have always been interested in the theory as well as the many practices of representation. During the last 5 years this interest has developed to the point that my research has become wholly devoted to the theory of representation and its wider implications, most especially as it pertains to perception, communication and consciousness.

What is representation? By "representation" I mean precisely what we commonly mean by the term in ordinary language. Representations are stand-ins typically, although not exclusively, used for the purposes of communication. Even political representatives act as proxies for the people whose political views they represent. In short, representations are useful substitutes for the things they represent (although substitution alone is obviously not sufficient for representation).

The means by which representations function might seem to be extremely varied, because they are can involve so many different techniques. But in fact representations can be divided into three distinct categories (vide Donald Brook 1997, 2013), two nonverbal and one verbal. The two nonverbal strategies of representation rely on two importantly different sorts of resemblance, whereas verbal communication relies upon the capacity to accept almost anything as a symbolic substitute for anything else. This is an extremely sophisticated skill, almost entirely restricted to most humans.

It is my view that our prodigious skill in the use of symbolic communication is the result of a long history of tool-use and especially of practices of social exchange in which objects and behaviours become interchangeable due to socially negotiated attributions of value that are ascribed to them. It is this process—this technique—of value attribution that I regard as such a vital factor in the emergence of practices of symbol-use and the ascription of symbolic meaning.

It will be clear then, that I regard symbolic meaning as an exclusively human invention because it is fully dependent upon the capacity to treat objects and circumstances as having significance that is not a quantifiable property of them. When we say that a ball means "playtime" to a dog, we need not be committing ourselves to the belief that the dog regards the ball as a symbol of playtime. We merely mean that the dog knows that the ball usually accompanies playful activity. In other words, we take the dog to be capable of forming associative responses to things and thus to know what they mean in this limited sense.


Returning to nonverbal representations, we find that language enables us to treat these also as verbal constructs, but it does not follow that it is always appropriate to do so; to treat them as messages, signifiers, descriptions or texts. What a nonverbal representation is of is what we might regard as a first-order representational feature. What it is about, on the other hand, is a second-order representational feature and thus depends on a different set of interpretive resources on our part.

There is another common confusion that my research seeks to disentangle. Images, for example, do not resemble things in the same way as models. Models, copies, replicas, reproductions, re-enactments, and exemplification etc. all function because they share features in common with the things they represent. These shared features may be approximate to various degrees, but they are not the result of effects. Images, on the other hand, are only fully like the things they represent in certain ways and under certain conditions. In other words, depictions can sometimes be mistaken for the things they represent, but such mistakes rely to a very significant degree on the particular circumstances of presentation or encounter (i.e. the level and evenness of light, our point of view, our level of attentiveness etc.). This is the basis of many illusions of course.

Finally, to say that an image is a description or that a description is a picture or that a picture is a model or that models delineate the world etc. is to talk in circles. My research is intended to show how the theorisation of philosophers, researchers and sometimes even scientists can go awry when they carelessly confuse, misunderstand or mischaracterise distinct categories of representation.