Monday 19 December 2011

Environmental and Semantic Information: Natural and Non Natural Meaning

Luciano Floridi, in his map of information (see, for example, "Information: A very short introduction" Oxford University Press 2010, or Semantic Conceptions of Information on the Stanford Encyclopedia of Philosophy), has the primary distinction between Environmental and Semantic information.

Floridi's informational map, Source: http://plato.stanford.edu/entries/information-semantic/
This is closely associated with Grice's distinction between Natural and Non Natural meaning (H. P. Grice, "Meaning" The Philosophical Review, 1957, Vol. 66(3), pp. 377-388 http://www.jstor.org/stable/2182440).

I talked about information being data plus meaning in a previous post, which is what Floridi refers to as "The General Definition of Information (GDI)", so we see that the two categories of information derive directly from the two categories of meaning.

Floridi points out that his concept of environmental information might not be 'natural'. Here's how he explains it in the Stanford Encyclopedia of Philosophy (op. cit.):
One of the most often cited examples of environmental information is the series of concentric rings visible in the wood of a cut tree trunk, which may be used to estimate its age. Yet “environmental” information does not need to be natural. Going back to our example, when you turned the ignition key, the red light of the low battery indicator flashed. This signal too can be interpreted as an instance of environmental information. 
Environmental information is defined relative to an observer (an information agent), who is supposed to have no direct access to pure data in themselves. It requires two systems a and b to be coupled in such a way that a's being (of type, or in state) F is correlated to b being (of type, or in state) G, thus carrying for the observer the information that b is G [...]:
Environmental information:
Two systems a and b are coupled in such a way that a's being (of type, or in state) F is correlated to b being (of type, or in state) G, thus carrying for the information agent the information that b is G.
However, I don't think this in any significant way makes it different from Grice's 'natural meaning', or not in any way that matters for my concerns at the moment at least.

My interpretation is that semantic information/non-natural meaning requires intention, whereas environmental information/natural meaning 'just happens'.

Switching to the language of semiotics, a point about signs is that the signifier can be arbitrary. I'm wondering if there's a distinction between, say, natural signs and non-natural signs, where the signifier in a natural sign is necessary, but the signifier in a non-natural sign is arbitrary?  We might interpret the rings visible in the wood of a cut tree as the signifier in a sign meaning the age of the tree (or maybe not, I'm not sure about that), but in that case the rings can't be said to be arbitrary: they are a necessary consequence of the way the tree grows. However the word 'tree' as the signifier for a tree is arbitrary, in the sense that we could equally say "arbre" or "der Baum" (http://translate.definitions.net/tree).

One thing I'm still not quite ready to concede, though, is that non-natural/semantic information/meaning/signs necessarily requires people.

Certainly I want to suggestion that animals can be responsible for semantic/non-natural signs. I want to say that when a peacock displays his tail this is a sign with meaning, that there is intention on the part of the peacock, and that the sign is arbitrary (since other birds manage with different signs meaning essentially the same thing).

Source http://commons.wikimedia.org/wiki/File:Peacock.jpg
I'm still trying to get to grips with what we mean by 'meaning', however, and I'm not sure saying there is a distinction between natural and non-natural meaning gets us any closer to what meaning is. I'm still ploughing my way through Ogden and Richards, 'The Meaning of Meaning' (Routledge and Kegan Paul, 1949) but can't say it has been especially enlightening so far, and has been made a little more difficult by their style and the casual 'of-their-time' racism in one the examples they use to illustrate their argument. Wittgenstein's perspective that I mentioned in the previous post sounds rather more promising so perhaps I need to track that down.

17 comments:

Derek Jones said...

Does information have to contain meaning? Is there a way of thinking of this more like your process diagram rather than a static triangle?

What if we remove meaning by requiring that it is a process of applying value.

Similarly, could we extend this to suggest that data on its own is completely one dimensional and can only be information through process (change, perception, interaction, cause, effect, etc).

Meaning then becomes an extension of the process of data as information (it comes 'after', if you like).

This allows perception of data to be information (meaning) but it also allow data to be information without the need for perception (Wittgenstein would definitely not approve...).

But if we go with this then the universe can continue to increase in entropy - data can continue to interact to provide information whilst we aren't looking...

There's a bit more amateur thoughts on this here (http://designthinkingthoughts.blogspot.com/2011/12/data-information-and-meaning.html) and apologies in advance if I have missed the point entirely :D

Great subject, though :D

David Chapman said...

Thanks for your comments.

You have introduced the word 'value' in here, and I'm not sure I understand what you mean by it.

The words data, information, meaning, knowledge, and maybe also now 'value', seem to keep slipping around and I'm not sure that they can be used in an absolute hierarchy. The semiotic triangle and the trapeziums, however, capture what seems a fundamental three-component relationship/process:

Signifier -> Context -> Signified
Data -> Context -> Information
Information -> Context -> Meaning
and, maybe
Meaning -> Context -> Knowledge

It is tempting to concatenate these to get something like:

Data -> Context -> Information -> Context -> Meaning

but this suggests some sort of absolute concept of data, information and meaning.

I'm thinking instead that the 'information' after the first stage is now the 'data' for the second, so 'information' is only ever relative to something else, the data. And I want to use the word 'meaning' by saying that the context gives meaning to the data in order to turn it into information.

Magnus Ramage said...

I also think that the data-information-knowledge hierarchy is unhelpful, but rather that there's a single process of transformation from non-meaningful stuff (call it data) to meaningful stuff (call it information). The last para of your response here has made me realise though that even the non-meaningful stuff is not subservient to the meaning/context - rather they exist together and each requires the other. Others have called this coevolution or sociotechnical, but whatever the term the key thing is the mutuality of the data & the meaning.

One more for the hierarchy though: Checkland and Holwell (in Information, Systems & Information Systems; but also in Holwell's chapter in Ramage & Chapman 2011) distinguish data from what they call capta. They say that data are the great mass of unsorted facts in the world, whereas capta are the subset of those facts which we choose to find relevant. They've thus inserted another stage into the hierarchy (data-capta-information-knowledge) but to my mind the same principle still applies.

Still in Checkland-related ideas, Derek's comments on value reminds me of Sir Geoffrey Vickers' concept of judgement as an 'appreciative system', the process of attributing value to the world. I think this is basically behind the idea of capta, but I've not seen this said in so many words.

Derek Jones said...

This is where blogs stop working - *this needs a few beers and good chat* ;D

@David - you are quite right, I did slip in a new phrase and you are also *absolutely* right about semantics, in terms of knowing what meaning we are applying to certain terms and the contexts in which they apply (does that count as irony).

By value, I intend (essentially) meaning - whereby there is a relative (additional?) piece of context applied to information or perhaps data. This will usually be consciously applied either by individuals or as a collective agreement. So, 'red' is only information in the sense that a value (name, example, description, symbol) has been applied to it. Without this value it is utterly meaning-less - some cultures have no word for certain colours and this social 'training' defines the value (meaning) for a range of electromagnetic radiation. Without this, it is simply data(?)

In the 'what is red' example, it is perhaps important that there is some sort of fundamental / absolute element (like the capta mentioned by @Magnus). In other words, we accept that, even though some cultures do not recognise green, there is still an electromagnetic frequency (capta) that exists. Or does the fact that we understand EM radiation enough to provide *all* colour capta a meaning without observation?

As mentioned, Wittgenstein would not approve of this at all - for anything to exist a priori or as anything outwith human perception in such a fundamental way would be quite meaningless to him (to the point of scorn...).

The other issue with context is that it (in itself) must also be either data or information and that gets you into a horrible snake eating itself philosophical argument.

But (being a Platonist) what if we do think about this in terms of information entropy and note that as entropy increases the amount of information increases but meaning *decreases*. Is this another way of sorting out a definition? This certainly ticks the 'information being relative to context' box - we can only measure the noise of the system by looking at the relationships of the individual components. But it would mean that information cannot have meaning until it is *observed* - i.e. information has no value in and of itself.

I need a lie down.

Randall Lee Reetz said...

David,

What do you know about basic computation/information theory? A la, Claud Shannon? Signal/cypher? "Information can't be formalized outside of an understanding of ordered data, the machinery to encode/decode it, and the value any such system has as abstraction, i.e. "compression", of reality. The cultural interpretive layers you are trying to falsely insert underneath this causal base will never fly. That is like defining a engine and drivetrain of a Buick by what your dad and your mom did in the back seat one night. There is a huge difference between the causal end of any system's hierarchy of influence, and that same system's effects. Cause and effect. You must learn the difference and respect the difference or you will be forever mystified by "information". It is a very very very good thing that the morphology and mechanism from which information precipitates, is so perfectly agnostic to content ("meaning"). Else, there could be nothing of the direction of time. Else there could be no information. Else the value of all data would be relative and equivalent. Else there could not be evolution of any kind. Yours is a top-down attempt at an understanding of a necessarily bottom-up phenomenon.

Randall Lee Reetz

Randall Lee Reetz said...

The arrogance necessary to initiate a conversation and then to maintain it when said conversation asserts, and is anchored to, the notion that information is somehow dependent upon humans is so far beyond what I am willing to accept as rational that I am ashamed to be of the same species as those asserting such claims. Let us remember (PLEASE) that the universe had to get a hell of a lot of information in order in order to get complexity handling to the point where humans could exist at all! That little exercise in information processing took about 13.4 billion years to accomplish. Should you be looking for "context" to frame your understanding of information or information processing, than I suggest you first put "human context" into the proper "universe context". Cause and effect. An influence hierarchy this large and deep isn't effectible by the Johnny come lately "human" no matter how colossal his ego.

Randall Lee Reetz

Randall Lee Reetz said...

David,

Please expose your motive. At base, what motivates your interest in "information" and "meaning"? What utopia do you seek? What fears do you run from? Hide from? Motive and the exposure of motive is everything. It is the "context" we need to interpret your every missive. So please expose your rhetoric. Lay it bare. Dig down to the quick. Splay it open. I think the very exercise will help you understand something of the complexity of the topics you here so awkwardly broach.

Randall Lee Reetz

Randall Lee Reetz said...

I have asked you to expose your rhetoric. But I really don't need you to. I already know. You are obsessed by and driven to find a way to invert the physical influence hierarchy such that the wants and fears of humans can be seen as universally causal. You want to play god. You are frustrated with physical reality. You seek an idea that will free you from the constraints of causal reality such that you might be able to effect the physical through thinking or feeling alone. You want a free lunch. You want the lies you tell your self to be made manifest in the universe at large. And, importantly, you really don't care what such selfishness might, should it come to pass, do to the very structure of reality. Your own angst is more important than the Universe and how it works. Your existential frustrations du jur, your ugly and embarrassing and petty pains and qualms must take precedence over all of reality! Why? Cause you just can't talk honestly about the source of your frustrations. You can't say, "I am lonely". Or, "I am afraid". Or, "I can't deal with the abuse, neglect, loss, I received/experienced as a kid". And these ignored and neglected authentic feelings have festered in the dark and mutated into unrecognizable facets of your ego and your rhetoric. As such, they are indistinguishable from the information arriving from your physical senses. You might still be able to tell the difference, but you simply can't afford the emotional confrontation such introspection would demand. So you bend an ignorance of theoretical science to fit your needs. Quantum, Information. These are topics rife for distortion. Especially should you remain blissfully ignorant. It is sooo sad. It is soooo common. It is why it is sooo hard for humans to do science.

Randall Lee Reetz said...

Now, for those who CAN see past their own distortion field, lets talk about "information".

1. Information is dependent upon a stable substrate within which data can be written. A sequence of bits laid down is meaningless unless the same sequence exist and continue to exist for extended periods of time.

2. Data must be readable such that the act of reading does not destroy the integrity of the order of that data. Reading always requires physically touching the data.

3. Data can only be written into a substrate that has the potential to hold many different configurations.

4. A mechanism (a cypher) (also data) must be built such that it can come into proximal contact with the data substrate. This mechanism must be capable of writing data, reading written data, and and performing computations based upon the data being read.

5. Said combination of data substrate and cypher must get better and better at building a computable abstraction of the actual causal hierarchy that is the Universe.

6. Writing, reading, and computing upon information are all energy expensive actions. As such, they must compete within the greater energy economy that defines the evolutionary filter that selects towards eventualities within any (all) environment(s).

Randall Lee Reetz said...

… information defined, continued…

7. Information is selected according to its fidelity to the universe it abstracts.

8. A tighter abstraction (more compressed) will demand less energy and yield greater results.

9. information has but one purpose: prediction.

10. Evolution is the process by which the universe hones in on its own prediction.

11. A better prediction increases the rate by which any system finds and moves towards its own eventuality.

12. There is no alternative ending for a universe, only heat death, the only variable is the time it takes to arrive there.

13. It doesn't matter at all what any one entity or species or system thinks of or feels about THE eventuality. But, to the extent that any system's goals are in discordance with the goals of the whole universe, that system will have to spend more energy and will therefore ultimately fail when competing with other systems more harmonious with the universe's largest eventuality.

14. There are no values in evolution or the information evolution generates. Only fidelity to the greatest eventuality, or a lack thereof.

Randall Lee Reetz

David Chapman said...

Randall,thanks for your comments. I hope to respond in a day or two.

Randall Lee Reetz said...

At minimum, any discussion about information should be bounded by and should sit firmly upon the base that is a causal physical definition. If you would there after prefer to bend the universe to fit your own personal rhetorical literary or philosophical perspective, then by all means go for it. But be warned, there is no way to get to an actual understanding of information as a universal phenomena should you begin at the human perspective, building downward. People who genuinely desire understanding don't try to fit the universe to their own agenda, fears, and wants. If you want to understand something, you had better get out of your own way. The universe performed 13.7 billion years of perfectly respectable operations before we humans arrived. What is the likelihood that what the universe did before we arrived had anything at all to do with what we find existentially attractive? If you can't find a path free of your own existential angst, at least show some humility and some respect for basic causal logic. A thing that doesn't exist can't influence anything.

I would love it, just love it, if for once, a philosopher would lead with their own bullshit, would say "I am afraid of death and loneliness and am in constant pain as a result of childhood trauma and neglect, I feel empty in the absence of drugs and alcohol and am constantly searching for an escape from the existential emptiness that results… to fill this emotional chasm I propose the following harebrained theory that demands that the whole universe operate as puppet, as theater, to produce an illusion that will quiet my inner chaos.

Not all scientists act or think as a scientist should, but the goal of scientific inquiry is different from rhetorical arts precisely and only because it seeks knowledge by looking outside of the self. Instead of asking, How can I find answers that resonate with how it feels to be alive?, the scientist asks, How might I figure out how this universe works, despite the noise generated by my wants and fears?

Randall Lee Reetz

Randall Lee Reetz said...

Look, you can build any number of false sets of information. Most environments are well removed from the edge of heat death, meaning there is plenty of slop, plenty of states between any now and equilibrium. So I can get up and yell to the world, "I am God!" or "I am a carrot!" and such statements, won't cause much of a crisis or collapse in the entropy cline upon which I depend for existence. Why is this true? Why can I think or say things that are absolutely false without causing an immediate cascade towards entropic oblivion? Because abstraction systems like our brains, and the structured products of our brains, only work because they are well insulated from causal reality. The linkage between abstraction and the abstracted, between language and the reality the language describes is necessarily weakly bound. Imagine, as illustration, the opposite situation, imagine that you are omnipotent, imagine that everything you think or say is instantly and physically manifest. Such a scheme would obliterate itself within moments. What makes an abstraction system powerful is precisely how de-linked it is from causality. You have to be able to think or say the words "I ate my house" without your house disappearing or your stomach blowing up. Words and thoughts have almost zero power and this is a good, a necessary, an essential aspect of language, and of thinking! It took the universe 13.7 billion years to get to this place of abstraction sophistication, this place where ideas are so comfortably removed from causal reaction. It is probably the greatest single example of irony that the species finally and uniquely endowed with the requisite distance between abstraction and abstracted, should spend so much time wishing for the very linkage the universe spent so much time removing itself from. It is an irony born of ignorance. Most humans are and will remain forever ignorant of the fact that the power of any thinking machinery is directly related to how removed the results of that thinking is from direct physical causality.

Randall Lee Reetz

Randall Lee Reetz said...

You want to understand "information"? Then ask your self how and why any change happens. Every configuration (all information) is the result of a collapse of a complexity cline. What causes complexity clines to collapse? What are the parameters of collapse? Should change occur, what can we know absolutely about the result? Why does change happen? What parameterizes change? What can we know about change independent of domain or situation? Answer these questions and you are well on your way towards understanding information.

Randall Lee Reetz

Randall Lee Reetz said...

In the longest run, fidelity between abstraction and reality, between map and territory, determine the survival of the abstraction and importantly, the influence exerted by any given abstraction system on the shape of the future. Why? Answering this question demands an understanding of the essential difference between map and territory. A map is valuable (worth the energy necessary to create and maintain itself), only if both it and the machinery necessary to navigate the it, is significantly less than the total information within the territory being mapped. A map must exist as a compressed version of territory. So what is compression? Compression means salience is favored over repetition. A compressed map of a crystal includes the minimal description of the repeating unit, and the bounds of the area within which the unit is repeated. The territory is the actual repetition of the crystal units, the map is its minimal essential description. And it is here that we begin to understand how knowledge and compression are linked. Compression demands knowledge. It might take x bits of data to describe a crystal, but should one understand and have described all atoms or a particle theory from which one can derive the table of elements, than one can reduce the number of bits necessary to describe all crystals. The universe selects for compression in abstraction schemes. This demands a better and better understanding of the universe and more specifically, the evolution of the universe. It is costly to derive such a map. Localized maps, less demanding of knowledge, less costly, will tend to out-compete better maps in localized and limited environments. But, in the long run, the universe will select for universally accurate maps. That is what we refer to when we use the word "eventuality". The eventuality of the universe, of any universe, is what cosmologists call "heat death". Getting to heat death is not a passive process. Evolution is constantly selecting for knowledge that allows a faster more inclusive path towards heat death. The universe gets better and faster at reaching its own eventuality. Why should this be true? Because a better understanding of itself allows the universe to fall faster, to collapse its own complexity cline at higher and higher rates of speed. Such loci of collapse have a tendency of increasing selective influence. Where the universe is collapsing the fastest, systems compete more feverishly for dominance. You either eat things faster or you get eaten by things that do. As the rate of collapse increases, so to do the forces that rip up and dissolve complexity. Should a complexity reduction scheme wish to survive the chaos of its own making, it must also have learned to maintain itself against the onslaught of the corrosive results of its own processes. It is easy to reduce a complexity cline by destroying your self. Far harder to reduce the same cline in such a way that you can keep doing so tomorrow. This is the double-edged sward that is evolutionary fitness. Evolutionarily (and intuitively), the goal of any local production of complexity, must be the universal reduction of complexity.

Sorry, if you want to understand "information" you must understand the whole universe and the base process that determines its shape and eventuality.

Randall Lee Reetz

Randall Lee Reetz said...

So, you can see, the Natural/Non-Natural division in meaning or information does not exist. All information exists at some point along a continuum that measured by fidelity to and compression of the salient attributes by which causal reality falls towards heat death. It matters not at all what or who wrote that information, only how succinctly the whole universe can be described such that the universe can use this information the accelerate its own path towards heat death.

So, do you want to talk about information as it really is or do you want to continue to talk of information in the context of ideas and framings that resonate with human emotions, needs and fears?

Randall Lee Reetz

Randall Lee Reetz said...

So, now that we have established an empirical and causal foundation as definition of information, let us carefully approach the issue of context and vantage. How an entity filters and makes use of information depends upon its unique goals and needs. To the dung beetle… etc. But ultimately, in the widest context, the entirely of the universe and in the full context of time, nature will always select for those entities who's vantage most closely mirrors that of the entire universe. What this means is that individual behavior and perspective allows for radical departure from universal truth, but only at a cost, a cost that will assure that adherent schemes can not succeed when in competition with schemes in higher accord with universal causality. A narrow context can't survive long as it depends upon a limited and statistically unique environment. However, in the short term, entities that specialize will always out-compete generalists.

Randall Lee Reetz