...out of the box...

No user-serviceable parts inside.

 

[trouble at mill]

This bit of stuff comes "as is" from the web of Stephen Downes. (Ok under Creative Commons, i'm not a thief)

http://www.downes.ca/cgi-bin/website/view.cgi?dbs=Article&key=1091120914&format=full

The reason I have republished it for you reading pleasure is because the first parts were a bit tangential. I've set it in a small typeface cos its lengthy.

Re: Unshaken Hands on the Digital Street, by Michael Bugeja.

The author assumes that interaction with the physically present must take priority over the physically distant (and electronically connected). Remove the assumption in this article, and require that it be supported through argumentation, and the impact of the dialogue is lost.

In fact, it seems to me, the order of precedence of interaction ought not not be resolved by proximity, which is typically merely accidental, but by two more salient factors: priority (that is, all other things being equal, the interaction that is most important to the actor takes priority) and precedence (all other things being equal, the interaction that began first takes priority). Most interaction is a case of these two stipulii conciding in two people: for each, the interaction with the other is the most important of those available at the moment, and will continue until concluded. 'Interruption' is the process of one person suggesting that the importance of an interaction is greater than one already in progress, and it is (of course) the right of the interrupted to make the determination as to whether this is so.

In the pre-digital age, priority and precedence coincided with proximity. That is, all the choices of interactive possibilities were of people located in physical proximity, and physical proximity being a limited quantity, predence assumed a much greater importance. But it would be a mistake to equate proximity withy priority and precedence; with electronic communications, it is now possible to have a situation in which a communication by telephone is of greater priority than a presently existing in-person interaction. When a telephone rings, this is an interruption, and the receiver typically makes an assessment (often by looking at the caller ID) as to whether the telephone call is likely to be more important than the present interruption.

What is also true, in an increasingly crowded and mobile world, is that the value of physical proximity is diminished. In less mobile, less crowded times, one could assign a high degree of probability that a person wishing communication while in close proximity was also a person with whom communication would be a priority - it would be a spouse or child, a business associate, or a customer. But today's physical interactions are increasingly with strangers with whom one has no prior attachment, and so the probabilities have now tipped the other way: it is more likely that a telephone call, from one of the few people in the world to know your number, is of greater importance than a conversation with a stranger on the street or in the office.

When a person in physical proximity interrupts a person using a mobile telephone or similar electronic device, the probability is that their priority to the person being interrupted is less than the priority of the person being talked to. Where once people apologized for being on the telephone when a stranger wished to speak, it became apparent that no person need apologize for talking with his spouse, child or friend, and that it is the stranger imposing the interruption and making the request. Breaking off a telephone call (or even shutting off an MP3 player) to help a lost tourist is a mark of altruism, and as the stranger had no prior claim on the person's time, such behaviour ought to be thanked rather than criticized when written about in an article.

The mistake being made in the article below is in the assumption that the virtual interaction is somehow less real, somehow inherently less important, than the proximal physical interaction. "By the time they attend college, they will come to view technology as companionship." But this is a fallacy, a confusion between the message, which is a product of the media (a "phone" call), and the content, which is a product of the interaction (a call "from John"). More straightforwardly, the vast majority of online and electronic interactions are with real people, and there is no a priori reason to assign a real person lesser importance on the basis that they are distance (and, given such a person's prior attachment with the caller in question, very good reason to assume the opposite, that the distant person is of greater importance than the proximal). Electronic communications may be caricatured as communications with the non-real, but to draw any conclusion of important from this characterization is to ignore an obvious and self-evident truth: real people communicate electronically.

The characterization of the product of electronic communications as "dumb mobs" is an assassination ad hominem. Were it true that drunken parties the only consequence of such forms of virtual communication (were it true that such parties were known to be caused by such communications at all, as though they had not occured prior to the advent of the telephone) then perhaps one might have a case. But electronic communications have conveyed messages as vital as the boirth of a child, the formation of a business, the death of a relative, humanity's step on the moon, and so much more. Empirical observation shows that the party-generation capacity of electronic communications is a minimal, and infrequently emnployed, use of the medium. It makes no sense, then, to assign to the communication the morality of the mob.

The reactions of a person who, by dint of physical proximity, assume priority and precedence over any and all electronic interactions, are, quite frankly, the reactions of a self-important boob. They convey the impression of a person who believes that his or here mere physical presence ought to command the over-riding and immediate attention of all who come within his or her purview. They show no respect for the importance a caller may place on communicating with friends, family or associates, and demand immediate and sole attention to the matter at hand, namely, him or herself. In a world of competing interests and of increasing demands for interaction, people have learning that they must from time to time take their turn. This author strikes me as one who hasn't learned this yet.

Four

While it is a fact that each of us, as knowers, is situated in the world (situated bodies) and we learn by bumping (commonsensical understanding) into the world; What constitutes knowledge is not reducible to any of us or to our bodily presence, any more that what constitutes the English language depends upon the use of English by any speaker of the language or what constitutes mathematical truths depends upon any person's calculations.

Trivially, this is an assertion to the effect that a recognizable entity (such as knowledge, langauge or mathematics) that has an existence outside ourselves is not reducible to states of affairs inside ourselves. If we argue from the outset that these are social phenomena, then it follows by a matter of definition that they are not reducible to mental entities. But this is no more revealing than to say that a house is not reducible to our perception of a house. Such a statement is necessarily true, assuming the independent existence of the house.

More interesting is the perspective where we are silent on the external existence of the house. We presume that our perceptions of a house are caused by a house, but it is also possible that our perception of a house was caused by something that was not a house, or caused by the convergence of discrete perceptions that have no discrete external status at all. After all, we can have perceptions (or, at least, thoughts) of a unicorn, without at the same time asserting that a unicorn has an independent external existence.

The real question is, is our concept of a house reducible to our perceptions of a house. That is to say, can we arrive at the idea of a house through some form of collection and organization of perceptions? The logical positivist answer to this question was that we could, though the entities and mechanisms proposed (an observation language, logical inference) proved manifestly inadequate to the task. A similar stance seems to be being taken here. Our concept of a house cannot be reduced to a set of mental entities; no mechanism of inference appears to be adequate to the task.

When we look at this more closely, we see that the assertion is that the entity in question - our idea of a house - is not composed of the entities from which is is supposedly derived. That is to say, we could replace one or even all of our mental entities (thoughts, perceptions, etc) with distinct instances of those entities, and yet the perception of a house would remain unchanged. This gives it a unique ontological status.

Consider, for example, what would happen were we to attempt the same thing with the Great Wall of China. The Great Wall is composed of bricks. Were these bricks removed, and replaced with new bricks, we would no longer say that the Great Wall of China exists; rather, we would say that we have constructed a facsimile of the Great Wall, and that the real Great Wall is now a pile of rubble somewhere.

By contrast, consider the image of Richard Nixon on a television set. This image is composed of pixels. Were we to replace one or all of the pixels (as happens 72 times a second, more or less, depending on your screen refresh rate) we nonetheless say that we are seeing the same image of Richard Nixon. The image has a continued existence even though all of its physical components have been replaced.

Why do we say that one set of pixels and another set of pixels constitute the same image? It is clearly that the two sets of pixels are organized in a similar way. For example, both sets of pixels have two clusters of dark pixels near the mid-point of the image - what we would call Richard Nixon's eyes. We say that the two sets of pixels constitute a single image because the organizations of the two sets of pixels resemble each other. Take one of the sets of pixels, and organize them randomly, and we would say that we no longer have an image of Richard Nixon, ever were we to have exactly the same set of pixels.

Now it is tempting, when identify a similarity such as this, between sets of unrelated collections of physical entitities, to say that some discrete physical entity must have caused this similarity to occur, that there is a real Richard Nixon that this image must be an image of. But of course the same reasoning would force us to agree that there is a real Donald Duck. Since Donald Duck is an animation, and does not exist except in the form of similarly organized pixels, it is evident that such reasoning is in error. But then we must ask, what is it that makes a collection of pixels into Richard Nixon or Donald Duck?

The being an image of Richard Nixon is not contained in any or all of the pixels. Nor may we assume that it is caused by an external entity. All external possibilities thus being exhausted, the explanation for the fact of an image being Richard Nixon must lie in the perceiver of the image. We say that the image on the screen is an image of Richard Nixon because we recognize it as such. This organization of pixels is familiar to us, so much so that we have associated it with a name, 'Richard Nixon', and even apparently unassociated utterances, such as 'I am not a crook.'

In a similar manner, entities such as knowledge, language and mathematics (as commonly conceived) exist only by virtue of the organization of their constituent parts. No particular instance of a fact, a word or a calculation is a necessary constituent of these. But something is called a piece of knowledge, mathematics or language only if it is recognized as such.

Most of our understanding in the world of what it is like to be embodied is so ubiquitous and action-oriented that there is every reason to doubt that it could be made explicit and entered into a database in a disembodied computer. We can attain explicit knowledge through our understanding with the world, by virtue of having bodies. We can find answers to questions involving the body by using our body in the world.

There is a lot packed into the seemingly innocuous phrase, 'made explicit', and the phrase is sufficiently distracting as to throw us of our course of investigation.

Consider, again, the image of Richard Nixon. What would it be to 'make explicit' this perception? One suspects that it needs to be codified, cast into a language. Thus, we say that our perception of Richard Nixon is 'made explicit' when it is associated with the phrase 'Richard Nixon'. (Is there another sense of 'made explicit'? Does the discussant have some other process in mind?)

When the image of Richard Nixon is made explicit in this way, however, a great deal of information is lost. The original perception is abandoned - nothing remains of the organization of the pixels; the pixels, and the organization that characterized them, form no part of the phrase 'Richard Nixon'. Nor either is the act of recognition contained in this phrase. The association of the image of Richard Nixon with similar, previously experienced, phenomena, can no longer be accomplished.

What is important to recognize here is that the information has been lost, not because the original image was produced by our bodies, and that the phrase wasn't (an assertion which is, as an aside, patently false - where else did the phrase 'Richard Nixon' come from if not from our bodies?). It is because the image of Richard Nixon has been completely replaced by this new entity, which represents the original entity only through association, and not through resemblance. Only if, on being presented the phrase 'Richard Nixon', we could call to mind the original image (the original organization of pixels) would we be in the position to make the same set of associations as the original viewer.

If I am presented with 'A' I can immediately infer that 'A is for Apple'. But if I represent 'A' with 'B', then I no longer have the capacity to make that inference. There is nothing in 'B' that would lead me to say 'Apple' (and the expression 'B is for Apple' even seems absurd). Presented only with 'B', therefore, I am unable to equal the cognitive capacity of someone who has been presented with 'A'. It is not therefore surprising to see people say that the accomplishment of such cognitive capacity on the part of a system presented only with 'B' is impossible.

But it is not impossible. It is impossible only if it is impossible to present the system with an 'A' instead of a 'B'. It is impossible, for example, if the having of an experience of 'A' is something only the first sort of entity can have, and that the second sort of entity cannot have. And that comes down to this: is the stimulation of a neuron by a photon the sort of thing that only a human can have? Put that way, the question is absurd. We know that photons stimulate things other than human eyes; that's how solar power works.

Perhaps, then, recognition is the sort of thing that can only be accomplished by a human. Presented with the same organization of photonic stimuli, is it the case that only a human can recognize it as Richard Nixon, while a non-human system is restricted to, say, associating it with 'Richard Nixon'? Again, the answer to this seems to be no. While it is true that most computers today think and store information only in symbolic form, it is not reasonable to asert that they must. A computer can store an image as an image, and given multiple images, nothing prevents a computer from performing the cybernetic equivalent of recognition, the realization that this is similar to that.

The question here is whether the perception of a given phenomenon - any phenomenon - dependent on the physical nature of that phenomenon, in such a way that the given instance of the perception could not be replaced with a similar instance without it becoming a different perception.

It is clear that the exact physical instantiation of the perception is not required. If I were to lose an eye, and were to have this eye replaced with a donor eye, such that the eye (and therefore any action of the eye) has a completely distinct physical constitution, it is not clear that I would no longer be able to see. Indeed, our intuitions and our research run in the other direction. We can replace eyes (and other body parts) without changing the perceptions that these body parts produce. Seeing with a donor eye is just like seeing with our original eye, or so much so that the difference is not worth remarking upon.

One asks, now, whether the physical constitution of the donor eye be the same as the physical constitution of the original. Is it necessary that the donor eye be a human eye. Were the donor eye to be instead an artificial eye, strikingly similar, or course, to the original eye, but nonetheless indisputably of non-human origin, is there anything inherent in the function of this new eye that would make it not capable of enabling the same perception as the original eye? It is true that today's artificial eyes produce only shadow-like vision. But this attests only to the fact that it is difficult to make eyes.

More significantly, would it be possible, with the replacement eye, to recognize an image of Richard Nixon as being an image of Richard Nixon? It seems manifest that it would. For, as observed above, what makes an image an image of Richard Nixon is not the physical constituent of the image, nor even the origin in an external cause of the pixels, but rather, the organization of the pixels and the recognition of this organization as being similar to other perceptions we have already had. And even were all of a person's perceptions obtained through this artificial eye, there seems to be nothing inherent in the physicality of the eye that would make this impossible.

As we more through the other organs of the senses, and as we move deeper into the cerebral cortext, we wonder, then, at which point this stops being the case. At what point do perception, recognition, and cognition, cease to be founded on the organization of the pixels, and start to be founded on the physical constitution of the pixels? At what point does it become necessary for a thought to be grounded in a human brain before it can be said to be a thought about Richard Nixon? The nature and function of the human eye is not different in kind from the nature and function of the deeper layers of the brain; what works with the eye would seem, in principle, to work with the deeper layers of the brain. So what is it about the human brain that makes it impossible for a computer to emulate.

If we think of computers as symbol processors, then the answer is evident. At some point, a translation from perception to symbol must occur, and at that point, so much information is lost that the processes behind that transformation are no longer capable of performing the same sort of inference a brain that does not make that transformation can perform. But is there anything inherent in computation that makes it necessary that all processing be symbolic? Is there any reason why a computer must store knowledge and concepts and ideas as strings of symbols and sentences? There is no doubt that today this is a limitation of computers. But it is not an inherent limitation; it exists because designers stipulate that at some point in processing physical input will be converted into symbolic data.

Yet, already, in some instances this never happens. When I capture an image with a digital camera, and upload it into my computer, the image is not converted into symbols (and it would be absurd to do so). The original state of the pixels, as they were inflenced by photons, is what is stored. Of course, this image is not intermingled with other images, as it would be in a human brain. It is stored separately as an 'image file' and displayed or transported as an entity when requested. Even so, this, at least, is an instance of non-symbolic data representation in a computer.

Suppose, instead, when an image were loaded to my computer, it were compared with every other image previously stored in by computer, and that the image displayed was not the original image, but rather, whatever image (or composite) was suggested by this presentation. Something like (exactly like) recognition will then have happened, and the second stage necessary for perception will have occured.

So long as we don't transform input into symbolic form, thereby stripping it of important information, there is no reason to assume that the cognitive capacity of a system handling that information is reduced. And if there is no reason to assume that the cognitive capacity is reduced, there is no reason to believe that the cognitive capacities of humans could be emulated by a computer.

Human beings respond only to the changes that are relevant given their bodies and their interests, so it should be no surprise that no one has been able to program a computer to respond to what is relevant. Bodies are important making sense with the world. Forms of life is organized by and for beings embodied like us. Our embodied concerns so pervade our world that we don't notice the way our body enables us to make sense of it. So, if we leave our embodied commonsense understanding of the world aside, as using computers forces us to do, then we have to do things the computer's way and try to locate relevant information replacing semantics. Prof. Dreyfus criticizes AI as the epistemological considerations concerning how human bodies work in intelligent behaviour.

It is evident that humans force computers to think symbolically, by virtue of such things as interface and operating system design. But do computers force humans to think symbolically?

The answer is no, and the reason for that answer is that humans are not symbol processers. Let me repeat that. A computer cannot force a human to reason symbolically because humans are not symbol processors.

Oh, sure, we have the capacity to understand and interpret symbols. But this is done in the same manner that we undertsand and interpret an image of Richard Nixon. The symbol is perceived, not as a symbol, but as an image (you have to *see* or *hear* the letter 'B'). The presentation of this symbol will call to your mind other perceptions with which it has become associated. And if you're lucky (most of us aren't, but that's another paper) the presentation of the associated 'A' will generate in you the capacity to draw the same associations as had you been presented an instance of 'A' in the first place, leading you to think, 'Apple'.

In other words, for humans, symbols are not what we use to think, but rather, what we use to communicate. We represent a mental state (a perception, say, of Richard Nixon) with a symbol (the phrase 'Richard Nixon') and send the symbol with the hope and expectation that the presentation of the symbol 'Richard Nixon' will generate in the receiver a mental state resembling your original mental state (a perception of Richard Nixon).

What is important to keep in mind here is that the information received from other people, by means of an encoded information transfer (ie., a sentence) does not become some kind of different and special *kind* of information in our brains. Information transferred to us as symbols does not remain exclusively as symbols in our brain, for the precise reason that the brain immediately wants to begin associating symbols with other types of perceptions.

The fact that we process symbols in the same way we process other types of information is what makes them work. Were we to process symbols differently, then they could not evoke other types of memories, and we would have two separate and distinct regions of thought, one for symbols, and one for images, and symbols could never be associated with images, and thus someone's utterance, expressed to us as a string of symbols, "Watch out!" would never translate into action.

To suggest that receiving information symbolically instead of receiving it directly causes us to assume a different moral, ontological, or inferential stance regarding this information is absurd. It is absurd, because it assumes that symbols never evoke other forms of perception, when it is manifest that the only reason symbols work at all is because they do.

Computers do not force us to leave our commonsense understanding of the world aside. Nothing could force us to do that. Not even one of Dreysfus's papers.

.

posted by: kenjprice at July 31, 2004 20:18 | link | comments |
tasite

posted by: kenjprice at July 26, 2004 21:15 | link | comments |
tasite

[now playing:  Warren Zevon,  Mr Bad Example]

posted by: kenjprice at July 12, 2004 23:29 | link | comments |
home and personal

Bought another camera on the weekend, a Horseman 45 FA (Fine Arts). Its so beautifully made I had to buy it. Its a large format field camera, its so big you could keep rabbits in it. It looks like someone has, and I spent half an hour re-folding the bellows. This is tapered and curved, no idea who thought this idea up. Probably the same person who first thought of milking cows.

Anyway, the Horseman will get cleaned up then I'll sell it, along with the 50 or so other classic SLRs in storage. One day we'll all wonder why 50 years of classic photographic engineering was thrown out to make way for digitals with a 3-year design life.

posted by: kenjprice at July 12, 2004 23:28 | link | comments |
home and personal