Monday, December 20, 2004

Blogdex 20 December 2004



More on the theme of

why isn't a computer more like a telephone
:

this by

"Chris Spencer [..] a Unix systems administrator for a decade, a Linux enthusiast since 1993, and Linux has been his desktop OS since 2002. He works for Western Illinois University ... but my opinions in no way represent them (they still use Windows). Above all he believes that open source software will cure the piracy problem."
about Linux over Windows came through Blogdex page #1, and this from the article's reference list.


And Spyware on my machine? So What? linked from ithe reference list at the bottom.



Wednesday, December 15, 2004

Why isn't a computer like a phone?



Having used a PC for a decade and a half, it never ceases to amaze how unsorted out the technology is, compared with say, the telephone. A PC does a lot more than a phone, but phones - which we have been using happily for, can we say it, centuries, without having to understand anything about how they work - cannot surely have created the anxiety that PCs do. In a sense we have been sold a lemon. No one would put up with a new car which had problems with the transmission or gears. Why have we accepted over the years PCs with "unfinished" software ? Everyone seems to have slipped into a mode in which a "patch" is not seen as what it is: the software was sold to you before they got it working properly.

Though all the recommendations such as Schneider's may or may not be wise, I'll put up similar sites as I come across them. Check out the debate that ensues at the bottom of his page.

Number 22 Blogdex 15 December 2004 is who links to me "for the ultimate narcissist within you". Even for someone like me who is sub-technical, this code seems harmless enough, but how can we tell what this is? Wonder how many pasted this into their weblogs today without asking what it might do apart from give feedback?

Monday, December 13, 2004

Google Search



Google search is what I think is called a Predictive Adaptive Lexicon. When my son was quite young - - 5-6 - - it was soon apparent he was dyslexic but the schools did nothing to help. Computers were becoming quite popular. I thought this was the way to go. Remember the Amiga 500/600 with the text to speech? Teachers in the main thought dyslexia a neat excuse for middle-class parents. There is still a lamentable ignorance amongst the generality of the profession - despite plenty of courses - which amounts to a sort of willful refusal to accept the obvious.

In the 80s and 90s a lot of effort was put into dyslexia software in the UK. I kept abreast of all the developments and kept a folder 6" thick. My son never got to use any of this stuff, but now text-to-speech and speech-to-text is pretty commonplace. Many companies have produced things like colour coded word-processing of text, where, e.g.,consonant groups are in blocks of colour. This is so easy to do in Word if the text is preprepared, so buying such software is hardly worth it.

A lot of diagnostics have been produced, for example, a series of programmes put out under the auspices of a set- up at the University of Hull.

Two things I learnt about dyslexia that changed my way of thinking about it were

  • An 70s article about reading by Sevrin Chin-Chance, then a Ph.D. student, and now in the ediucation hierarchy of some U.S. State or other, which I will look up and explain in a bit more detail. C-C did some work on the types of reading strategies in ordinary readers (as opposed to slow readers) and found some very interesting things which no one seemed to take much notice of in the literature I read subsequently and to date.
  • The work on saccades. This showed - they still do a lot of this stuff - that a mature reader fixes on the middle of each word as the eye skips along the line of text. The eye is primed, according to what I read the other day, to pre-plan where to hit the page, dead smack in the middle of each word, time after time. It was obvious to me that the type of dyslexic my son was probably couldn't do this saccading properly, but scanned back and forwards over the words.
Many of the early dyslexia books such as Reading and the Dyslxic Child by R M N Crosby show flow charts explaining the various processing weak points such as
- visual input
- auditory input
- motor output
It is not difficut to work these out from scratch with no specialist knowledge. Then you can see where it could go wrong and work back to the actual problems that dyslexics exhibit.
The dyslexia syndrome includes a broad spectrum of difficulties including such such things as difficulty learning to tie shoes laces and being unable to read a clock face. Parents, look out!

  • It is not quite as simple as that, both because it is a spectrum of difficulty and there is the learning of the words in the first place, both aurally and visually. A normal reader uses a mixture of visual and phonetic means, whereas the severe dyslexic is forced back on phonetics which makes learning the words much harder and longer.
  • The brain of this type of dyslexic doesn't work in the same way as a normal reader. Much brain research was done implicating the cingulate gyrus, and much earlier work suggested the corpus callosum joining the two sides of the brain, which they said had not lost enough of the neurons it should during the later stages of development, hence too much traffic between left and right side of brain. The original term for this type of dyslxia was "developmental dyslexia" a term used by British experts 40-50 years ago.
  • Work by Galaburda et al showed that "Fast words speed past dyslexics" New Sci 27 August 1994. It would be quite east to test this using a tape recorder if your kid is dyslexic: read out the same piece of text [or three pieces of text of equivalent difficulty] at different speeds and see how much they comprehend in each.
  • simply spacing words out on the page probably helps too

  1. Do an experiment on the screen with different spacing to get the optimum spacing for your sprog. Make the words different colour too [ one version coloured/one, monochrome].
I will come back and fill out some of the details on this later.

Meanwhile, get your kids to have Google Suggest on when they are wordprocessing their home work. What is good about it is the context to words. Say, typing t-th-thr-thro-throu -thoug (and back again if it doesn't work) produces a list of phrases like "through the dark continent","through the looking glass" which is very helpful in understanding that this is through not thorough or though.

Wednesday, December 08, 2004

binary code and brains

Learning for the first time many years ago, that the text on the wordprocessor page was "virtual" - the "reality" from which it was created plucked from a series of 0s and 1s in a continuous chain - made a big impression on me. I can see how empowering writing software is because I have done a few simple ones that do calculations by copying the code out of a magazine, but it will always be impossible for me to grasp the arcane world of complex programming.

Even someone with little understanding of computers can work out that pressing the letter A on the keypad involves quite a lot of activity "under the bonnet". A signal is sent into the "box" to tell it to recover some of the bitstream it holds in memory. Then, presumably, some clever software is asked to translate that series of 000011111010101s in such a way that a series of pixels - each themselves represented by a few 0s &1s - are placed on the screen in fixed positions. Press DELETE and the whole effort is wiped out by a further set of programmed instructions. If you have the imagination or, should it be wonderment, this whole business does have a magicalness or, perhaps, what might be described as a transcendent quality. Others insist on the illusional and dellusional aspects of software.

I have an image of the bitstream on my PC contiguous with the bitstream that is the internet. In reality this bitstream, in toto, is physically discontinuous, but this doesn't matter when it can be stitched together by software and communications.


Trying to find analogues between this digital sytem and the brain is fraught with problems.

Firstly, though the brain has been recognised for a long time not to be a computer - rather, to have computational ability - people still insist on saying that a computer is a good model for the brain.

If the brain is a computer pure and simple, where are the correspondences, analogues, at the basic level ? What is the the binary code and its control mechanisms equivalent to? What is the neural network, the p.d. along neuronal axons? The neurotransmitters in the synapse are "gates" equivalent to those a computer processor. But there many types of neurotransmitter, divided into inhibitors and stimulators. What corresponds to the central processor? The brain is widely accepted as a modular structure but the modules don't relate one-to-one in the way the working parts of PCs do. A brief foray into brain damage shows things are more complicated in the brain. There is an Italian man who had brain damage - stroke, whatever - so cannot write vowels. Another man cannot write consonants. A lady, DF, is known to be able to put cards through slots arranged at differnet angles but her brain doesn't allow her to "see" the slots!

In the brain, the role of the DNA code must not be forgotten, though it is a more long-term facet than the action potential and the synapse.

Any undergratuate biology student learns how the mammalian brain has arrived at its present state, rather like the software in a nuclear power station: by repeated add ons, the pattern of which can be see through comparison of fish, reptile, bird and mammal brains. I like image of the the men who wrote the original code for the power-station software being dead and none of the current programmers working to keep the control and safety machanisms running having any idea how these dead men wrote their code. They write new bits of software to keep the various old bits working together without understanding completely how it originally worked.

Although there may be some correspondences between computers and brains, as soon as you get to things like consciousness and meaning, there is a parting of the ways. What is on the PC screen could be called a sort of "consciousness", but what is missing from this is the complex feedback mechanisms that make consciousness what it is.

Happening to be mad, knowing Freud called dreams the "Royal Road...", I think the dream is simply another form of consciousness. What the dream shows, according to my deranged thinking, is more of consciousness than waking consciousness shows. Included are some of the workings of consciousness. "Waking consciousness" is bound in to perception and realtime events. Dream "consciousness" shows more of what the brain does. In other words, the craziness of dreams is an expose of the way reality is constructed from separate components. It shows how what comes into consciousness , especially in thoughts, evolves. How out of all the masses of perceptions, cognitions, whatever, can we come up with a single, coherent conscious experience, which is not a psychopathic like an LSD trip? Sometimes only metaphors explain. I once had a weird "theatre" dream, in which I watched a play with players coming on and off in different parts, argiuing , shouting, moving and so forth, while at the same time seeming to be taking part in the drama myself. If any actor could pop in and out, different costume, different voice, different mannersims, randomly interferring with the flow of the written play, how ever do we get the coherence of the scripted play?

The image of a seen red cup is not held in the brain as an entity, but in the myriad features of its "redness" (that mysterious qualia business) and "cupness", scattered all over the brain. This includes the ability to imagine [from memory] a cup revolving round to show its various sides(e.g. handleless view) ,even though we are only looking at the one view. In other words, we are not taking red cup in perception/cognition anew each time, but taking inputs plus memory to create red cup. In this view even consciousness is not as "real" as we imagine. We don't really know how much is a construct of the brain and how much is real-time correpondence. It is asaid that the brains main fuctions are differentiation and integration. In vision we see the former: remocing all the confusion around us to present a useful picture. The brain even alters reality tp create foregoun/background, when this does not exist "out there". Integrative functions more easily understood in thought, where we canbreak down a problem into easier to understand components, then build things back up to the more complex answer arrived at.

Neural network computers have been designed which are said to "learn" like brains, but, so far, the tricky goings on in the synapse, the control that emotion plays in what the brain does, has not really been included in any big way. But the best way to confirm how a brain works by making an artifical one that can do everything a human brain can does. Then again, faulty thinking, because we don't know what real brains "do" except overtly. The bits and pieces of philosphical and scientific knowledge about the brain still don't add up to a complete understanding of how the brain works at every level. We have an inckling of what it would be like to do without what the brain does for us from our imagination and from malfunctions such as mental illness and physical damage through e.g. strokes.

Is it useful to think of unconsciousness as a sort of binary code with consciousness being the brain's virtual creation such as those letters on the PC screen? And yet unconsciousness would, or might not be, played out like consciousness. Nature is parsimonious.

Dreams might be a sort of "secondary" consciousness, one up from the bare bones of code. After all visual basic is not the same as working in primary programming. The pictures and sometimes sounds of dreams seem as real as waking experiences. It would seem they are produced in the same way as conscious experiences. But it may present like "real life" while actually working at a lower level, one or two up from the basic neural processes which store and manipulate the data of experience, because the senses are not involved. The "consciousness" could be playing at the lower level yet be perfectly understandable " to the brain". waking consciousness requires another layer of programming to make it consciousness. And yet that is loopy. Why two systems?

Maybe the test of whether a neural network has reached our level of sophication and complexity is if it has dreams, rather than if it has "emotions". We know emotions in brains have a great effect on what we see and think. Adding "emotion" to neural networks, in the sense of sophisticated weightings in the switches or gates, has begun. But we have to be able to experience the NNW's overt "emotions" in order to be sure they are like ours and not simulated. Seeing a display screen with the neural network's dream on it, or receiving reports from a robot while "asleep" that it is dreaming, might be an easier route to checking if a neural network was beginning to be a brain? How would it be possible to determine if NNW dream corresponded in any way to brain dreams? Have a chat with the robot! Compare notes on dreams. Shades of Turing.