Author: Psybertron

List of Semantics References by Greg Sanders
From a 1993 Usenet correspondence from Greg Sanders, recently brought to the top of comp.ai.nat-lang
Quote … the following (alphabetically by author) are some of the books
I have come across that I regard as interesting (omitting material on
Montague semantics). I make no claims that this covers the field. I
have not read any of “Meaning and Grammar.” I particularly recommend
Pinker’s “Learnability and Cognition” as a good entry point to this list.
“Natural Language Understanding” by James Allen
“Situations and Attitudes” by Jon Barwise and John Perry
“Meaning and Grammar” by Gennaro Chierchia and Sally McConnell-Ginet
(appears to be THE heavy-duty book on formal semantics)
“Language and Problems of Knowledge” by Noam Chomsky
“Matter and Consciousness” by Paul Churchland
“Mental Spaces” by Gilles Fauconnier
“Philosophy and Cognitive Science” by James Fetzer
“Meaning and Truth” edited by Jay Garfield and Murray Kiteley
“The Artificial Intelligence Debate” edited by Stephen Graubard
(contains what I think is the best paper by Hubert and Stuart Dreyfus)
“Consciousness and the Computational Mind” by Ray Jackendoff
“Metaphors We Live By” by George Lakoff and Mark Johnson
“Women, Fire, and Dangerous Things” by George Lakoff
“Cognitive Grammar” (volumes I and II) by Ronald Langacker
“Learnability and Cognition” by Steven Pinker
“Connections and Symbols” edited by Steven Pinker and Jacques Mehler
(see especially the paper by Steven Pinker and Alan Prince
End Quote.
Interestingly, in Dupuy’s book reviewed below, several of these are obviously referenced, but more importantly is Dupuy’s basic point that any “science” would do well to reflect on earlier attempts to solve its problems and not presume that new ideas actually supercede the old. 1993 is a long time ago in web-enabled knowledge management !

Willard Van Orman Quine – Word and Object (1960)
Reading Quine’s work, from the Peirce, Carnap, Davidson school of linguistic modelling of what we can know about the world. Interesting because the book is written entirely from a linguistic perspective – components of language learning and usage, and translation between languages – in no way aimed at Cybernetics or Knowledge Modelling, so none of the messages are biased to that purpose. Very readable with lots of common sense examples. Starts by illustrating the social processes of learning language and meaning. Then, in “Language and Truth” chapters, he covers use of language in objective “scientific method” – and the apparently natural concepts of simplicity used in induction of facts – Occam’s razor if you like. A promising start, relevant to several threads of interest here.
(Q: is it Quine as in queen, or Quine as in dine ?)
(Post note – corrected Pierce to Peirce in response to query from Davin Enigl.)

Resurgence of AI in WWW – Story from McKinseys via CNet News
Link from Danny Ayers Blog
(Note : Danny is one of the authors in WROX XML Meta-Data Text, along with Pepper, Moore, Rivers-Moore, Wrightson et al – it’s a small world in the XML / XTM / RDF / SemanticWeb / Knowledge practicioners field.)

Searle vs Pinker correspondence at New York Review
Good reference list too.
Debated on comp.ai.philosophy
Oh no I didn’t. Oh yes you did. Pity to see another binary argument. Can’t get logged on to enter the debate, however the disagreement is 80% about representation of each others thoughts, not about the actual philosophy being debated. Both agree 100% on …
Pinker : “words and rules are necessary for understanding but not sufficient.”
Searle : “words and rules are never enough to determine interpretation, not even in the simplest cases”
The other unnecessary binary argument …
“There are brute, blind neurophysiological processes and there is consciousness,” Searle wrote, “but there is nothing else.” He also goes on to say “I have never relied on common sense [Searles emphasis]. I appeal to a logical distinction between the syntax of the implemented program and the semantics of actual human understanding, and the thought experiment in question is designed to illustrate the distinction between the syntax and the semantics. I am not sure I know what “common sense” is, but I doubt that it contains theories about the distinction between syntax and semantics.”
Surely common sense and experience remains a good test of deductive reasoning, even if not a sound basis of inductive logic. Cognitive “science” may deserve a bad name if it insists on the rationale of scientific method (Bacon), and fails to suspend disbelief in common experience (Feynman). We are talking about the human mind, a social “science” here – as with anthropology (Pirsig) and ethnography (Walsham), the rules of scientific method need not be presumed to address the whole problem. Why throw the baby out with the bathwater ?

Three more good links from Jorn at RobotWisdom

J.K.Galbraith Interview
Take on Enron / WorldCom from CommonDreams Newscentre

The TAO of Topic Maps
By Steve Pepper, who acknowledges contributions from Graham Moore, with whom he co-edited XTM. (See ealier notes on Graham / Empolis and SWAP)

HumanML Taxonomy from OASIS
Hopeless quest for ontology of human emotions and intent ?

The Mechanisation of the Mind – On the Origins of Cognitive Science.
by Jean-Pierre Dupuy, translated by Malcolm DeBevoise
(Princeton University Press, New French Thought series ISBN 0-691-02574-6)
Original reference provided by Jorn the original blogger.

Earlier partial review here, with some updated highlights as follow:

Dupuy’s main theme is the disappointment that Cybernetics and subsequently Cognitive Science failed to exploit opportunities for important links with Philosophy of Mind and Social Sciences, which could have generated important advances and assured the future importance of these subjects. Unfortunately the arrogant temptations to claim a Grand Unifying Theory of Mind and Matter were too great and will prove to be the downfall of Cybernetics and Cognitive Science and spin-off subjects like AI and Second Order Cybernetics, unless the lessons are learned. Each new philosophy which believes itself to contain the sole answer is doomed to repeat the mistakes of its predecessors without “modesty and critical reflection”.

Interestingly, the whole story of these subjects and their near misses with countless schools of philosophical thought, proceeding more by misunderstanding than any common knowledge, is a microcosm of the subject itself – a case of a too complex system with emergent “organic” properties that cannot be attributed to any one of the elements, and who’s observable “phenomenal” outcomes could have been vastly different with only a few minor changes to internal or external connections or events.

For someone who claims “literature [to be] the highest form of knowledge”, it is surprising that Dupuy ultimately considers that “the existence of a symbolic level that is structured like a language and functions like a machine – does not stand up to critical analysis” Presumably its the “and” that he has trouble with, so perhaps he is not dismissing the lingusitic element of knowledge, just the finite, programmable aspects of a language built into the Turing machine assumptions.

Ultimately the theme of “subjectless processes” or “knowledge without a subject” is decisive in the view of knowledge “that cannot be appropriated by any individual consciousness” …. quoting Hayek …. [Knowledge] is embodied in norms rules, conventions, intutions, which are themselves incorporated in individual minds in the form of abstract schemata. We can make use of so much experience, not because we [literally] possess such experience, but because it has become [sub-consciously] incorporated into the schemata of thought which guide us.” I think Heylighen et al would approve of the “process view” of knowledge.

Knowledge, nor any mental state, is no more attributable to a single human subject, than it is to a group of individuals. Both are complex systems of which knowledge is an emergent property attributable to neither the whole “pseudo-subject” nor the individual. For the single human the complex system consists not only of the physical neural network of the brain, and the physio-biological systems around it, but also the complexity of both the conscious and sub-conscious mental systems. For the group, the arbitrary interactions on many levels are the complexity.

Other goodies. On my thread of rationalisation / reification destroys knowledge ; Axiomatisating complexity [has] the effect of making it disappear, [rather than explaining or understanding it]. Also paradoxical element of “reductio ad absurdum” – beware the pitfalls of reductionism (Leon).

Many, many secondary references on every “ism” imaginable. Hayek looks like a good place to go next. So many individuals, works and schools of thought referenced that Dupuy probably offers an excellent framework to organise the analysis around. An advance on my manifesto and dissertation in any event.

Finished Dupuy – review to follow.

E-Mail Working Again
Since yesteday evening UK time.

Some holding thoughts …
Binary vs three-layer world. Mind & Matter debates. Mind – Conscious & sub-conscious thinking and behaviours, as well as non-thinking reflex stimulus / response behaviour. Every time I see a binary argument, I suspect a three layer problem – net result is that every layer (having two interfaces) also has three layers, thus infinite onion-skins in every issue. Simple topology, but naturally fractal.

Beware definitions – defining hard boundaries – this and not this. (someone said that ?) Also see my caveat on the “Glossary” page – One of the web dictionaries of philosophy has an ancient quote to the same effect (cannot immediately re-locate ?). (Caveats to texts – never start a presentation with an apology adage ? This is not an apology , or since I’m starting with one, I can’t be selling you anything etc … Double-bluff / games theory etc, where will this end ?)

Not enough words in the English (or any natural) language to reserve hard definitions for each – example – behaviourists would like to define behaviour as meaning only stimulus / response behaviour. Note also conditional statement in Turing machine definition (it works … “provided the problem can be unambiguously defined by a finite set of words” … or words to that effect). One feature of “isms” is that they become “schisms” – divisions divided by defintions which were really only intended to be working definitions against which to apply the chosen logic in the chosen space. From the outside the natural language words look like they’ve been “hi-jacked”. This gets more problematic the more its the meaning of “natural language” we are concerned with – the context becomes the subject content too – tricky one [Post Note – not come across Quine at this point?] – very close to the Catch22 line I keep raising – a circular reference – a bootstrapping problem. (Q – is this also related to the “there is no such thing as a private language” debate ?).

On a similar problem with naming or definition of words representing nouns (tangible or conceptual) – often observe the behaviour that people use the name for a main sub-type which is the same as the common name of it’s parent. It’s a kind of arrogance that the sub-type currently under consideration is somehow all important – often means siblings have trouble resolving their parent from the “main” sibling. In complex, uncertain, situations involving learning by discovery (heuristics) it becomes essential that long-winded precise naming is used until such time as the common ground is firmly established. “Why use one word when a sentence will do ?” – is an adage I’m often heard to utter.