Search This Blog

Wednesday, August 8, 2018

Grice v. Carnap on what philosophy isn't

Here's a draft for "The City" which dates back a while.  I stumbled across it and made a pass at bringing it to a close.

In (once) recent postings to the Grice club Speranza and I have touched anew a topic on which we have not wholly agreed in the past, perhaps in the manner in which Grice and Carnap might have differed.

At its most stark there is a radical difference between Carnap and Grice on the scope and nature of philosophy.

For Carnap philosophy is confined to a priori reasoning yielding analytic truths expressing necessary propositions, which he would rather have in some formal language by way of escaping the inescapable ambiguity and logical incoherence of natural languages.

Our knowledge of natural languages falls under Hume's "matters of fact", hence a posteriori, synthetic and contingent.
For Carnap the study of such things, a posteriori knowledge, is, by definition empirical science, and for this he is called scientistic.
Well perhaps not just for that.  Its not just that he calls it science, but also that the "unity of science" demands that these phenomena be considered continuous with, and addressed by the same methods, as the rest of science.

I have tended to talk in a similar manner, at least some of the way.
So I talk about the study of natural languages as belonging to empirical science, even though I don't myself subscribe to the "unity" thesis.
(and I might add, that Carnap's pluralism is a reason to wonder how substantial the "unity of science" was for him).
Against this, Speranza, who knows more about natural language, philosophy of language, ordinary language philosophy and the science of linguistics (none of these to be confused) objects, quite rightly.

There are a few points here on which I will gently criticise Carnap (where others would do so enthusiastically).
A lot of it is "mere" terminology, some of it is demarcation, which is also terminology.
Thus, in my case if not in Carnap's, its little more than inept terminology to talk of all a posteriori knowledge as belonging to science, and it's worth pleading that if that's as far as it goes, then it doesn't really amount to Grice's devil of scientism.
On the other hand, insisting as Carnap did (though he did soften) that philosophy is confined to analytic pronouncements, is the kind of terminological eccentricity which appears as a controversial demarcation, and an intolerant ejection from the status of philosophy of much that philosophers have done.

The suggestion of dogmatism may be contested, for Carnap was more conspicuously pluralistic than he was an advocate of the unity of science, and these two are in tension.
The unity of science is most forcefully presented as a doctrine about language, as some kind of reductionism of the whole of language to one special language.
But Carnap's pluralism rejects the thesis that any language has a special status. He wants phenomenalistic, physicalistic and "theoretical" language all to be equally acceptable, and expected each science to have its own special language.

Grice has an incompatible principle which also looks as if it might be one of those terminological/demarcation things.
It might just be methodological, that all philosophical problems should be addressed in the first instance by a careful study of the relevant ordinary language, but it is hard not to see this as a refusal to accept as philosophy those kinds of problem for which ordinary language can provide no illumination, and thus as a matter of demarcation.
Could Grice really have believed that there are no philosophical problems for which a preliminary study of ordinary language is not valuable, or even relevant, and which his doctrine would therefore be excluding from philosophy?

By way of a speculative gesture I'm going to suggest a concession to which I think Carnap might possibly have been amenable, and a complementary concession which, if extracted from Grice, might draw the teeth from this conflict.

It's principally about adjustments to terminology and demarcation.
The demarcation issue concerns the respective scopes of philosophy and science, and the point to press upon Carnap is that the real world is rather messier than his principled division of academic disciplines along lines inspired by Hume's fork (but not actually in conformance with Hume's conception of philosophy).  Not all scientists are concerned with empirical matters, there is such a thing, for example, as "theoretical physics", which is an entirely mathematical discipline concerned with the mathematical consequences of scientific theories such as the general theory of relativity, rather than with the empirical, experimental confirmation or falsification of scientific theories.

Likewise there always have been, and possibly always will be, kinds of philosophising which involve reasoning a posteriori to conclusions which are not purely logical, but which for one reason or another cannot be addressed by the methods of empirical science, or which will only become science after some kind of pre-scientific investigation (perhaps "conceptual elucidation") has rendered them fit for scientific investigation.

The concession for Carnap is to give up his simplistic conception of the words "science" and "philosophy", and allow that these be decoupled from the rigid association which he preferred between these disciplines and the search for synthetic and analytic truth respectively.

Carnap was capable of making this kind of terminological adjustment, bowing thus to necessity.  He did in fact concede on the scope of philosophy, and he also gave signs of conceding on the usage of the term "logical truth".
Certainly he did shift his usage of the technical term L-truth which for many years stood for "logical-truth" in the broad sense in which for Carnap took to be the same as analyticity.  In the transition from the first edition of Meaning and Necessity and the paper on "Meaning Postulates" which was to be included as an appendix in the second edition, he shifts to using the term "L-truth" for a narrow conception of logical truth and introduces the term "A-truth" for analyticity and the broad conception.

There is no sign that he is personally inclined to take the narrow view, but by this time it perhaps seems a fait accompli, that the community now takes the narrow view (without in general recognising that this is mere terminology).

Turning back to Grice, the concession we need is that, even if a study of the relevant ordinary language were allowed as an important preliminary (to philosophical insight), maybe sometimes there isn't any.  Surely there may be recondite and esoteric corners of mathematics or physics so far removed from everyday life that terminological issues can only be progressed through a detailed analysis of the scientific origins of the particular esoteric language.  Must these corners of science be devoid of philosophy?




Tuesday, August 7, 2018

Cognitive Architecture

Following a slow moving theme on which I posted here some 18 months ago, I wanted to write something about a kind of "cognitive architecture" in which I am now engaged, and I thought it might be suitable material for this blog.

Eternal truths, if such there be, are of course, once established, a kind of knowledge, and knowledge is the stuff of cognition.  So, the place where one might hope to find eternal truths could be some ideally conceived cognitive system.  A philosophical exposition of what eternal truth might be, how and where we might find it, and how it might be brought to bear in improving our mundane existence might be thought of as a kind of cognitive architecture, and also as a special species of philosophy, which I now like to call synthetic philosophy.

Having thus made a pass at arguing the relevance of this topic to the city of eternal truth, I will try to elucidate more carefully the kind of cognitive architecture which I think might serve the purpose.  I'm going to do this by throwing a bunch of adjectives at it, and explaining how these are to be taken, so that we arrive at a more definite conception.


  • architecture - I think  of this as the highest levels in the design of some system, concerned with the system as a  whole, the principal subsystems and how these fit together, and all the considerations which might precede these matters in the development of the system, particularly, in the case of cognitive systems, a whole raft of considerations which are recognisably philosophical.
  • cognitive - this tells us what the system is intended to do, viz. to acquire, hold, reason with and apply knowledge.
  • synthetic - there are two distinct ways in which the proposed architecture should be considered synthetic.  The first is that the architecture is not intended to be descriptive of any existing cognitive system, but rather prescriptive of various possibilities, which we offer for consideration as models for a future cognitive system which we might strive to realise.  The second is that the system is intended to be in part manufactured, and ultimately, overwhelmingly manufactured.  A third sense is that the aspects of the architecture which may be regarded as philosophical in character (e.g. the account of what in the system counts as knowledge) will be examples of this kind of philosophy which I am calling synthetic philosophy (a term I will elaborate on in due course).
  • hybrid - the cognitive systems for which an architecture is offered are intended to be hybrid, a mix of organic and inorganic constituents.  Initially the intelligence will be supplied by organic components (human brains), but eventually some inorganic subsystems will deserve to be considered intelligent as well.
  • cosmic - If we first consider our globally networked computing machinery, together with all the human brains which in one way or another connect into that network, as being a computational system slowly morphing into a cognitive system, and think forward to a period between one million and one billion years in the future, when the physical reach of humanity and our progeny extends spatially over a significant part of our galaxy, then we have the context at which the proposed cognitive architecture is aimed.  It is nevertheless intended to be immediately applicable, and hence an architecture for the transition from our present global system to a future cosmic system.  The hybrid/cosmological pair are also intended to help de-anthropomorphise our conception of cognition, we are not concerned merely with human cognition, but with cognition in very different systems.
That's too many adjectives to throw into a name, so I'll stick with cosmic cognitive architecture or just cognitive architecture, and you may understand that, unless the context dictates otherwise, I mean the whole kit and kaboodal.

The first port of call in the exposition of this kind of cognitive architecture is philosophical, it is the articulation of a systematic philosophy in terms of which the cognitive system may be understood, and which will also form bedrock in the "belief system" of this cosmic cognitive system.  Since this most closely relates to the philosophy of Rudolf Carnap, I will talk about it at the Carnap Corner Blog.

Sunday, February 12, 2017

To mock a mockinbird: Carnap, Grice, and Smullyan

Speranza

Smullyan shared some passions with Carnap and Grice, if you can believe it.

Smullyan, the author of a very influential, "First-Order Logic" believed, well, in the power of first-order logic.

His education was peripatetic. An MA followed by a PhD from Princeton on "Theory of Formal Systems."

His main contribution was puzzling: it was his puzzles.

He was an admirer of Goedel, and his list of favourite authors should please both a Carnapian and a Griceian: Dodgson (aka Lewis Carroll), Boole, Cantor, and a few others!

And he had a sense of humour.

When Grice left Oxford, he had to give some reason. He said he moved because he was looking for the assistance of logicians and he couldn't find ONE in Oxford!

This was a bit hyperbolic. He, for one, was one (Strawson credits him in "Introduction to Logical Theory" as the tutor "from whom I have never ceased to learn about logic"). In "Vacuous Names," Grice's convoluted exposition of a System to allow for names like "Marmaduke Bloggs," who refers the hero who climbed Mt. Everest on hands and knees ("the invention of journalists, as it happened," Grice adds), there is a passing reference to Smullyan. With Myro, and others, Smullyan provided Grice with a way to provide a neat exposition of first order predicate logic -- of which Smullyan was an expert.

The fact that both played the piano helped!

Saturday, February 11, 2017

Reason and Reasoning -- and Grice's Bootstrap

-- as it combines with Carnapian tolerance!

Speranza

R. B. Jones has shared with us (that's R. B. Jones and Speranza -- and all the other followers of the yellow brick road to the City of the Eternal Truth) some comments under "Deep Learning Eternal Truth".

Jones notes:

"I thought it might be of some interest to the followers of this blog (both of them) for me to say a few words connecting my present pre-occupations with the Carnap-flavour of the City of Eternal Truth, and this is it."

Good. I love 'pre-occupations'. They contrast of course, Carnapian pre-occupations, with Griceian post-occupations. Which reminds me of the Roman occupation, as satirised by Sellars and Yeatman in "1066 and all that" -- nice figure!

Jones goes on

"I spent a while a few years ago failing to complete a short book-shaped work entitled "Positive Philosophy and The Automation of Reason". "

Lovely title. Jones of course is aware of Comte's positivism, and the new logical positivism. He prefers the mere 'positive', as applied to 'philosophy'. "The automation of reason" is a genial turn of phrase.

Jones:

"It ran into the sand in a very incomplete state at about 130 pages, and though I still felt positive about the enterprise I couldn't find the way to make it move again."

The keyword, though, remains, 'automation of reason', and we should use that phrase more frequently!

Jones:

"Of course (!) "reason" is the key weapon we have in the search for eternal truth."

While Grice uses 'eternal' truth -- as applied to the city thereof -- the phrase also occurs in Quine. I think it's ultimately 'metaphorical', "eternal" -- Grice's 'timeless' won't do here either.

i. Either it will rain or it won't.

is possibly NOT a 'so-called' "eternal" truth, in that in trivalent logics, is not even tautologous!

Jones:

"For a man of logical bent, surely the truths which most deserve that plaudit are the logical truths (though what are they?)."

Indeed, if 'rational' and 'logical' are interchangeable, there's also the possibility to play with 'analytic a priori' truths, as Ayer (or Freddie to his friends) calls them in his Gollancz book -- a classic of logical positivism. Ayer however notes that 'a priori' is otiose, and 'analytic truth' will just do.

Jones:

"Notwithstanding the foundering of that project, it remains my life's pre-occupation to find some way of progressing that topic, or simply of articulating the ideas on it which jostle for attention in my head. A few months ago I made a fresh start at that, shifting the context in which to progress the ideas. The whole thing has always been for me a fence-sitting between the fields of philosophy and information systems engineering."

Beautiful. But cfr. Cole Porter, "Don't fence me in"!

Jones:

"To automate reason is to develop software, and maybe, as is happening right now, to re-architect the hardware we use to execute the software (new hardware architectures for "Deep Learning",  beyond Von Neumann). "

Hence the title of his post, with "deep learning" as yet another keyword.

Jones:

"An architecture for reason and its application, depends on philosophical foundations.  The articulation of appropriate such foundations is an essential and should be a prominent feature of any such architecture. "

Hence the first part of his conjunctive title, "Positive philosophy AND the automation of reason". The implicature being that only positive philosophy can provide such a foundation. This reminds me of Husserl who said that he dreamed of a philosophy without foundations!

Jones:

"On the other hand, even if the primary purpose is philosophical, the architectural application is a valuable way of testing the practical significance of various issues at stake. One problem with a purely philosophical approach is the enormous difficulty in swimming against the tide of contemporary philosophical opinion, which in this sphere is unduly negative about the status of formal deductive systems as a result, for example, of the Godel's incompleteness results, and of Quine's skepticism about semantics and his consequent dismissal of the notion of logical truth as it was conceived by Carnap (aka analytic truth)."

Indeed, not to mention ("then why do you?") Derrida and all the irrationalists. There is some remarks by Grice on the analytic-synthetic distinction on his Valediction (to his life, almost), in "Retrospective Epilogue" to WoW. He notes that one should indeed take a pragmatist approach to the notion of analyticity.

Jones:

"A shift of thinking from a philosophical perspective on this problem to an architectural engineering perspective is liberating in a way which Carnap's principle of tolerance would endorse.
It allows the adoption of philosophical terminology, in the service of architectural exposition, on a pragmatic basis, sidestepping side issues which in this context may be regarded as metaphysical."

Indeed. Which brings us back to Grice's Bootstrap ("Try to pull yourself by your own bootstraps"). He is (in "Reply to Richards") considering

L1

and

L2

or object-language and meta-language. And he is saying that if L1 is first-order predicate calculus, L2 should NOT contain too much metaphysical jargon; for the idea is that L2 should be reduced to L1, as far as terminology is concerned. So the less technical one is with one's L2, the fewer problems with the morrow, he hopes.

Jones:

"This is what I have done to progress the body of ideas with which I approach the "City of Eternal Truth". I have moved the locus of my creativity from my web domain, hosting philosophical web pages and abortive book projects, to my github account (see: rbjones.github.io),"

which EVERYBODY should check -- where 'everybody' is a universal non-substitutional quantifier, if you must!

Jones:

"where I now seek to articulate a 21st century successor to the idea of demonstrative science  found in Aristotle's "Organon," taking science here just as broadly as Aristotle did, encompassing theoretical, practical and productive sciences, and the role which deduction and logical truth plays in this broad arena."

Good. I like the idea of 'demonstrative' science alla Kantotle. Grice used to lecture on the Organon, if you can believe it! And, to me, the most important thing of his seminars on Aristotle's Organon at Oxford is that J. L. Ackrill attended them! (And credits Grice as having taught him Aristotle "so well" -- In Oxford, Plato is almost non-existant, unless you are Hegelian or Bradleyian).

Jones:

"Of course, putting the material at Github creates an expectation that these architectural philosophies will ultimately be translated into code.  It's a dream..."

worth pursuing.

"Code" is perhaps not a Griceian word. There is usually the distinction between a code-based model of communication and an inference-based model of communication. The idea is that codes tend to be otiose if we can provide a 'rationale' in terms of inference patterns for them.

Grice said, "Do not multiply senses beyond necessity", and I don't think there is a necessity to multiply the senses of 'code', which is UNI-guous, or monosemous. But Jones uses it alla Jones, and I use it alla Grice.

The code relates to the automation of reason, that Hobbes would have loved!

Cheers

Aspects of reason and reasoning

Speranza

In his interesting "Deep Learning Eternal Truth," R. B. Jones "thought" (and rightly so) that "it might be of some interest to the followers of this blog (both of them) for me to say a few words connecting my present pre-occupations with the Carnap-flavour of the City of Eternal Truth, and this is it."

Of course there may be more than two (one of which is him) who read this if not follow the yellow brick road (to the City of course!)

"I spent a while a few years ago failing to complete a short book-shaped work entitled "Positive Philosophy and The Automation of Reason".  It ran into the sand in a very incomplete state at about 130 pages, and though I still felt positive about the enterprise I couldn't find the way to make it move again."

Blame the sand -- but I love the phrase, 'automation of reason', which of course reminds me of Grice's John Locke Lectures on Aspects of reason and reasoning (previously given as the Immanuel Kant lectures, under the same title, at Stanford -- trust Grice to turn from rationalist to empiricist as he crossed the pond!)

"Of course (!) "reason" is the key weapon we have in the search for eternal truth. For a man of logical bent, surely the truths which most deserve that plaudit are the logical truths (though what are they?)."

Well, for Cicero, 'ratio' was a problem. The Grecian (as in "Ode on a Grecian urn") keyword is merely "logic".

Jones:

"Notwithstanding the foundering of that project, it remains my life's pre-occupation to find some way of progressing that topic, or simply of articulating the ideas on it which jostle for attention in my head. A few months ago I made a fresh start at that, shifting the context in which to progress the ideas. The whole thing has always been for me a fence-sitting between the fields of philosophy and information systems engineering.  To automate reason is to develop software, and maybe, as is happening right now, to re-architect the hardware we use to execute the software (new hardware architectures for "Deep Learning",  beyond Von Neumann).  An architecture for reason and its application, depends on philosophical foundations.  The articulation of appropriate such foundations is an essential and should be a prominent feature of any such architecture.  On the other hand, even if the primary purpose is philosophical,"

Grice and Carnap would be delighted to hear that!

"the architectural application is a valuable way of testing the practical significance of various issues at stake."

Jones:

"One problem with a purely philosophical approach is the enormous difficulty in swimming against the tide of contemporary philosophical opinion, which in this sphere is unduly negative about the status of formal deductive systems as a result, for example, of the Godel's incompleteness results, and of Quine's skepticism about semantics and his consequent dismissal of the notion of logical truth as it was conceived by Carnap (aka analytic truth)."

Not to mention ("when why do you?", I can imagine Jones retorting) Derrida and all the continental irrationalists!

Jones:

"A shift of thinking from a philosophical perspctive on this problem to an architectural engineering perspective is liberating in a way which Carnap's principle of tolerance would endorse.
It allows the adoption of philosophical terminology, in the service of architectural exposition, on a pragmatic basis, sidestepping side issues which in this context may be regarded as metaphysical. This is what I have done to progress the body of ideas with which I approach the "City of Eternal Truth".

This reminds me of Grice's bootstrap principle ("try to pull yourself by your own bootstraps", in "Reply to Richards"): the less metaphysical the metalanguage, the less problems for the morrow!

Jones:

"I have moved the locus of my creativity from my web domain, hosting philosophical web pages and abortive book projects, to my github account (see: rbjones.github.io),"

which everybody should check, not just two followers!

"where I now seek to articulate a 21st century successor to the idea of demonstrative science  found in Aristotle's  Organon, taking science here just as broadly as Aristotle did, encompassing theoretical, practical and productive sciences, and the role which deduction and logical truth plays in this broad arena. Of course, putting the material at Github creates an expectation that these architectural philosophies will ultimately be translated into code.  It's a dream" but one worth pursuing.

That's the stuff that dreams are made of, as the refrain to that catchy song goes!



Deep Learning Eternal Truth

I thought it might be of some interest to the followers of this blog (both of them) for me to say a few words connecting my present pre-occupations with the Carnap-flavour of the City of Eternal Truth, and this is it.

I spent a while a few years ago failing to complete a short book-shaped work entitled "Positive Philosophy and The Automation of Reason".  It ran into the sand in a very incomplete state at about 130 pages, and though I still felt positive about the enterprise I couldn't find the way to make it move again.

Of course (!) "reason" is the key weapon we have in the search for eternal truth. For a man of logical bent, surely the truths which most deserve that plaudit are the logical truths (though what are they?).

Notwithstanding the foundering of that project, it remains my life's pre-occupation to find some way of progressing that topic, or simply of articulating the ideas on it which jostle for attention in my head.
A few months ago I made a fresh start at that, shifting the context in which to progress the ideas.

The whole thing has always been for me a fence-sitting between the fields of philosophy and information systems engineering.  To automate reason is to develop software, and maybe, as is happening right now, to re-architect the hardware we use to execute the software (new hardware architectures for "Deep Learning",  beyond Von Neumann).  An architecture for reason and its application, depends on philosophical foundations.  The articulation of appropriate such foundations is an essential and should be a prominent feature of any such architecture.  On the other hand, even if the primary purpose is philosophical, the architectural application is a valuable way of testing the practical significance of various issues at stake.

One problem with a purely philosophical approach is the enormous difficulty in swimming against the tide of contemporary philosophical opinion, which in this sphere is unduly negative about the status of formal deductive systems as a result, for example, of the Godel's incompleteness results, and of Quine's skepticism about semantics and his consequent dismissal of the notion of logical truth as it was conceived by Carnap (aka analytic truth).

A shift of thinking from a philosophical perspctive on this problem to an architectural engineering perspective is liberating in a way which Carnap's principle of tolerance would endorse.
It allows the adoption of philosophical terminology, in the service of architectural exposition, on a pragmatic basis, sidestepping side issues which in this context may be regarded as metaphysical.

This is what I have done to progress the body of ideas with which I approach the "City of Eternal Truth".
I have moved the locus of my creativity from my web domain, hosting philosophical web pages and abortive book projects, to my github account (see: rbjones.github.io), where I now seek to articulate a 21st century successor to the idea of demonstrative science  found in Aristotle's  Organon, taking science here just as broadly as Aristotle did, encompassing theoretical, practical and productive sciences, and the role which deduction and logical truth plays in this broad arena.
Of course, putting the material at Github creates an expectation that these architectural philosophies will ultimately be translated into code.  Its a dream,




Thursday, February 9, 2017

Rudolf Carnap, Herbert Paul Grice, and Marcus William Dick

Speranza

There was a post in the Grice Club which was basically a quotation from P. M. S. Hacker's essay on Witters (or Wittgenstein, if you must). Hacker has a thing, like I do, matter of fact, for alphabetical ordering. And he manages to try to list the members of what J. L. Austin called his 'kindergarten' -- i.e. the members of that play group that he led and that met on Saturday mornings. The list starts -- I shall use set-theoretical formulation to please Jones:

PG = {Dick, Grice, ...}

i.e. Dick is the first in the list, followed by Grice. I don't think Hacker is trying anything too deep, since, give me five minutes or so, and I shall find members for the play group starting with A, B, and C -- so Dick would NOT be the first!

In any case, why does this relate to The Eternal Truth?

Well, in "Dear Carnap, Dear Van: The Quine-Carnap Correspondence and Related Work," edited by Richard Creath, we read:

"Among it various communications from Marcus Dick, trying to get in touch with us."

So, yes, Dick made it to the Carnap annals, too! The interesting thing is that the communications could not be reciprocated because Marcus Dick left no 'future address', if you can believe that!

Ah, well.

In his biography, Quine notes that Dick had been a Commonwealth Fellow at Harvard, under one of Quine's own courses, and that he had displayed (or done, if you must) some outstanding 'work in logic'.

The connection between Dick and Grice is more tenous, or shall I say, implicatural, in nature!