On Peirce's Four Incapacities of Man -- RE: Natural Language, Thought, and Ontology -- RE: SUO: Proposed Changes to Merged Ontology
John,
Thanks very much for your message.
You wrote:
> In my 1984 book, Conceptual Structures, I spent more time and
> effort on Chapter 2, in which I surveyed all of cognitive
> psychology and psycholinguistics, than I did on any of the
> other chapters.
First, I should say that your 1984 book has always been one of my favorites,
and your 1999 book is a wonderful successor. It is a distinct intellectual
pleasure and honor to have this correspondence with you.
> In that chapter I tried to put together
> whatever evidence there might be for whatever might be called
> "natural logic" or "natural knowledge representation" or
> "natural reasoning". I learned a lot while studying the
> literature during the preparation time, but precious little
> of it gave any hard evidence that would guide the knowledge
> representation. The study of linguistics, however, was a much
> richer source of insights. During the past 17 years, I have
> continued to keep an eye on what the psychologists have been
> doing, and there still isn't any hard evidence. But I still
> find linguistics to be a rich source of ideas.
From what I've seen, I would certainly concur with all of your statements
about the rate of progress, difficulty of finding "hard evidence", and value
of linguistics as a source of insights.
It seems to me that the experimental results cited by Pinker are at least
very close to "hard evidence" suggesting a distinction between "natural
reasoning" and the ability to speak and understand "natural language".
However, I would agree that these results are not hard evidence about the
nature of "natural knowledge representation" as it occurs within the brain.
And on the other hand, there is somewhat hard evidence of a very close
relationship between natural reasoning and natural language, e.g. in the
reports of cognitive impairment for children raised without being taught a
(spoken, written, or sign) natural language during their formative years.
Of course, many people in the fields of cognitive science and linguistics
have attempted to apply evidence in one form or another to deduce
characteristics of natural knowledge representation. To the extent that such
efforts are successful they raise important issues for the development of
artificial knowledge representations and ontologies, which may ultimately be
significant to the success of SUO. One interesting paper along these lines
is:
Pinker, Steven & Prince, Alan (1999) The nature of human concepts: evidence
from an unusual source. In Language, Logic, and Concepts: Essays in Memory
of John Macnamara, edited by Jackendoff, Bloom & Wynn (1999), MIT Press,
pp.221-261.
I would agree that the jury is out on the relative success of such efforts,
yet I am glad to see people working along these lines.
> For some idea of the rate of progress in psychology, I recommend
> the "Principles of Psychology" by William James (1890), which is
> available as a Dover reprint. Or you can get the text online:
>
> http://psychclassics.yorku.ca/James/Principles/index.htm
>
> Although it's 111 years old, the blurb on the back of the
> Dover edition says "Reading James brings a sense of perspective
> and even a little humility to our regard for more modern
> achievements."
I have read some of James' works (long, long ago) and been very impressed by
his thinking. However, I cannot claim to be a scholar of his works -- there
are so many great authors, whom I have too little time to read -- So, it
helps me greatly to have specific references pointed out, and I appreciate
your providing this reference.
> James and Peirce were lifelong friends, who discussed many of
> their ideas about language, thinking, the nature of signs, and
> thought, language, etc. I recommend the following article by
> Peirce, which he wrote early in his career (age 29):
>
> "Some Consequences of Four Incapacities Claimed For Man",
> Journal of Speculative Philosophy 2 (1868), pp. 140-157.
>
> Available online at:
> http://www.door.net/arisbe/menu/LIBRARY/bycsp/conseq/cn-main.htm
Likewise, thanks very much for this reference. This is quite an interesting
paper, along with its companion paper "Questions Concerning Certain
Faculties Claimed For Man".
I am also not a scholar of Peirce, though I have been impressed by his
thoughts in these papers, and by the accounts that you have presented in
your books, and what I've seen posted of his work in SUO correspondence. So,
I am eager to learn more about Peirce's philosophy.
Therefore, I wonder whether you would be kind enough to answer some
questions about the "four incapacities", speaking either on behalf of Peirce
or your own views? Since it might take several years for me to become a
scholar of Peirce, this could hasten my progress and might also suggest some
interesting lines of inquiry...
> Following are the "four incapacities", which he discusses in
> further detail:
>
> 1. We have no power of Introspection, but all knowledge of the
> internal world is derived by hypothetical reasoning from
> our knowledge of external facts.
I would agree that introspection is limited, but am not yet persuaded we do
not have such a power.
For example, what about the simple knowledge that we have thoughts? What
about the knowledge that we can "self-verbalize" in our minds, without
speaking physically or externally? Is this not knowledge (at least of the
existence) of the internal world? I can talk to myself, in my mind, and be
aware of my thoughts -- this awareness is not based on knowledge of external
facts, is it?
If I can observe my thoughts, then I can think about them, and reflect upon
them, and develop beliefs about them, even if I cannot trace their origins.
All this is true even if in fact, many of my thoughts are about external
things. For example, I can develop beliefs that some of my thoughts might be
inconsistent with other thoughts -- Also, is the predicate "inconsistent"
based on external knowledge, or is it possible that such predicates are
internally "hardwired" and innately given?
At least in CP5:244-249, I do not see any direct consideration by Peirce of
these questions, and of these reasons for granting the existence of some
direct knowledge of the internal world.
Would you agree or disagree? I'd very much appreciate your insights on this
question.
> 2. We have no power of Intuition, but every cognition is
> determined logically by previous cognitions.
Peirce does not seem to consider the possibility that some thoughts might be
generated internally and randomly, without being determined logically by
previous thoughts. In the same way that genetic algorithms may randomly
generate structures, perhaps some thoughts are randomly composed by
processes in the brain, and have potential influence on other thoughts.
Mechanisms of this kind could account for some degree of creativity, or some
degree of "inspiration" in problem solving.
So my question is, did Peirce consider this possibility and rule it out
somehow, or is this something he overlooked? Or is it something that is
subsumed somehow in his definitions?
> 3. We have no power of thinking without signs.
So far, I cannot think of any objections to this conclusion of Peirce, and I
like very much his definition of signs:
“A sign, or representamen, is something which stands to somebody for
something in some respect or capacity. It addresses somebody, that is,
creates in the mind of that person an equivalent sign, or perhaps a more
developed sign. That sign which it creates I call the interpretant of the
first sign. The sign stands for something, its object. It stands for that
object, not in all respects, but in reference to a sort of idea, which I
have sometimes called the ground of the representamen.” (CP 2.228)
I do have some questions, however: Did Peirce deduce the need for internal,
combinatorial systems of signs, to represent thoughts and support thinking?
Did he present an argument similar to that of Jackendoff (1995) in favor of
the need for an internal mentalese?
Incidentally, it does not seem to me that Jackendoff's argument is based on
introspection. Rather, it seems based on hard evidence and logic: The hard
evidence that the brain is finite and is the locus of thought; the hard
evidence that natural language permits combinatorial expression of a
virtually unlimited number of concepts; and the logic that this implies some
mechanism for combinatorial representation of concepts within the brain...
>
> 4. We have no conception of the absolutely incognizable.
I am inclined to agree with this in some ways and disagree in others.
To me it seems that we must have some internal framework of cognition and
perception, as well as a shared cultural framework of cognition and
perception, and that what we can cognize is limited by the expressiveness of
these frameworks. These frameworks are extensible, yet again there may be
limits to their extensibility.
It would follow that we cannot cognize specifically with concepts that are
outside these frameworks, and outside the reach of possible extensions to
these frameworks.
On the other hand, our meta reasoning and reflection enables us to have some
conception of the fact that there may be concepts outside the reach of
extensions to our thought frameworks. We can have some conception of the
fact that our frameworks may be limited. And in this sense, we can have
*some* conception of the absolutely incognizable, of the fact that there may
be things that are absolutely incognizable to us.
In this regard, the work of Turing and Godel during this century would
surely have been of interest to Peirce... Yet again, the fact that we can
cognize about (at least some) paradoxes (and not go into infinite loops
while doing so) and cognize about the proofs of Turing and Godel, and the
implications of these proofs for our own thought frameworks, gives us some
(admittedly limited, indirect, and general) conception of the possibility of
an absolutely incognizable.
Also during this century, physicists have confronted the mysteries of
quantum physics. Some (e.g. Feynman) have advised that these mysteries are
incognizable, while others continue to grapple with understanding them.
Perhaps only time will tell whether these mysteries must remain beyond us.
It does not seem to me that Peirce addresses these perspectives in
CP5.254-258. Rather, he states that "there can be no conception of the
absolutely incognizable, since nothing of that sort occurs in experience"
and that "cognizability (in its widest sense) and being are not merely
metaphysically the same, but are synonymous terms." In making these
statements, Peirce seems to be saying that the reason we have no conception
of the absolutely incognizable is because the absolutely incognizable does
not exist -- he seems to be saying that the frameworks of human cognition
and perception are so complete (or so completely extensible) that no form of
knowledge about the universe is beyond us. Such a viewpoint would have been
consistent with a spirit of optimism, and belief in human progress, possible
in the 19th century...
So, my general question about this incapacity is, how do you think Peirce
would have responded to the perspectives I've presented above?
> He was writing this in criticism of Decartes, but it has
> a lot of relevance to the study of thinking in general.
Yes, I agree. I remain very impressed by Peirce's contribution to human
thought about thought.
Sincerely,
Phil Jackson
-------------------------------------------------------------------
"A man's got to know his limitations" - Clint Eastwood
-------------------------------------------------------------------
Standard Disclaimers. www.philjackson.prohosting.com
==================================================
> > -----Original Message-----
> > From: Philip Jackson [mailto:phil.jackson@computer.org]
> > Sent: Sunday, February 11, 2001 12:37 PM
> > To: John F. Sowa; pat hayes; Robert E. Kent;
> > standard-upper-ontology@ieee.org
> > Subject: Natural Language, Thought, and Ontology -- RE: SUO: Proposed
> > Changes to Merged Ontology
> >
> >
> > Along the lines of this discussion, here are some further comments:
> >
> > It is tempting to say that if we restrict "Language" to verbal or
> > written, serial human natural languages such as English, Chinese,
> > etc., then thought is possible without language: People can solve
> > some kinds of problems using reasoning (e.g. "spatial reasoning")
> > and perception that are at least not easy to express in English,
> > etc. People can display intelligence and thinking even if they
> > haven't learned a language such as English, etc. Pinker ("The
> > Language Instinct", 1994) cites a list of very incisive
> > experiments that show conclusively that thought and intelligence
> > are not identical to the ability to understand spoken, natural
> languages:
> >
> > A variety of psychological experiments show that babies can do
> > various kinds of thinking before they can speak.
> > People who suffer strokes may have impairments to use of natural
> > language, yet otherwise still be in possession of their
> > faculties, and able to think.
> > An (apparently genetic) impairment called "Specific Language
> > Impairment" can impede use of natural language, yet leave the
> > rest of intelligence intact.
> > Conversely, some people are "linguistic idiot savants" -- very
> > good at using language, yet may be intellectually impaired. This
> > may occur as a result of a variety of diseases, including
> > hydrocephalus, schizophrenia, Alzheimer's disease, autism,
> > Williams syndrome, and aphasia.
> > Pinker also argues strongly against the "Sapir-Whorf hypothesis"
> > that language determines and limits our thinking abilities,
> > providing a variety of arguments and evidence to refute a strict
> > interpretation of Sapir-Whorf.
> >
> > On the other hand, there is an elegant argument that concepts
> > must be expressed as sentences in a mental language (viz.
> > Jackendoff, "Languages of the Mind", 1995), based on the
> > apparently unlimited number of sentences that can be expressed in
> > natural language: Since natural language sentences can describe
> > concepts, and the brain is finite, we cannot suppose that all
> > possible concepts are stored as "atomic" propositions,
> > pre-existing in the brain. This implies that concepts must be
> > represented internally within the mind as structures within a
> > combinatorial system, i.e. a language of some sort.
> >
> > So, despite the evidence and arguments against thought being
> > based on spoken natural languages, it seems clear that human
> > intelligence must rely on an internal language for thinking, a
> > "mentalese" distinct from spoken natural languages.
> >
> > In this case, spoken natural languages would be ways of
> > "serializing" mentalese concepts for communication between
> > people. The psychological experiments cited against the
> > equivalence of natural language and thought may show cases where
> > the mechanisms for spoken language are impaired, while the
> > mechanisms for mentalese continue to function, or vice versa.
> >
> > The expressive capabilities of languages like English may reflect
> > the expressive capabilities of mentalese. The ability to express
> > arbitrarily large, recursively structured sentences is plausibly
> > just as important in a mentalese as it is in English. The ability
> > to metaphorically weld concepts together across arbitrary,
> > multiple domains may plausibly be just as important in a
> > mentalese as it is in English.
> >
> > This is not to say that mentalese would have the same limitations
> > as spoken natural languages. Perhaps in mentalese, sentences can
> > have more complex, non-sequential, graphical structures not
> > physically permitted in spoken languages, which might make
> > mentalese more expressive for problems involving spatial
> > reasoning, etc. Perhaps in mentalese sentences can more clearly
> > express the kinds of logical precision possible in formal
> > languages like predicate calculus.
> >
> > The study of natural language may yield some very important clues
> > about how concepts are represented and processed by humans. For
> > example, consider the interrelated problems of disambiguating
> > polysemy and parsing natural language sentences. How do humans
> > overcome the exponential explosion of possible sentence parses
> > (which is a major problem for computational linguistics), in view
> > of the fact that the most highly polysemous words are also the
> > words used most frequently in discourse? (I like to think of this
> > as the "paradox of polysemy".)
> >
> > Phil Jackson
> > -------------------------------------------------------------------
> > "Imagination is more important than knowledge. Knowledge is
> > limited. Imagination encircles the world." - Einstein
> > -------------------------------------------------------------------
> > Standard Disclaimers. www.philjackson.prohosting.com
> >
> >
> > > -----Original Message-----
> > > From: owner-standard-upper-ontology@ieee.org
> > > [mailto:owner-standard-upper-ontology@ieee.org]On Behalf Of
> John F. Sowa
> > > Sent: Friday, February 09, 2001 4:16 PM
> > > To: pat hayes; Robert E. Kent; standard-upper-ontology@ieee.org
> > > Subject: Re: SUO: Proposed Changes to Merged Ontology
> > >
> > >
> > >
> > > Pat,
> > >
> > > Just a couple of quick comments:
> > >
> > > >My basic grumble is that there seems to be a widespread assumption
> > > >that it is just kind of obvious that linguistic data is
> > > relevant to
> > > >ontology design.
> > >
> > > I think that NL is "obviously" relevant, but I don't think that
> > > it is at all "obvious" in what way it might be relevant.
> > >
> > > >I think that for an inference-supporting ontology
> > > >this view is extremely naive, based on very daft assumptions
> > > (such as
> > > >the idea the people think in natural language) and more likely to
> > > >cause trouble than to be of use, in practice.
> > >
> > > Anybody who claims to understand how people think is a crank.
> > > That is especially true of anybody who claims that their own
> > > particular theory, device, notation, program, or whatever
> > > "thinks the way humans do".
> > >
> > > And by saying that, I do not imply that studying psychology,
> > > linguistics, or psycholinguistics is a waste of time. On the
> > > contrary, the more you study those subjects the better you
> > > understand how little you understand about how humans understand.
> > >
> > > For more on this point, see my review of the book by Lakoff
> > > and Johnson, Philosophy in the Flesh:
> > >
> > > http://www.bestweb.net/~sowa/direct/lakoff.htm
> > >
> > > John
>
>