A New Look at Language, Thought and Reality

Tom Johnston
TM1972.com
[email protected]
November 30, 2015
minor revisions January 2017

keywords: analytic, ascription, brains, concepts, declarative sentences, descriptive,

diachronic semantics, dictionaries, dispositions, sensory stimuli, facts, gestalts,
hardwired gestalts, hyponymy, information, intensional sets, kinds, language,
learned perceptual gestalts, lexical items, lexicon, memory, motor responses,
neural links, neural processes, neural states, observation language, patterns,
perceptual gestalts, physical objects, pragmatics, predicates, prescriptive, pro-
attitudes, propositional attitudes, propositions, reality, reference, rules, semantic
content, sense data, sensory stimuli, sentences, statements, theory of meaning,
things, thought, TM, tokens, types, word space.

names: Aristotle, Kant, Nowell-Smith, Quine, Whorf.

Table of Contents
A New Look at Language, Thought and Reality.................................................................1
Language.........................................................................................................................2
Thought...........................................................................................................................4
Reality.............................................................................................................................7
Language, Thought and Reality......................................................................................8
Appendix A. Early Material..........................................................................................10
[1972-02-10 - PhilNotes72-pdf]...............................................................................10
[1972-07-23 - Green]................................................................................................10
[1972-08-07c – Green].............................................................................................10
[1972-08-30 – Green]...............................................................................................10
[1972-10-00 - PhilNotes72-4-pdf]............................................................................11
[1972-10-05 – PhilNotes72-4&5-pdf]......................................................................11
[1974-12-15 – PhilNotes73-74-pdf].........................................................................11
Diachronic Semantics..........................................................................................11
[1974-12-21 – PhilNotes73-74-pdf].........................................................................12
From my Dissertation...............................................................................................12

__________________________________________________

A New Look at Language, Thought and Reality. p. 1 of 13.

(c) 2016, Tom Johnston. Non-commercial copies permitted.
I begin with a basic description of the view that I have on language, thought and reality.1
In a set of extensive notes that I wrote in the early seventies, I called this view my "theory
of meaning", or "TM" for short. I'll continue to call it that.

This description will not reference any of the relevant academic literature, nor will it
express my views on any of the currently active academic topics. That comes next.

note January 2017. This article was originally written before I began my research
into the externalism/internalism and broad/narrow mental content debates. And
my conclusions on these topics should be kept in mind in reading what I have
already written here. They do not contradict anything already written here. But the
perspective they provide is important.

Externalism and broad mental content are philosophical comments on our

commonsense folk theory whose ontology is realist and whose epistemology is
one of more or less accurate direct contact with the external world. As I
claim/acknowledge elsewhere (see the section "Alice Through the Looking
Glass", in my commentary on the SEP article "Narrow Mental Content"), this
commonsense theory is the unavoidable basis for all our talk about the world. And
yet as recent work in the neurophysiology of language (2015. Kemmerer) makes
clear, physical objects in particular, and the things we talk about in general, are, as
Quine said, no more than posits, on a par with the posits of the Homeric gods.

Internalism, and narrow mental contents realized as perceptual gestalts and

dispositions regarding co-ascriptions (see several other articles of mine at
tm1972.com), compel the conclusion that this commonsense theory is a
universally-shared projection of structures and processes which exist inside the
brains of language users. And yet the very science of the neural basis of language
is not, and cannot be, carried out except in terms of this commonsense projection.
1 Language, Thought and Reality is the title of a book by Benjamin Lee Whorf (1956), but my use of the
phrase does not indicate any commitment to (or, indeed, disagreement with) the Sapir-Whorf
hypothesis. It's just that, although Whorf beat me to it, those three words capture the scope and limits of
what I am interested in better than anything else I can think of.

__________________________________________________

A New Look at Language, Thought and Reality. p. 2 of 13.

(c) 2016, Tom Johnston. Non-commercial copies permitted.
Language
Language is a set of tags and rules for combining them. The tags are the lexicon of the
language. As for the rules, there are both context-independent and context-dependent
rules. The context-independent rules constitute the grammar of the language. The
context-dependent rules constitute the pragmatics of the language.

When lexical items are combined in accordance with these rules, well-formed sentences
are created.2 All meaningful sentences are well-formed although not all well-formed
sentences are meaningful.

Meaningfulness is important because the point of language is to express information. It is

through our ability to share information with others that we have become the dominant
species on the planet.

Information is the representation of a feature of something. We express information by

picking out something and saying something about it. In doing this, we make statements
by uttering or inscribing declarative sentences, or by uttering or inscribing fragments or
other indicators of declarative sentences.3 When things are as a statement describes them
to be, the statement is true, and otherwise is false.

True statements represent facts. Facts are what are represented by true statements. No
fact, no true statement, of course. But also: no true statement, no fact.

The something picked out by a statement may be an object or an event or a process; it

may be abstract or physical; it may exist, have once existed, may at some future time
exist, or may exist in a way that has nothing to do with time. It may not exist at all, or
may be taken to exist in some possible world other than the real world. What we say

2 So instead of using "well-formed" to mean "in conformance with grammatical rules", I use it to mean
"in conformance with grammatical and pragmatic rules". The intent is to indicate conformance with all
required constraints on meaningful combinations of lexical items.

3 For the most part, I'll use "utterance" and "inscription" interchangeably.

__________________________________________________

A New Look at Language, Thought and Reality. p. 3 of 13.

(c) 2016, Tom Johnston. Non-commercial copies permitted.
about it is that it has one or more properties, one or more relationships with one or more
other things, or has undergone, is undergoing or will undergo one or more kinds of
change.

We also make statements about statements. Some statements express our propositional
attitudes to other statements, for example, "I believe that John went to the store".

Some statements result from the negation, conjunction and/or disjunction of other
statements.

Material implication is the statement that, for an ordered pair of statements, whatever
truth-value they each have, that pair of truth-values is not, respectively, true and false.
Entailment is the statement that, for an ordered pair of statements, whatever truth-value
they each have, that pair of truth-values cannot be, respectively, true and false.

Statements must be distinguished, on the one hand, from their utterances and/or
inscriptions and, on the other hand, from their semantic content. The inscription of a
statement is a (declarative) sentence. The semantic content of a statement is the
proposition it expresses. Orthographically (phonetically) identical sentences express the
same statement; synonymous statements express the same proposition.4

Thought
Thought is the mental process by which we create, acquire and use information.
Knowledge is stored and accessible information. This information is stored as neural
states, and accessed by neural processes.

Thought is always about something. When expressed in statements, it is about what those
statements pick out.

4 January 2017. I would no longer say this. I think that the notion of a proposition, as the semantic
content which a set of synonymous statements share, has shaped the minds of analytic philosophers in
ways at least as harmful as they are beneficial.

__________________________________________________

A New Look at Language, Thought and Reality. p. 4 of 13.

(c) 2016, Tom Johnston. Non-commercial copies permitted.
Now for some specifics.

Our brain is a neural net. It is constantly changing in response to external events and
internal events. External events affect our brain by means of stimulating our sense organs.
These stimuli are photons (sight), air pressure (sound), volatile organic compounds
(smell), physical pressure (touch), and specific organic molecules (taste). For most
sensory experience (smell is an exception), what happens between the initiating stimulus
and its (conscious or unconscious) recognition in the brain is more than the simple
transduction of stimuli into neural impulses. At one or more stages in its journey towards
the brain, these signals are processed. They are organized.

Natural selection has endowed organisms with brains with useful sensory patterns. These
are hardwired perceptual gestalts. The link from a hardwired gestalt to a motor response
is often an "express lane" more or less directly from sensory stimulus to muscular
response. The frog's eye may indeed tell the frog's brain that there is a fly-like shape
moving across its visual field, at which point the frog's brain tells the frog's muscles to
flick out its tongue towards the object. A monkey gathering fruit on the ground isn't likely
to need to spend much time concluding that a moving colored shape nearby is likely to be
a tiger. Flies are hard to swat. Natural selection hardwires the gestalts each species needs
for its survival.

Other perceptual gestalts are learned. Indeed, every time we perceive a physical object, a
corresponding gestalt has been activated in the brain. This gestalt is the memory, in the
brain, of a perceived pattern of sensory stimuli, and the reception of that pattern of
stimuli activates the pattern.

Most perceptual gestalts are learned early in life. But there is one area where perceptual
gestalts are learned by adults. That is in using instruments to extend the range of our
senses. For vision, those instruments are telescopes and microscopes, and in many cases,
those who use them must learn to identify what they see. A molecular biology student,
when first using a microscope to observe different bacteria, will likely mis-identify many
of them initially. It is only through repeated trial and error, and correction by an expert,

__________________________________________________

A New Look at Language, Thought and Reality. p. 5 of 13.

(c) 2016, Tom Johnston. Non-commercial copies permitted.
that the perceptual gestalt is eventually formed. A young child may initially use "kitty"
for both dogs and cats; but with the help of his parents will quickly learn "doggie" and
the distinction between kitties and doggies. In both cases, perceptual gestalts are being
learned.

For the layperson, another way of becoming aware of perceptual gestalts is by means of
stereoscopic images. Two photographs of the same image are taken, from slightly
different angles. By unfocusing your eyes as you look at them side-by-side (which takes
some effort), a three-dimensional image will suddenly appear. There are lovely books of
stereoscopic images in which each page is two photographs of abstract colors and shapes
out of which arise a three-dimensional image, such as a mother duck leading her
ducklings to a stream.

This is also a metaphor which, I believe, is more than a metaphor. Do tables and chairs,
cars and highways, lions and sheep, men and women, quanta and black holes, really
exist? Does the real world contain these things? It does, but in the same sense that that
page in the stereoscope book contains the image of the mother duck and her ducklings.

Gestalts are usually acquired, usually after a training period, all at once. The occasion is
the "Oh! Now I see it!" experience.

Some perceptual gestalts correspond to kinds. When I see a tiger, I see an instance of a
kind. That instance is a physical object which I identify as a tiger, i.e. an instance of the
kind Tiger.

But that gestalt is not yet the concept of a tiger. For that, something more is needed.

Concepts are represented by lexical items, just as propositions are represented by

statements. In our brains, we create neural links between perceptual gestalts and the
lexical items we tag them with. The activation of either one – its introduction into
"working memory", as I shall say – will activate the other.

__________________________________________________

A New Look at Language, Thought and Reality. p. 6 of 13.

(c) 2016, Tom Johnston. Non-commercial copies permitted.
Reality
Here I make a Kantian point, but without the "transcendental", i.e. that gestalts,
perceptual and conceptual, are not fixed, immutable and necessary, any of them. Also that
my gestalts are an extension of Kant's (and Aristotle's) categories up or down a
hyponymy/ontology tree.

So do physical objects exist? I take seriously Quine's comment that physical objects are
posits, in the same sense that the Olympian gods are posits. Are there, for example,
tigers? Are there cheetahs?

Our ontology of physical objects consists of those types that we find it useful or
interesting to distinguish. But what does it mean to distinguish a type of physical object?
Well, if I distinguish tigers from cheetahs, but you do not, and call them both tigers, what
is going on? Am I making a distinction among really distinct types of things, whereas
your conceptual apparatus is not fine-grained enough to enable you to recognize that
distinction? Or are the animals I call cheetahs really tigers, and I am drawing a distinction
that may make sense to me, but that doesn't correspond to any real distinction in the
animal kingdom?

In this case, we may be inclined to say that the difference between us is "merely verbal", I
having a richer vocabulary, in this area, than you do. But suppose that we both believe
that there are tigers and also cheetahs. I pick out felines in Africa in a way that those
familiar with these animals recognize as correct, while you insist on calling a certain kind
of leopard a cheetah, along with those animals we both agree are leopards.

Does the real world contain these two types of animals? And do some individual big cats
fall into one or the other of these types? Really?

What is the difference between a coffee cup, a coffee mug, and a bowl used to drink
coffee from?

Stepping back for a moment, here's what's going on. We find it useful to recognize

__________________________________________________

A New Look at Language, Thought and Reality. p. 7 of 13.

(c) 2016, Tom Johnston. Non-commercial copies permitted.
different kinds of things. Physical objects are basic among them because our interactions
with them can enhance or degrade our chances to survive and reproduce our genes. So
evolution has hardwired into all animal brains certain basic perceptual gestalts. In some
cases, with such simple organisms as krill, for example, these hardwired gestalts may be
all they need to identify predators and swim away from them, and to identify food and
swim towards it. In neurally more complex organisms, higher-level perceptual gestalts
are formed. There are gestalts for objects, and for the movement of objects.

For the organisms with those gestalts, each triggering of the gestalt is an encounter with
the stimuli that triggered the gestalt, or with a memory of such an encounter. For
language users, we tag the gestalt with a lexical item. That item is associated with the
type of which each instance of activation of the gestalt is interpreted as a present,
remembered, anticipated or imagined encounter with a physical object which is an
instance of that type.

Language, Thought and Reality

Language is the expression of thought. Its primary function is to express information. But
language is constantly changing. Lexical change is reflected in the periodic revision of
dictionaries. These changes record changes in the meanings of words and expressions.

And, now and for the most part, I am more interested in lexical change than in changes in
grammar or in such pragmatics phenomena as prosodic and paralinguistic conventions.

Given that language and thought exist in human brains, which are constantly changing,
the basic unit of both is a brain at a point in time. A word or other expression has meaning
for a given person, at a given point in time. A statement may represent a different
proposition/fact for the same person at different times, or for different people at the same
time.

So dictionaries represent the best guess of acknowledged expert language users of what
the defined words in the dictionaries mean, more or less, to the majority of competent
users of that language, at the then-current time. Dictionaries can be prescriptive to less

__________________________________________________

A New Look at Language, Thought and Reality. p. 8 of 13.

(c) 2016, Tom Johnston. Non-commercial copies permitted.
competent users because they are descriptive of definitions generally accepted by more
competent users.

__________________________________________________

A New Look at Language, Thought and Reality. p. 9 of 13.

(c) 2016, Tom Johnston. Non-commercial copies permitted.
Appendix A. Early Material
Note: all headers and page numbers reference (2015. Johnston. Transcriptions of
Philosophy Notes: 1972 – 1979).

[1972-02-10 - PhilNotes72-pdf]
The experience of seeing, say, a table, is not simply the having of certain sense
impressions; in order for certain sense-data to elicit the experience of seeing a table, a
perceptual gestalt organizing those sense-data in certain ways, is necessary. Without it,
there is no experience of seeing a table. (p.19)

[1972-07-23 - Green]
.....perceptual gestalts are associated with words. The most frequently used ones extend
over a clearly-defined range (not of possible objects, but of objects we usually
encounter). This is ..... reference.

.....words have analytic ties such that given that one word had been asserted in a given
context, other words cannot be denied to hold, and others cannot be asserted to hold.
(And this is a continuum, not a dichotomy.) This is ..... sense. (p.56)

[1972-08-07c – Green]
A sentence is analytic for a given person on a given occasion to the degree that he would
not allow empirical evidence to count against its truth-value, on that occasion. (p.73)

[1972-08-30 – Green]
The observation language becomes associated with perceptual gestalts. The combination
creates a sense of that language being purely descriptive. With common-sense
observation language, the association took place at too early an age for us to remember it
occurring. But with the conscious adoption of other observation languages (in science,
among art critics, or any profession), we can often catch the associations forming.

Initially we use the observation language, but must “figure out”, i.e. go through a time-
extended process of reasoning – in order to correctly use the language. But after awhile,
we don't have to anymore. We suddenly “see” a phenomenon, and realize that a certain
element of the observation vocabulary refers to it. Then, with that element, we no longer
have to “figure out” when to use and not use it. (This association accounts for many
prejudices, and the strength with which people adhere to them.)

The gestalts formed are, of course, a function both of factors external to the subject, and
also neurophysiological biases within the subject – of the external as well as of the
internal, world. Thus a complete non-overlap of sense organs would produce a great
divergence in observation language, and in gestalts. It might not be possible to

__________________________________________________

A New Look at Language, Thought and Reality. p. 10 of 13.

(c) 2016, Tom Johnston. Non-commercial copies permitted.
communicate across such gaps, unless instruments could extend sense organs into
overlapping regions. (pp.79-80)

[1972-10-00 - PhilNotes72-4-pdf]
Perceptual input from analytic sentences activate those dispositions to respond to concept
pairs which constitute the meaning of concepts. Thus analytic sentences are a way of
getting at the psychological factors which constitute meaning. This is the most important
connection between analyticity and meaning. (p.86)

[1972-10-05 – PhilNotes72-4&5-pdf]
A change in ontology is a change in the observation language's referring expressions. A
change in interpretation is a change in the theory's referring expressions. The difference is
psychologically based only. (p.103)

[1974-12-15 – PhilNotes73-74-pdf]

Diachronic Semantics
i) The extension of a concept shifts – the set of objects to which it is applied changes
membership. The causes are well-known, analogy being the principle one.

ii) The disposition to disallow the assertion of the concept of an object or object-kind
weakens (these are the objects moving into the set), and the disposition to disallow it of
others strengthens (these are the objects moving out of the set). (And, of course, we can
use "allow" as well as "disallow" to characterize these dispositions.)

iii) As extensions shift, concepts form new extensional pairs (pairs in which the necessary
relation of containment of one in the other obtains), and break their pair-relation with
older concepts. As the range of "A" and "B" shifts, "All A are B" begins to admit of
exceptions. When an exception is acknowledged, the disposition to disallow a non-B A
has faltered. The faltering, over a sufficiently long period of time, signals the eradication
of the disposition. Conversely, new dispositions are formed, as the new extensional
relationship entrenches.

iv) The new disposition ensures the continued dominance of the new pattern; the
perceived dominance of the new pattern contributes to the formation of the correlative
disposition. The new pattern is thus self-sustaining, which is why so few instances of
synthetic statements are to be found; they transform themselves into analytic statements
by this reinforcing mechanism of the formation of the disposition.

v) The new pattern establishes itself within socially dominant groups over a sufficiently
long period of time. Ceremonies of semantic rule change take place, as new editions of
dictionaries and other semantic guides (Eric Partridge, for example) revise their rule
entries to conform to the new pattern and dispositions.

__________________________________________________

A New Look at Language, Thought and Reality. p. 11 of 13.

(c) 2016, Tom Johnston. Non-commercial copies permitted.
vi) Given that changes in meaning are logical, not physical, phenomena, these
ceremonies are the occasions on which semantic changes actually take place. (pp.123-
124)

[1974-12-21 – PhilNotes73-74-pdf]
The processes by which semantic rules change under the pressure of changing patterns of
usage are physical and sociological, not logical, processes. (p.127)

From my Dissertation
Few predicates in a natural language have an exact meaning. The principal reason, as
Nowell-Smith observed, is that when we talk about the meaning of a word, we are talking
about the use of the word extended over all the speakers of the language who use the
word and over the extended period of time during which the word is commonly used. The
patterns of predicative behavior, unevenly spread over this large two-dimensional space,
are what ultimately constitute "the meaning" of the word.

Each token in this "word space" has a unique location. It also has a set of characteristics
which constitute its meaning. In my own view, this set of characteristics is the set of other
predicates whose ascription, if queried, would be said to follow from the ascription of the
given predicate. This alone would explain why the meaning of most predicates in a
natural language is not an exact thing. Even allowing for the rejection of a large number
of tokens on the basis of semantic anomaly, it would be surprising if the remaining token
meanings, spread over such a large word space, were exactly the same.

But the blurring of the boundaries of a predicate's meaning is due to more than the sifting
out of a common denominator of token meaning from as large a number of tokens as
possible. For consider the set of predicates whose ascription, if queried, would be said to
follow from the ascription of the given predicate. These predicates can be arranged
according to the strength of the "following from" relation. At one end of the sequence are
those predicates whose non-ascription would be inconceivable, given the ascription of the
given predicate. At the other end are those predicates whose non-ascription would merely
seem somewhat odd or peculiar. For any given predicate, the predicates in this
"intensional set", as I call it, are not usually distributed evenly; they tend to cluster. A
common-sense belief about meaning is that we can bisect this set, distinguishing
predicates at the strong end from those at the weak end; in non-technical parlance, we
talk about a word's denotation and about its connotation. In place of this, I distinguish
"layers" of meaning corresponding to the predicate clusters within the intensional set.

The predicate clusters can, in theory, be seen pretty clearly in the intensional sets for each
token. The boundaries of these clusters are occasionally obliterated, but more often
blurred, as we move from tokens to type.

So even at the token level, the "great dichotomies" of analytic vs. synthetic and the
various "world vs. words" dichotomies come to nothing more than arbitrary bisections of
token level intensional sets. In place of them, we can say only this: to the extent that a
predicate in the intensional set is close to the strong end of the "following from"

__________________________________________________

A New Look at Language, Thought and Reality. p. 12 of 13.

(c) 2016, Tom Johnston. Non-commercial copies permitted.
sequence, the witholding of its ascription is a semantic mistake; to the extent that the
predicate is close to the weak end of the sequence, the witholding of its ascription is a
matter of peculiarity in "pro-attitudes" or in beliefs about the world.

At the type level, there is even less justification for the imposition of semantic
dichotomies. For at that level, a predicate in the intensional set occupies not a specific
point in the sequence of predicates, but rather a range along that continuum from strongly
following to weakly following, a range which will usually overlap the ranges of several
other predicates. So it does not merely seem difficult, as Hare suggested, to use the
analytic-synthetic and "world" vs. "words" dichotomies. (pp.131-133.)

__________________________________________________

A New Look at Language, Thought and Reality. p. 13 of 13.

(c) 2016, Tom Johnston. Non-commercial copies permitted.