Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

CONTENTS

I. PRESCRIPTIVE METAPHYSICS I

2. THE LOGICAL CHARACTERISTICS OF


METAPHYSICS 4

3· MODELS TO MECHANISMS I8

4 . EXISTENCE-CLAIMS 42

5 · OBJECTS 54

6. ONTOLOGICAL EXPERIMENTS 67

7 · PYTHAGOREAN INDUCTION 7I

8 . THE PARADIGM-CASE ARGUMENT


CONSIDERED 77

g . Two CAsE-HISTORIEs 83

I O. APPLICATION OF THESE IDEAS TO A


CURRENT PROBLEM 93

I I. PHILOSOPHICAL BACKGROUND • I06

REFERENCES 114
I. PRESCRIPTIVE METAPHYSICS

Metaphysics is an old subject, and some


philosophers would also want to say that it is
decrepit if not altogether defunct. Since I
respect the views of many who would make this
judgement, I want to prefix to my attempt at
some modern µietaphysics a defence and an
explanation of what I am setting out to do. I do
not think that my argument· can secure con­
viction against the run of anyone's metaphysical
predilections, nor do I think that any sudden
enlightenment will necessarily be vouchsafed to
anyone who cares to follow it. Large claims
were made for traditional metaphysics, so large
that the subject could not meet them and was
declared bankrupt. If my claims are smaller, I
at least hope to be able to fulfil them. Recently
an old branch of metaphysics has been revived
by Mr P. F. Strawson under the name "descrip­
tive metaphysics", in his book Individuals. Des­
criptive metaphysics is concerned with very
general features of the conceptual scheme we
ordinarily use; in particular it is concerned with
the categories of existence. This is sometimes
said to have been the task of traditional meta­
physics-the construction of a science of being.
The mistake of traditional metaphysics was not
so much to say that metaphysics is about
what exists, but in supposing that it was a
science. This is completely to misunderstand its
role in intellectual constructions. Descriptive
THE ORIES AND THINGS
metaphysics concerns itself with the existence
presuppositions and priorities of our conceptual
schemes as they are; prescriptive metaphysics
with the existence presuppositions and prior­
ities of conceptual schemes we might or should
adopt. In neither branch are we doing any kind
of science. You can take or leave a metaphysical
view in a way in which you can't exercise
options in science.
It is my belief that useful and respectable
prescriptive metaphysics arises naturally in the
discussion of the sciences. Incidentally, this was
where Aristotle put the subject-in the discus­
sion of the presuppositions of the science of
physics. In this present study I want to concern
myself with two systems of prescriptive meta­
physics: positivism, and another theory, which I
shall call the doctrine of ontological depth. My
method will be to try to make clear what posi­
tivism is by expressing it, not as a system, but in
a variety of principles. Against these principles
I shall set up principles in accordance with the
theory of ontological depth. I shall then produce
several "layers'' of necessarily less than con­
clusive arguments designed to weaken positiv­
ism and to strengthen the doctrine of onto­
logical depth. I shall first briefly try to show
that this is the only possible method in prescrip­
tive metaphysics. My intention is to continue
this process of contrasting principles and build­
ing up arguments until I feel that enough· has
been said to recommend the doctrine of onto­
logical depth and to show up positivism un-
P R E S C R I P T I VE MET A P H Y S I C S 3
tavourably by contrast. Metaphysical systems
cannot be proved, nor can they be refuted; but
this does not mean that they cannot be adopted
or �ejected under a weight of rational dis­
cussion.
The peculiar logical characteristics of meta­
physical systems derive from the place that
these systems hold in our thinking. A meta­
physical system is a construction designed to
force the attachment of existential priority to a
certain class of entities. What- does this mean?
It might mean a number of different things.
For instance, if you were to decide that the
fundamental existences were sensations, what
philosophers call sense-data, then one effect of
your adherence to this doctrine as a meta­
physical theory would be that all perception
statements would be, by you, reduced, for pur­
poses of verification, to statements about the
relations of sense-data. For instance, instead of
talking about tables you would talk about
brown, paralleloid sense-data. To argue for a
certain direction of reduction among the kinds
of things which we ordinarily do not dispute
exist, is to produce a metaphysical theory. To
produce reasons for adopting one direction or
another of reduction and one or more basic
kinds of object is to do prescriptive metaphysics.
Scientists are often engaged in prescriptive
metaphysics. For instance, if someone wants to
get all the entities that his science recognizes,
defined, say, in terms of spatio-temporally
located electri�al charges, then one way of
4 .T H E O R I E S AND T H I N G S
expressing this reductive preference is to argue
that only electric charges exist. This move is
prescriptive, if, until the argument for this
reduction is adduced, the science in question
tends to define its concepts in terms of some
other class of entities.
When these prescriptions begin to have an
effect outside the technical convenience of some
science or other then we begin to have a kind of
prescriptive metaphysics which must become the
concern of philosophers. Sooner or later the
prescription is going to impinge on conceptual
schemes other than those used in the original
science. (Compare, for instance, the way the
conceptual recommendations of Darwin about
the way we should define species spread over
even into moral and political philosophy.) This
spread is found in the history of positivism too .
Positivism has always been a vigorous element
in physical speculation in one form or another,
since there is always a strong feeling for admit­
ting nothing but the observable to our onto­
logy. The doctrine has been so vigorous that it
is always breaking out of its natural confines
into other fields. I want to argue that the doc­
trine of posivi tism is by no means even a
necessary feature of physics.

2. THE LOGICAL C HARACTERISTICS O F META­


PHYSICS
All metaphysical schemes are logically im­
peccable. This follows from the way they are
LO G I C AL C H A RA C T E R I S T I C S 5
constructed. To see this let us examine a simple
model. Suppose there is a universe where there
are two kinds of things held to exist by the
inhabitants, Xs and rs. There are several meta­
physical possibilities. It might be alleged that
only Xs are real (exist) and that rs have to be
understood in terms of them. Everything is then
reduced to Xs and constructions out of Xs. But
nothing could refute this theory, "for the world
still presents the appearance of consisting of Xs
and rs. All that is different is t�e inhabitants' way
of understanding rs. Another possibility is that
both types of object should be held to be illusory
and only some third kind of thing, ..(s, held to
exist. But just the same argument to the truth
and irrefutability of this theory holds. Both
theories can be "true" in one sense, i.e., empir­
ically they can't be refuted, logically they are
equally coherent; how then can we ever get rid
of a particular metaphysical theory?
Metaphysical theories of this prescriptive
kind can collapse in several different ways.
( I ) They may cease to be fruitful. To suggest
that matter is really (ontological priority claim)
made up of spatio-temporally located electrical
charges has consequences, though it cannot
itself be established or refuted. For instance, we
should expect to find an electro-magnetic
susceptibility in all matter, etc., etc. (2) Some­
times they fail because of "ontological dullness".
It may well be true that all the statements of the
physical sciences can be reduced to statements
about pointer-readings, but this claims that
6 TH E O R I E S A N D T H I N G S
there is only one ontological class. It is a very
peculiar one, because it includes nothing that is
in the natural world, only items that are arte­
facts or the properties of artefacts. There have
been several such positivist doctrines, but none
of them has survived. The reasons for their lack
of permanent following we shall see in what
follows. (3) They may fail through theoretical
inelegance. It may turn out to be theoretically
more elegant to admit Xs and rs and omit .(s
rather than make this reduction. Sometimes it
is otherwise; and the reduction gains great
power of conviction from this source. I call this
process of quasi-aesthetic selection, Pythagorean
Induction.
If metaphysics does consist in existence­
discussions, what sort of discussions are these?
We must specify what it is we are going to
understand by "existence" (and any other
member of the "ontological" family; "reaP',
"genuine", "basic" . . . ) . I shall simply lay down
my criteria; and so determine the style of this
discussion. No doubt the criteria could be laid
down in many other ways. If this way of laying
down existence criteria does not appeal to you,
a problem sets itself for the reader--do my
arguments in what follows fit your criteria too?
I shall define "existence" in terms of "find­
ing" and "fulfilling". Existence may be attrib­
uted to something when it fulfils the set of
criteria below-and what it means to attribute
existence to something is that these criteria have
been satisfied by a unique individual. We �ust
L O GIC A L C H A R A CTE RISTIC S 7
then first specify some criteria for "unique
individual" . In the style of Strawson, I shall say
that a unique individual is that part of a
domain which (i) can be identifyingly referred
to; and which (ii) can, with the help of the
same identification procedure, be re-identified;
and which (iii) can, with the help of the same
identification procedure, be discriminated from
all other parts of its domain . For instance, we
say that the moon is a unique individual
because :
(i) It can be described in such a way that it
is the only object fulfilling the description (or
pointed to in such a way that it is the only
object in the direction indicated; and here it
doesn't matter for some purposes whether the
pointing is done with a �nger or with a radio­
telescope) .
(ii) Each month it can be re-identified by
the use of the same identification procedures.
(iii) It can be discriminated from other
heavenly bodies by its track, its size, appearance
and so on.
The criteria for existence, then, are as follows :
An object, 0, exists if and only if unique
a
individual can be found which fulfils a
description of 0, sufficiently detailed to
enable it to be discriminated from other
members of its domain.
For example, provided there are heavenly bodie�
(Empedocles for example, denied this) , the
earth's one natural satellite exists.
8 T H E O R I E S A N D TH ING S
According to this criterion, which includes
the unique individual criterion above, o�jects
emst only in domains. Once 1a domain is
accepted then one can ask whether or not such­
and-such an object e�ists. Here we get our first
characterization of positivism :
Posi.tivism is a doctrine according to which there is
on!J one acceptable domain.
Of course, as there are different choices of
unique domain, so there are different varieties
of positivism. Prescriptive metaphysics is con­
cerned with the advocacy or condemnation of
domains. The evidence brought forward in this
trial may be very varied; from reductive
Pythagorean inductions to assertions of incon­
sistency, claims of convenience, claims of in­
corrigibility, etc., no one of which could secure
conviction. However, when taken together such
considerations have often made people raise or
lower the existence status of a domain (e.g.,
Gaiileo on colours, Occam on universals, Hume
on necessary connections, Moore on sense-data,
Descartes and Ryle on mind-substance) . Such
a domain I shall call an ontological class or kind.
By object I shall henceforth mean an attested
member of an accepted ontological class; and
by entity, I shall mean that to which a unique
reference can be made, though it is not neces­
sarily either an individual (re-identification) or
an attested member of an ontological class. For
instance, "after-image of my window seen by
me at 12. 30 p.m. today" is an entity.
LOGI CAL C HARACTERISTICS 9
There are powerful arguments in Strawson's
Individuals, as there are in Kant's Critique of Pure
Reason, fo r the contention that we do in fact
take our ontological start from spatio-tempor­
ally locatable entities. I shall take for granted,
then, that ordinarily our ontology proliferates
from material objects; and t�at a doctrine that
refuses them the status of an ontological
class needs to be argued for, while a doctrine
that accepts them as an ontological class
does not need arguing for-that far, at any
rate.
Before we turn to the detailed examination of
various forms of positivism it will be useful to
run over the arguments of this study in outline .
The logical form of the individual arguments in
what follows is mostly that called modus tollens.
An argument by modus tollens goes something
like this : first you establish that from some
proposition, say, "There is an anti-cyclone over
the North Sea", another proposition follows,
say, "The weather is sunny in East Anglia." If
it turns o ut to be raining in East Anglia then
the proposition "The weather is sunny in East
Anglia" is false. Now, this falsifies the conse­
quent of "If there is an anti-cyclone over the
North Sea then the weather is sunny in East
Anglia", which is the simplest general proposi­
tion allowing us to infer the East Anglian
weather from the North Sea meteorology. But
since the consequent is false, we can conclude
that the antecedent is false-that is, we can
conclude that it is false that there is an
IO T H EOR I E S A N D T H I N G S
anti-cyclone over the North Sea. Such an argu­
ment is in the modus-tollens form. In general, if
p implies q
and
not-q
then it follows that
not -p
In the form in which the argument is used in
this study it goes something like this :
p (a certain form of positivism, the philo­
sophical position under discussion) :
p implies q (where q is some co�sequence
for the theory or practice of science that
follows from p) :
not-q.
The consequence q is shown to be in some way
unwelcome, either because it is inimical to
scientific theory or because it runs counter to
scientific practice . We then use the argument
modus tollens to deduce
not-p.
And so we reject the form of positivism under
discussion.
The common form of all positivist doctrines
is that they recommend that for the purposes of
science we adopt only one ontological class.
Only one sort of thing is, it is claimed, legi­
timately said to be in existence. All other sorts
of things either don't exist at all and have at
L O G I C AL C H A R A CTE R I S T I C S Il
best psychological or heuristic value, or are
really constructs out of members of the accepted
ontological class. Certain things are given, all
else is, for positivism, non-existent. For ex­
ample, one form of positivism takes the given
ontological class to be sensations- ·that we are
having a certain sensation at a given instant
cannot be refuted-and hence any belief that
we may have that the things which cause or
generate the sensations exist is a more or less
hazardous inference from restricted evidence.
The doctrine which I wish to recommend in
this study, what I have called "ontological
depth", has a much less stringent existence
requirement. According to that doctrine we are
entitled to say, for instance, that both sensa­
tions and material things exist. We shall see in
Section 5 that the minimum criteria for exist­
ence go far beyond sensations and permit us to
include among the things which exist for certain
much more than any positivist would allow.
Indeed, we shall see that if we admit sensations
then the same eX!istence criteria which admit
them to our ontology also admit material
things, magnetic fields and things of many
different ontological kinds. We shall see, too,
that just because there can be a magnetic field.
and a material thing at one and the same point
in space-time, this is not a ground for claiming
the existence of only one of these.
In general, positivist forms of science take as
their ontological class what a scientist is given.
This usually turns out, in the less stringent
H? T. H E O R I E S A N D T H I N GS
forms of positivism, to be what can be observed.
It is not always clear whether a positivist
would admit that observation with the help of
instruments enriches the given ontological
class. In the more stringent forms of positivism
observations are restricted to various immedi­
ate data. For instance, Eddington reduces the
given to pointer-readings, and out of these the
universe is constructed. Bridgman has a slightly
less stringent requirement. For him concepts
only have empirical meaning-that is, refer to
members of the acceptable ontology-if this
meaning can be given in terms of operations of
measurement. Time, for instance, is the opera­
tions of clocks; length is the number of times a
yardstick can be laid down in a defined set of
operations.
The first step in my arguments in Section 3
against positivist views of science is to consider
what principles of scientific methodology emerge
from these doctrines. According to the prin­
ciple of ontological depth, when we invent
imaginary mechanisms to account for pheno­
mena, be they pointer-readings or observations
in general, we are entitled, under some condi­
tions, to ask whether the imaginary mechanism
or model exists. To do this we have to devise
experiments which go beyond what we are
given, either by extending an ontological class
we already have (when, say, we extend the
class of micro-organisms to include viruses); or
by including in our ontology an altogether new
ontological class, as Maxwell and Hertz did
LO G I CAL C HARAC T E R I S T I C S I3
when they respectively predicted and discov­
ered electro-magnetic radiation. The first case
-extending some given ontological class­
might be just acceptable to less stringent forms
of positivism, but it is hard to see how the
doctrine of the "given" could accommodate the
second case. I shall argue, then, that the
methodological principle of positivism, accord­
ing to which concepts originally of a purely
theoretical kind cannot acquire existential
status, is false. .
My method of reaching this conclusion is to
show that there are certain statements which
necessarily occur in a theory which are mis­
construed by positivism. These statements are
those which link our imaginary mechanisms
with empirical concepts. For instance, such a
statement is:
Gas pressure is rate of change of momentum
of gas molecules.
According to positivism, rate of change of
momentum of gas particles has to be identical
with gas pressure and has no empirical meaning
independently of that empirical concept. That
is, attempts to measure the rate of change of
momentum of gas molecules independently of
gas pressure are a waste of time. This has the
ontological consequence. that molecules have
either no existential status at all, or at least a
weaker claim to existence than manometric
operations. Admittedly no ·practising scientist�
however positivistically inclined in theory,
14 T H EOR I E S A N D T H I N G S
could ever go to these lengths about molecules,
but consistency would demand that he should.
As we shall see, the theoreticians of quantum
mechanics do go this far, denying existence to
properties and/or entities between successive
observations. I shall argue that the kind of
statement above, in which gas pressure is
equated to change in momentum, is not an
identity but a synthetic empirical statement. It
begins, perhaps, as the statement of an ana­
logy, but takes on a more and more �mpirical
character as the possibility of demonstrating
the existence of its imagined terms becomes
more feasible. That is, I shall argue that there
are experiments by means of which models are
transformed into mechanisms.
How do we do this? To answer this1we need
to go a bit deeper into the logic of existence­
claims. There are two cases: extension of a
given ontological class to new members, and
the discovery of a new ontological class
altogether.
(a) When we look into the logic of existence­
claims of the former kind we find that they
depend upon establishing some kind of con­
tinuity between observations of things accepted
in our ontological class and observation of new
members. I define a number of such continu­
ities-optical family continuity (for instance
when we move from a drop of blood to a virus
infecting a corpuscle, by progressively increas:
ing the magnification), kinaesthetic family
continuity, etc. It turns out that when we are
L O GI C A L CH A RACTERISTICS 15
extending an ontological class (though theory
was originally required both in postulating the
existence of the entity which we attempt to
discover, and the instruments by which we
discover it) the setting up of a family contin­
uity-for instance, by means of microscopes­
eliminates the theoretical factor. We do not
need, then, to be concerned about the possi­
bility of being forced to revise the extension of
our o"ntological classes every time we revise our
theories. For instance, modern revi5ions of
optical theory do not lead to our becoming
uncertain about the existence of bacteria, even
though we can only see them in a microscope
whose principles of operation are theoretical
and have changed radically twice since the
microscope was invented. Why this is so we
shall see in our detailed discussion of family
continuities.
(b) Theory sometimes leads us to postulate
a new kind of object-an electromagnetic wave
for instance, or a genetic factor. We come to
think that the model we have constructed to
help us understand certain phenomena really
does exist. We come to think this when the
model has been very successful in accounting
for what we know and for predicting new
phenomena of a kind with which we are
already familiar. When we find that the theory
begins to predict phenomena of a new kind,
then we are 'often forced to consider how far our
model 'is real. I attempt to show that there are
two main criteria which have to be satisfied
I6 T H E O RIE S A N D T HIN G S
before we can make out an existence-claim to
a new kind of object. They are:
( 1 ) Is the entity in question spatio-tempor­
ally locatable? This turns out to be the weaker
of the criteria.
(2) Do we have experimental grounds for
some stable expectations of the entity's behavi­
our? This is the stronger criterion. Sometimes
we are quite prepared to give spatio-temporal
location to an entity, but withdraw an exist­
ence-claim because of our lack of stability of
expectation of its behaviour. This is the case at
present with some sub-atomic particles.
The discussion up to this point concerns the
possibility of and criteria for making onto­
logical experiments . However, there are other
criteria than experimental grounds for saying
that things don't exist. And these affect our
ontology. Two famous criteria that have been
used in the past are:
(i) Occam's Ra zor: Don't invent any more
entities than you really need for an explanation.
(ii) Einstein's Chopper: The simpler a, theory
the more acceptable it is, provided that it
accounts for all the facts; that is, don't invent
any more processes than you really need for an
ex;plana ti on.
I attempt to show that both these principles
derive from a yet more fundamental principle­
Pythagorean Induction. According to this
principle theories are to be constructed with the
maximum elegance.
The last ontological principle that I discuss
L O G I C AL C H A R A C T E R I S T I C S 17
is the paradigm-case principle. This is yet more
general than either Occam's Razor or Einstein's
Chopper. According to this principle there
must be something in existence for every mean­
ingful referring expression in a language to
refer to. I argue against this principle on the
grounds that it is too general, for in its baldest
form it can only be true if we determine whether
a referring expression is meaningful by seeing
whether there is anything to which it refers, and
this brings us back to the ontological experi­
ments we have already discussed.
I then examine two case-histories of famous
ontological experiments in some detail. I use
these to try to bring out the implausibility in
any positivistic interpretation of them. I hope
then to have secured the conviction that a'V'
one-level ontology, which is the general form
of posivitism, is inimical to science. Hence
positivism does not have the prestige which it
pretends to borrow from that source.
The last step in my argument is to apply all
these ideas to a problem in the philosophy of
science that is, in one form or another, peren­
nial, and which depends upon hidden positivist
principles. What are we to say when we have
two explanations that between them cover the
known facts, or both cover the known facts
equally well, but which depend respectively on
incompatible models? Are we to say that this
situation marks the end of ontological experi­
ments, or are we to hope for a unified theory?
The example I choose is quantum theory,
18 T H E O R I E S AN D T H I N G S
though the same discussion was carried on in
the sixteent� century around the ontological
status of Ptolemaic and Copernican models of
the solar system. In that case it was resolved in
favour of a modified form of one of the com­
peting theories; how we may hope for a similar
outcome in quantum theory I suggest at the end.

3· MODE LS TO MECHANISMS
When Abdul Khayyum began public bus
services to remote areas of the North-West
Frontier Province certain mullahs objected on
the ground that the absence of horses drawing
the vehicles implied the presence of djinns under
the bonnet. They were induced to withdraw
their objections by being shown the actual
mechanism by which buses were propelled.
This story illustrates a fundamental principle
of scientific theory construction:
If you don't know why certain things
happen then invent a mechanism (in accord­
ance with the view you take of how the
world works)-but it is better still if you find
out how nature really works.
Let us call this principle PI. This principle is,
in effect, a statement of the doctrine of onto­
logical depth. It clearly presupposes that we
can assume and tolerate the existence of things
not immediately given in observation, but
which are in several different ways the condi­
tions, causes or determiners of those observa-

You might also like