| |
Peirce,
Retroductive Logic, and Semantical Constraints in
Discovery
Hanson was influenced by Charles S. Peirce,
but he did not accept Peirce's views on observation.
In his "How to Make Our Ideas
Clear" (1878) Peirce set forth his pragmatic
maxim, which says that our conception of the
practical effects that we conceive an object might
have, is the whole of our conception of that object.
He distinguishes observed facts from
judgments of fact, and says that observations have
to be accepted as they occur, while judgments of
fact are controllable.
According to Peirce's theory of scientific
discovery, hypotheses are judgments of fact
expressed in propositions, and all such propositions
are additions to observed facts that are sense
impressions of singular events associated with
particular circumstances.
That which is added to observed facts by
propositions Peirce calls practical knowledge, and
it is something that is controllable and subject to
error.
Hypotheses are the result of inference, and
Peirce distinguishes inductive and abductive types
of inference.
Abduction (which Hanson also calls
retroduction) involves both formulating of
hypotheses and then selecting of one hypothesis by
testing its ability to account for surprising facts.
The difference between abduction and
induction is that the former involves guesswork and
originality, while the latter only tests a
suggestion previously made.
Once the hypothesis is formulated, abduction
is an inference that satisfies the following form:
1) a surprising fact, C,
is observed; 2) if A
were true, then C
would be a matter of course; 3) hence, there is
reason to hypothesize that A
is true.
This is actually a logical fallacy known as
affirming the consequent clause of the conditional
statement.
Peirce says that Kepler's development of his
three laws is the greatest piece of retroductive
reasoning ever performed.
He rejects J. S. Mill's view that Kepler
merely generalized on Tycho's data, and that there
was no reasoning in Kepler's procedure.
Peirce maintains that at each step of
Kepler's investigation, Kepler had a theory which
approximated the data, that Kepler modified his
theory to make his theory closer to the observed
facts, and that the modifications were never
capricious.
Hanson adds that given a choice between two
hypotheses, the simpler is preferable, where
simplicity is to be understood not as a logical
simplicity but rather as an instinctive simplicity,
because unless man has a natural bent in accordance
with nature's, he has no chance of understanding
nature at all.
In his chapter on theories in Patterns
of Discovery Hanson rejects both the hypothetico-deductive
and the Positivists' inductive accounts of
scientific discovery.
He rejects the inductivist thesis that
scientific theories are developed by an enumerating
and summarizing of observable data, as the
Positivists maintained for the development of
empirical generalizations; he states that empirical
laws explain, they do not simply summarize.
He also rejects the hypothetico-deductive
thesis that scientists start from hypotheses for the
development of theories, as Popper maintained; he
says that scientists do not start from hypothesis,
but rather they start from data.
The initial inference is not from higher
level hypotheses to observations, but the other way
around. The article setting forth his most mature
views on retroduction is "Notes Toward A Logic
of Discovery" in Perspectives
on Peirce (ed. Bernstein, 1965), which includes
summaries of Hanson's earlier papers.
The logic of retroduction pertains to the
scientist's actual reasoning, which proceeds from an
anomalous situation to the formulation of an
explanatory hypothesis that fits into an organized
pattern of concepts.
In Patterns
of Discovery Hanson refers to the pattern of
concepts as a conceptual gestalt, which functions to make the anomalous situation appear
intelligible.
The conceptual gestalt
supplies the semantics for the theory or hypothesis.
In Hanson's philosophy the semantics of
observation is variable, while in Peirce's it is
fixed and uncontrollable.
In “Notes...” he says that the formal
criteria for the retroductive logic of discovery are
the same as those for the hypothetico-deductive
logic of explanation.
They both contain the same elements: a
hypothesis, statements of initial conditions, and
the conclusion deductively derived from the
hypothesis and statements of initial conditions.
One difference between them is the direction
of the inference.
In the hypothetico-deductive logic the
inference is from the hypothesis and statements of
initial conditions of an experiment, to the
statements describing the observed outcome of the
experiment as a conclusion.
This process is used for experimental
testing, and if the results are not anomalous, it
also serves as the logic of the explanation of the
resultant phenomenon.
But in the retroductive logic the direction
of inference is in the opposite direction.
The statement reporting an observed
experimental outcome describes an anomaly relative
to what is expected, and the problem is one of
finding the hypothesis capable of functioning in a
hypothetico-deductive account that will explain the
anomalous situation as occurring as a matter of
course.
But the difference between the hypothetico-deductive
and the retroductive types of inference is not just
a matter of the directionality of the inference.
They are also different because the former is
determinate, while the latter is not.
In hypothetico-deductive inference consistent
premises must produce consistent and unique
conclusions, while in the retroductive inference
there may be many alternative and mutually
inconsistent hypotheses that are able to explain
deductively the formerly anomalous test outcome from
the same set of statements of initial conditions.
From this nondeterminate character of
retroductive inference Hanson concludes that
retroduction cannot yield a uniquely specific and
detailed hypothesis.
But he maintains that it can yield an
indication of the type of hypothesis that is most
plausibly to be considered as worthy of serious
attention.
And the decision about what type of
hypothesis is the most plausible depends in turn on
the structure of presently accepted theories and on
the shape of the most reliable conceptual frameworks
that highlight hypothesis types for the problem
solver.
Therefore, much as it is only against the
background of the intelligible and the conceptually
comprehensible offered by existing theories that the
anomalies stand out at all, so it is also in these
same terms that the scientist comes to know which
types of hypotheses will do the job and which do
not.
Reflection on this analysis reveals why
Hanson defends the Copenhagen interpretation,
understood as the semantics that is defined by the
formalism of the quantum theory.
The Copenhagen interpretation is the type of
hypothesis that (in Hanson's view) will most
plausibly resolve the current anomalies to Dirac's
relativistic quantum theory, just as it had enabled
Dirac to develop his quantum theory in 1928.
Hanson also maintains that the conceptual gestalten
constituting the semantics for currently accepted
theories not only supply some guidance for the
creation of new theories, but also offer what he
calls conceptual resistance, which must be overcome
for scientific discoveries.
The development of a new theory requires a
new gestalt
just as in the reinterpretation of the ambiguous
drawing, and similarly there is a resistance to such
a change.
In Patterns of Discovery Hanson illustrates this in the historical
episode in which Kepler developed the theory that
the orbit of Mars is elliptical.
In formulating this theory Kepler had to
reject the traditional belief held since Aristotle
that the orbits of the planets are circular, because
unlike sublunar motions the celestial motions are
perfect.
This is also the thesis in Hanson's most
significant historical analysis set forth in his Concept
of the Positron.
This work is original historical research in
which Hanson interviewed several physicists
including the three principals in the episode: Carl
Anderson, P.A.M. Dirac, and P.M.S. Blackett.
All three physicists discovered the positron,
but only Blackett recognized that the particle
discovered experimentally by Anderson was the same
one that was postulated theoretically by Dirac.
Dirac's 1928 paper offered a relativistic quantum
theory that was Lorentz-invariant, but it also
contained negative energy solutions that could not
be eliminated.
Originally he had hoped that these strange
solutions could be construed as protons, and then he
thought of them as vacancies which are positive
charge solutions with the mass of the electron.
This constituted the gradual development of
his prediction of the existence of positive
electrons before they were observed.
Anderson made photographs of electron tracks
in the cloud chamber, and he concluded that one of
them showed a positive electron, because the change
of the particle was positive while its mass was too
small to be that of a proton.
Dirac had published his theoretical paper on
the positron in 1931, a year before Anderson's
photograph.
In 1933 Blackett and Occhialini reported that
the Anderson particle and the Dirac particle are the
same thing, by using the new photographic technique
in which the particles took photographs of
themselves.
Hanson states that one reason Anderson did
not recognize any connection between his cloud
chamber experiments and Dirac's quantum theory, is
that such experiments rely on concepts that are
largely classical in nature such as track-leaving
particles.
But the greatest conceptual constraint, the
one that led many physicists to reject the idea of
the positive electron, was in the semantics of the
concept of the electron.
That semantics was such that an intimate
association between the electron and the proton, and
between the two basic units of electricity, negative
and positive, made the very idea of a particle other
than a proton or an electron very difficult to
conceive.
Just as positive/negative exhaust the
totality of electrical charge, so too the
proton/electron was thought to exhaust the totality
of charged particles, since the proton and the
electron came to be viewed not as carrying the
charge but as being the charge.
Hence there was a conceptual resistance to
the idea of a third charged particle built into the
structure of classical electrodynamics and the
elementary particle theory.
Hanson
on Perception, Observation and Theory
Hanson defends the Copenhagen interpretation,
and he criticizes the hidden-variable interpretation
and Bohm's agenda.
He maintains that in microphysics all the
limitations placed on our conceptions of what the
microphysical world is like and what we can observe,
are really limitations arising out of the linguistic
features of the formal languages available.
Such is particularly the case with
Heisenberg's uncertainty relations.
The uncertainty relations and Heisenberg's
thought experiment involving a gamma-ray microscope
are often said to state limits to the possibility of
observation within microphysics.
Hanson says that this is true in an
unsuspecting way: there never have been nor could
there ever be experiments or observations pertinent
to the establishment of the uncertainty relations,
because these relations are the conceptual or
logical consequence of the language of quantum
theory.
In the formalisms for modern quantum physics
there is a logicolinguistic obstacle to any attempt
to describe with precision the total state of an
elementary particle, and if there is a conceptual
limit to such a description, then there is ipso
facto a limit to such observation.
The conceptually impossible is
observationally impossible.
Hanson's thesis is that theory is integral to
observation or, as he also says, observation is
theory-laden.
This is also implied by Einstein's admonition
to Heisenberg that it is the theory that decides
what the physicist can observe.
Hanson's is the same philosophy of
observation that Einstein told Heisenberg in 1925,
and that Heisenberg used to develop the uncertainty
relations.
But Hanson was not led to develop his
philosophy of observation by reflection on
Heisenberg's autobiographical chronicles, in which
Heisenberg relates his discussion with Einstein and
the use that he made of it.
Hanson identified Heisenberg's views on
observation with those of Bohr, which Heisenberg
included in his explicit and systematic philosophy.
Nor was Hanson led to develop his philosophy
in response to Feyerabend's criticisms of Bohr's
dogmatic interpretation of quantum theory; Hanson's
philosophy of observation was developed many years
previously.
His philosophy of observation was drawn from
Wittgenstein's Investigations
and from the gestalt
psychology.
It is necessary, therefore, to consider
briefly Wittgenstein's ordinary-language philosophy
and Hanson's use of it in his philosophy of science.
Ludwig Wittgenstein (1889-1951) was a
somewhat reclusive individual who wrote a somewhat
unsystematic philosophy of language in a somewhat
obscure style, and who is thought to have
anticipated certain ascendant trends in
philosophical thinking.
In fact Wittgenstein seems twice in his
lifetime to have anticipated successfully an
ascendant trend in philosophical thought with his
two principal works: firstly his Tractatus
Logicus-Philosophicus (1922) and then later his Philosophical
Investigations (1953).
The thesis of the latter explicitly includes
a repudiation of the thesis of the former, yet each
work gathered its own retinue of sympathetic
interpreters and devout disciples.
Both the Tractatus
and its author attracted the attention of Schlick
and his Vienna Circle (with the noteworthy exception
of Carnap, who after his one and only meeting with
Wittgenstein was unforgettably unimpressed).
But in spite of Schlick's invitations to join
the Vienna Circle, Wittgenstein remained aloof from
them, just as he remained aloof from all other
sublunar states of affairs.
About thirty years later the
Investigations inspired philosophers who were
becoming disillusioned with the technical pedantics
of Logical Positivism, and its thesis occasioned the
formation of a new philosophy of language.
Conventionally historians of philosophy now
refer to the two opposing dogmas in these two books
as the ideal-language tradition and the
ordinary-language tradition respectively.
The ideal-language view set forth in the Tractatus
has a reformist flavor, which accords special status
to symbolic logic, such as may be found in Russell's
Principia
Mathematica.
The Tractatus
advanced an ideal (not metaphysical Idealist)
interpretation for symbolic logic, consisting of
what is called a picture-theory semantics.
This is one of many variations on the
naturalistic theory of the semantics of language,
and it is also the most naive.
This first book also advanced a
constructionalist view of language.
It described all sentences in the ideal
language as consisting of elementary sentences,
which in turn consist of semantically independent
names of simple objects.
All nonelementary sentences are constructable
from the elementary ones.
The former is said to be truth functional,
which means that the truth of the constructed
compound sentences depends completely on that of
their component elementary sentences.
As a result of this semantical atomism and
logical constructionalism, the understanding of any
sentence ultimately reduces to knowing its logical
structure and what its constituent names reference.
This is a variation on the mechanistic
philosophy of the semantics of language, and was
called logical atomism.
The principal argument in defense of the
ideal-language tradition is that ordinary language
is unsuitably vague and misleading for philosophy,
just as it is unsuitable for empirical sciences like
modern physics, which rely on mathematics.
The initial attractiveness of symbolic logic
to philosophers of science was the expectation that
it could serve philosophy as mathematics serves
physics.
This programme evolved into the Logical
Positivist reductionist programme of Carnap and
others such as Feigl and Hempel, in which the
controlling agenda was the logical reduction of
theories to a semantically significant observation
language, in order to demonstrate the meaningfulness
and semantics of the scientific theories.
But experience with the reformist efforts of
the ideal-language philosophers, notably the Logical
Positivists, led some younger philosophers to charge
that ideal languages are even more unsuitable than
ordinary language for philosophy, and that
philosophical analysis should be directed toward the
examination of colloquial language.
The outcome was a new folk philosophy that is
self-consciously naive.
Wittgenstein anticipated this reaction,
perhaps because it was also his own reaction to his
own Tractatus,
and he was led to develop his ordinary-language
philosophy.
Early statements of his new philosophy were
set down in a set of notebooks later published as The
Blue and Brown Books (1958), and the more mature
statement is the Investigations.
The latter work describes philosophy as a
kind of empirical linguistics, and its main themes
are (1) the variety of uses of language, (2) the
need for the philosopher to consider statements not
in isolation but in the context that occasions their
utterances, and (3) the definition of meaning in
relation to usage.
Wittgenstein maintained that the problems of
philosophy originate in philosophers'
misunderstanding of certain crucial terms such as
“know", "see", "free",
"true", "reason", and that the
resolution of these problems requires examination of
the uses of these words as they occur in
ordinary-language discourse.
The later Wittgenstein seems clearly to have
rejected the naturalistic theory of the semantics of
language.
He asks rhetorically in the Investigations,
if the formation of concepts can be explained by
facts of nature, then should the philosopher not be
interested not in grammar, but rather in that in
nature which is the basis of grammar.
He answers that the philosopher is not
interested in natural science or in natural history,
and he affirms an artifactual theory of the
semantics of language stating that a concept is
comparable to a style of painting.
But the artifactual theory that he accepts
seems to be a wholistic one, since he states in the
opening pages of The
Blue and Brown Books that understanding a
sentence means understanding a language.
Hanson was of the generation of philosophers
who took their professional education after the
Second World War, and he was also one of those who
looked to Wittgenstein's new philosophy to rise
above the inadequacies of the Logical Positivist
philosophy of science.
He was not an ordinary ordinary-language
philosopher; he was firstly a philosopher of
science, and if there was an ordinary language of
interest to him, it was the language ordinary to
contemporary physics including most notably
microphysics.
He was specifically drawn to Wittgenstein's
comments in the Investigations
about seeing, in order to re-approach the subject of
observation in physics, which modern quantum theory
had made so problematic.
Hanson's discussions about observation and
theory are set forth in Patterns
of Discovery, in "Observation and
Interpretation" in Philosophy of Science Today (1967), and in Perception and Discovery.
Hanson rejects the Positivist view that
seeing is merely a matter of predetermined
sensations, sense data, phenomena, or retinal
reactions in the eye, and that interpretation is
something added to the predetermined perception as a
secondary and discrete step in the perceptual
process.
Instead he says there is more to seeing than
meets the eye, and he follows Wittgenstein's view
that interpretation is an integral component of
seeing instead of something forced on it.
The significance of this point is that
perception is not predetermined and fixed by nature
but is variable, and he illustrates this variability
in perception by using both Wittgenstein's and
others' ambiguous drawings that admit to reversible
optical interpretations.
He explicitly invokes Gestalt psychology
(something that Wittgenstein did not do), to explain
the reversibility of interpretations of ambiguous
drawings as changes in the conceptual organization
of what is observed.
In this context Hanson references Duhem's
example in The
Aim and Structure of Physical Theory of the
layman visiting a physicist's laboratory.
The layman would have to learn physical
theory before he could observe what the trained
physicist observes.
Duhem had described this commonplace state of
affairs in terms of his Positivist semantics of
observation and theory.
But Hanson is a critic of Positivism, and he
does not maintain any such two-tiered semantical
thesis, as Duhem had.
Hanson maintains that the postulated
laboratory situation reveals that the elements of
the laboratory in the visitor's field of perception
are not organized as they are for the trained
physicist.
Physical theory provides the physicist with
patterns within which data appear intelligible; it
is what makes possible observation of phenomena as
being of a certain kind and as related to other
phenomena.
To illustrate his thesis that perception is
theory-laden, Hanson uses the example of the
second-century and the seventeenth-century
astronomers who both look at the dawn.
They both have the visual experience of the
rising sun, but they do not see the same thing,
because each believes different astronomical
theories: the former, Ptolemy, believes in the
geocentric theory, the latter, Galileo, in the
heliocentric theory.
Nevertheless, it can still be said that they
see the same thing, since the sun could be described
by both as a brilliant yellow disk.
Hanson calls this latter kind of description
phenomenal seeing, but he maintains contrary to the
Positivists that such phenomenal seeing is not the
ordinary way of seeing.
It is something that requires special effort,
because seeing is normally interpretative, and is
used when the observer is confronted with a new
seeing experience in which case what is seen cannot
be characterized by reference to his background
knowledge.
Observation in science aims to pass beyond
the phenomenal seeing occurring in the case of the
new experience, and to get the visual experience to
cohere against a background of accepted knowledge.
The differences between gestalten
are due to differences in previously acquired
background knowledge, knowledge that involves
language. Hanson is therefore led to follow
Wittgenstein's ordinary-language analysis, because
examination of commonly used locutions in colloquial
discourse reveals the relation between language and
the variability of interpretation in observation.
The locution "seeing as" reveals
that seeing is to see an object as a certain kind of
thing, which is brought out by the verbal context in
which the locution occurs.
The text in its context supplies the
interpretation.
But his thesis is still stronger than merely
stating that language reveals an interpreting
conceptual component; he invokes the locution
"seeing that" to exhibit a necessary role
for language in interpretation.
The idea of "seeing that" explains
the relation of "seeing as" and the
observer's background knowledge: to see something as
a certain kind of thing is to see that it behaves in
a certain known and expected manner.
The "seeing that" locution supplies
a statement of the background knowledge, which can
be true or false.
Seeing is therefore a theory-laden activity
in the sense that the seeing is interpreted by
reference to our background knowledge.
Without a linguistic component to seeing,
nothing we saw could be relevant to our knowledge.
Before the wheels of knowledge can turn
relative to a given visual experience, some
assertive or propositional aspect of the experience
must have been advanced.
Only statements can be true or false; visual
experiences must be cast into the form of a language
to be considered in terms of what we know to be
true, i.e. in terms of our theories.
Furthermore, Hanson's thesis is not only that
language is necessary for the interpretation that is
integral to perception, but also that the
logicogrammatical form of the language used for
description exercises a formative control over the
interpretative thinking that occurs in perceiving.
Just as seeing may be stated locutions which
are "that..." clauses, so too can facts
and theories.
For this reason Hanson says that Ptolemy
could not express in the second century what were
facts for Galileo fifteen centuries later.
Physical concepts are intimately connected
with the formalisms and notations in which
scientists express them, including the formalisms
used today in contemporary microphysics.
The dependence of physical concepts on the
mathematical formalisms is a very strategic
consideration in Hanson's rejection of attempts by
Bohm and Feyerabend to propose interpretations of
the uncertainty relations and the Schrödinger wave
function that are alternatives to the Copenhagen
interpretation of modern quantum theory.
For Hanson the Copenhagen interpretation is
precisely that interpretation which is supplied by
the formalism of the modern quantum theory, because
contrary to both the Positivists and to Bohr, it is
the formalism that supplies the intelligible
patterns and conceptual organization in perception
for the observations relevant to microphysics.
Interestingly in his Primer
of Quantum Mechanics (1992) Chester Martin
explicitly exhibits Dirac’s notational system as a
language, and references the linguistic philosophy
of Benjamin Lee Whorf.
|
|
