[Return to Dmitry Sepety’s
Personal Page]
In
Defence of Reasons
and
Other Criticisms of D.Miller’s Critical Rationalism
by Dmitry Sepety,
“The principles that form the basis of every
rational discussion, that is, of every discussion undertaken in the search for
truth… 1. The principle of fallibility… 2. The
principle of rational discussion: we want to try, as impersonally as possible,
to weigh up our reasons for and against a theory… 3. The principle of
approximation to the truth…”
(Popper 1992, 199)
One of the Myths about Popper’s
Philosophy
There are
several widespread myths about Popper’s philosophy of knowledge. Historically
the most influential is the positivist myth that Popper’s philosophy is a
variety of positivism. This myth seems gradually to come out of fashion to be
replaced with another one: that Popper’s philosophy is ‘eccentric views’ on the verge of irrationality or
beyond it.
My first
encounter with this myth was in the books of Stephen Law, an ingenious
populariser of philosophy (author of two bestsellers – “The Philosophy Stories”
and “The Philosophy Gym”), though not always competent in what he writes about.
S.Law attributes to Popper the view "that we never possess any grounds for
supposing that a scientific theory is true” (Law 2004, 126). Moreover, he
writes that Popper, in effect, accepts Hume’s conclusion that beliefs (1) that tomorrow
morning the sun will appear over the horizon and (2) that tomorrow morning a
million-mile-wide luminous panda will appear over the horizon are equally
reasonable (Law 2007, 175). I had e-mail discussion with mr. Law, and he has
remarked that this is "a very common understanding of Popper". I
don’t know if I have succeeded to change mr. Law’s views with my explanations
and arguments.
At that
time I thought that this myth, which directly contradicts many Popper’s statements,
is due to his incompetent and ill-disposed opponents. A few years later, when I
started my familiarizing with works of Popper’s followers, I was amazed to
discover that some of them interpret Popper’s philosophy of critical
rationalism in a way very much like that in S.Law’s quotations. The most
reputed of them is David Miller. (Another example is the article of Antoni
Diller “On Critical and Pancritical Rationalism”, which I criticize in the
article “On Alleged Flaws of Pancritical Rationalism”.)
Miller’s
high repute is based on a friendship and close collaboration with Popper, and
also on his incontestable critical contribution to the philosophy of critical rationalism
which consists in being one of the first logicians who showed Popper that his
definition of verisimilitude as a matter of more truths and less falsities in
the logical content of a theory is untenable.
Because of
Miller’s high repute, he is often taken as the first modern authority on
critical rationalism, and Popper’s own philosophy is understood in the light of
Miller’s explanations and developments.[1] In this article I argue that this
may be misleading in some important respects; that Popper’s philosophy was both
much more commonsense and more rich and complicated, and that some Miller’s
developments are hardly beneficent for critical rationalism.
I consider
as the most important point of difference the question about reasons – their
"job" and importance for rational thought. Miller expounds critical
rationalism as banishing reasons of all kinds, without informing that this was not Popper’s view, that Popper
really considered some kind of reasons – which he did call 'critical reasons' –
as integral for rationality. Besides, I think there are other important
differences between Popper’s and Miller’s critical rationalisms. They concern
with questions: what theories belong to empirical sciences? if they all have
the equal status? if there is any practical need to estimate reliability of
theories and probability of predictable events?
There is
yet another important question where I disagree with Miller: whether science needs
some metaphysical presuppositions (assumptions) about “immutability or order of
nature”, “that there is some regularity in the world”? (On this question,
Miller and Popper seem to agree, though Popper’s view is not so unequivocal.)
1. Critical Rationalism and Reasons
Miller on Reason and Reasons
In Chapter 3 of his Critical Rationalism: A Restatement and
Defence Miller wages war against reasons – of all kinds: sufficient, good,
positive, negative...
I am afraid that I don’t
quite well understand what Miller really means. He doesn’t quite deny that
there are reasons, but state that they “have no job” and have nothing to do
with rationality:
“Reasons exist, no doubt, at least in the
subjective sense, but not good ones... But this does not imply that rationality
is impossible, either in intellectual affairs or in practice. Reason has a job
to do in every sphere; reasons, poor things, have not.” (Miller 1994, 52)
Possibly this is to be
understood in the meaning: all there is to "reasons" is just subjective feelings which “do no job” in
the process of acquisition and development of knowledge, except possibly bad
job, – if it does any, it is quite irrelevant to rationality. Miller directly
states this about “good reasons”:
“The subjective feeling of being in possession
of good reasons may exist. But as far as rational thought is concerned,
evaluation in terms of good reasons is a pure epiphenomenon.” (Miller 1994, 66)
I still wonder how to
understand Miller’s banishment of reasons from epistemology. Does Miller means
that arguments and criticisms “have no job” and are irrelevant for rationality?
No, he doesn’t, at least with respect to criticisms (negative arguments). But
if so, I am puzzled about 'reasons'. May be, Miller uses that word in some specific
technical meaning which I fail to catch clearly enough. So, I must admit that I
don’t know what Miller means by 'reasons'. The pity is, Miller doesn’t explain.
In my understanding, 'reasons'
means whatever is relevant to judgement on truth/falsity, or truthlikeness, or plausibility
of position under examination. Arguments (criticisms) are just statements of
reasons. To argue for the position X, or to argue against X (to criticise Õ) is
to say: “I think X is (not) the case, for {reason}” or “I think "X"
is true (or false or truthlike or untruthlike or plausible or implausible), for
{reason}”. Reasons are nothing but contents of arguments (criticisms). As for predicate
"good" or even "sufficient" I think it can be well used in
the meaning: making the case for belief "X" or against "X" plausible
enough with respects to purposes at hand.[2]
Now let us look at
specific kinds of reasons Miller rejects.
Sufficient reasons. By sufficient reasons Miller means those which guarantee the truth of the
position under examination. I, as fallibilist, agree with Miller that such
reasons don’t exist.
Good reasons.
Here I am not sure what Miller means. Miller describes good reasons as
“sufficient or even partly sufficient favourable (or positive) reasons for
accepting a hypothesis rather than rejecting it, or for rejecting it rather
than accepting it, or for implementing a policy, or for not doing so” (Miller
1994, 52).
Now I am completely at
loss how to understand denying existence of such reasons, except if Miller uses
phrase "sufficient reason" in the strong meaning, as described above.
But if so, I don’t understand what "partly sufficient reasons" (later
usually referred to as "insufficient reasons") means. Or may be
Miller denies only favourable (or
positive) reasons? No, Miller includes also reasons “for rejecting it
rather than accepting”, – i.e. unfavourable (or negative) reasons.
Really, Miller admits
that his idea of what "good reason" means is rather vague. He writes:
“It would help if I had a better idea of what
good reasons are supposed to be good for.” (Miller 1994, 60)
As I sometimes use the
word-combination "good reason" or "weighty reason" (on
supposition that such things exist), as I sometimes judge some reasons as good
(or weighty), I can readily explain what I mean. And I guess that usually those
who talk about good reasons mean the same.
For me, good (weighty)
reasons are those which I judge good (weighty) enough for the purpose at hand, –
for judging, or classifying (however conjecturally and tentatively) the position
under examination as true (or truthlike, sound, plausible, commendable etc.) or
false (or untruthlike, unsound, implausible, censurable etc.). Or, as weighty
enough to influence considerably the balance, to make the case for/against the
proposition under consideration stronger/weaker.
Miller seems to mean by
good reasons something different, for he writes:
“Although there are such things as good
arguments, and it is these that the rationalist strives to provide, there are
no such things as good reasons…” (Miller 1994, 51-52)
As I already said, I
fail to understand what Miller means by divorcing arguments from reasons. For
me, good arguments and good reasons are inseparable: good reasons are meanings
(contents) of good arguments; good arguments are statements of good reasons.
Now about positive and negative reasons.
(1) Miller rejects
positive reasons on the ground that they, on his view, necessarily beg the
question under examination and entail infinite regress or vicious circle. Miller’s
argument goes as follows:
B is a positive reason for A
if and only if A is logically derivable (deducible) from B. But it is possible
only if A is a part of logical content of B. If so, proposing B
as reason for A is the same as proposing A as a reason for itself. It is
begging the question. And if we continue the chain to C as positive reason for B,
and D
as positive reason for C etc. we will get infinite regress
or logical circle, and each link of this chain is begging the same question.
So, there can be no positive reasons.
(2) Miller discusses the
question: if the same holds for negative reasons? Prima facia it seems that it holds. Miller describes the situation as
follows:
Suppose e
is logically derivable from h. In this case, not-h
is logically derivable from not-e. Can not-e be proposed as
reason for not-h? The fact is, that logical relation of not-e
to not-h
in this case is the same as relation of B to A in previous case. To
make it more obvious let me designate not-e as ê and not-h
as ĥ
and rewrite:
ê is negative reason with respect to h
if and only if ĥ is derivable from ê. But it is possible only if ĥ
is a part of logical content of ê. If so, proposing ê
as reason against h is the same as proposing ĥ as a reason for itself. It is
begging the question.
But this doesn’t discourage
Miller. He proposes “two responses open to critical rationalist here: one
specious and superficial (though not wrong), the other simple and satisfying”.
(2.1) Miller writes:
“Although not-e may beg the question of not-h,
it does not beg the question at issue, namely h.” (Miller 1994, 68)
I am very much amazed at
Miller’s qualifying this response as spurious though not wrong. It is stark wrong. There is no question of not-h
as distinct from question of h. The question is one: “If h?”
(which can be as well formulated as “If not-h?” or “What is the case: h
or not-h?”
– the meaning is the same). h and not-h are possible
answers, not question. And argument begging the question against h
(in favour of not-h) is as fallacious as argument begging the question in
favour of h.
Miller sees the
difference which excuses negative reasoning in the fact that continuing the
chain of criticisms doesn’t retain begging the initial question. Really, if ê
is criticism for h; ĥ belongs to logical content of ê. Now if some g
(different from h) is criticism of ê, then e belongs to logical
content of g. But neither h nor ĥ needn’t belong to
logical content of g. So, there is no that sort of regress as in the case of
positive reasoning, when each new knot in the chain of "reasons" begs
the initial question. Miller states that here we have not regress, but
progress.
I don’t see how there
can be progress. Really, moving this way we can change the question which is
begged with each turn, but each step of this movement is as fallacious as in
the case of positive reasoning. (Nor do I see why this chain of reasoning can
be stopped at some point while the chain of positive reasoning can’t; but this
is of little importance.)
Miller finishes discussion
of this “response open to critical rationalist” with admission that “it smacks
of casuistry” (Miller 1994, 70). I contend that it doesn’t smack, – it is
casuistry, brain-muddling. It is spurious, fallacious, wrong.
(2.2) The second
“response open to critical rationalist” which Miller appraises as “simple and
satisfying”:
“…critical rationalism need have no more truck
with negative reasons than it does with positive reasons. There are no negative
reasons either: nor do we need them for rational thought and action. We don’t
need reasons against a hypothesis in order to classify it as false (nor against
a course of conduct in order to classify it as foolish). All we need is a false
consequence of it…” (Miller 1994, 70)
There are quite a many critical things which I
have to say about this response, but I will postpone most of them. Now I
wonder: how this “simple and satisfying response” is supposed to solve the
problem of begging the question?
Suppose, that, as Miller
states, all we need to classify a hypothesis as false is a false consequence of
it. Now, if Miller’s previous argument about begging the question is correct, then
any judgement that some consequence of a hypothesis is false is begging the
question against that hypothesis!
On Miller’s above
exposition, if we judge (or "classify") as false the consequence e
of the hypotheses h, we beg the question against h (in favour of not-h).
Now do we really beg the question or no? If we do, then judging
("classifying") e as false would be begging the question
against h (not-h is part of logical content of not-e). If we don’t,
then proposing not-e as a reason
against h (in favour of not-h) is not begging the question
too.
Miller seems to
recognize that the charge of begging the question applies to arguments as well
as to reasons, for he writes:
“All valid arguments beg the
question.” (Miller 1994, 58)
Now the question is:
does this mean that all valid arguments are fallacious? Or there is nothing
fallacious about begging the question?
To recapitulate preceding
discussion.
If there are arguments
that are not begging the question, then there are reasons that are not begging
the questions. And if there are no reasons that are not begging the questions,
then there are no arguments that are not begging the question.
If begging the question
is fallacious, and all good arguments beg the question, then it is equally
ruinous for arguments and for reasons. If begging the question is not fallacious,
then reasons are to be allowed begging the question as well as arguments.
If Miller’s argument
about begging the question undermines reasons, it undermines arguments as well.
If it doesn’t undermine arguments, then it doesn’t undermine reasons.
Whatever is the case, arguments
and reasons are in the same basket. You can save either both or none.
As for my view, it is as
follows. There is the third way open for critical rationalist: to see that all
Miller’s exposition about begging the question is mistaken. And this mistake is
rooted exactly in Miller’s attitude towards reasons as “poor things” that have
nothing to do with rationality.
Distinction between begging and not begging the question, – in any sense
in which there is something wrong with begging the question, – is exactly the
matter of reasons, – not the matter of logical content.
Surely, you can define
"begging the question" as a matter of logical content. But if begging
the question is so defined, then begging the question is not fallacious.
On the other side, as far as I know, all sense of
qualifying some arguments as "begging the question" consists exactly
in blaming them as fallacious (logical fallacy) – in the sense of having no
force, providing no reason at all. All traditional usage of term "begging
the question" have sense only so far as there is something wrong with
arguments begging the question. As far as I can judge, usually there is indeed
something wrong with arguments charged with begging the question. But what it
is?
I think that it is not the fact that conclusion of
argument belongs to logical content of conjunction of its premises. There is
nothing wrong with this. For this is the case with all logically valid
arguments – whether negative (criticisms) or positive (favourable).
On my view, what is really wrong with arguments
usually charged with begging the question is that their premises are spurious,
sort of ad hoc – some of their
premises are forged specially for the case under discussion, and there are no
(good enough) independent reasons for their acceptance.
So, to retain its raison d’etre, "begging the
question" is to be defined not as a matter of logical content, but as a
matter of reasons.
If A
is derivable from B (is part of logical content of B), proposing B
as a reason for A is not begging the
question if your reasons for accepting B are independent of A.
B
can be a reason for accepting A (without begging the question), if
A
wasn’t a reason for accepting B.
I will illustrate it
with two examples, – negative and positive.
Suppose, we examine the hypothesis
H
“All swans are green”. Now the fact that white (or black) swans are observed
can be used as argument against H. It is even more natural to say
that observations of white (or black) swans is the (good) reason to reject H.
There is no begging the question, for our reason to think that there are white
(or black) swans is neither our acceptance of H, nor our rejection of H
(acceptance of not-H). Our reason to think that there are white (or black)
swans is our observations of white (or black) swans. But considering the
situation in terms of logical content we will see that not-H (“Not all swans are
green”) is a part of logical content of “White (black) swans exist”. This
clearly demonstrates that logical content is not the thing that demarcates logical
fallacy of begging the question from not committing that fallacy.
Situation is analogous
when we use some well corroborated or well-argued-for theory in criticisms of
other theories or singular statements.
Now the next question
is: can there be, likewise, positive
reasons which don’t beg the question? I think in this respect situation
with positive reasons is the same as with negative ones.
As example, we can take
any scientific prediction. If prediction p logically follows from universal
statement S in conjunction with conditions C (p is part of logical
content of S&C), and S is scientifically acknowledged law
of nature, than the fact that scientists have accepted S on reasons independent
of p
and the fact that we know about C independently of p
constitute a reason for our acceptance of p.
Or we can take classical
simple example. Mortality of all humans and humanity of Socrates can be taken
as a reason to think (judge or "classify" as true) that Socrates is
mortal. And there is no begging the question, if our (conjectural) knowledge
that all humans are mortal is accepted (classified as true) on reasons independent
of the case of Socrates – as it was amply tested by great multitude of other
cases and have passed all examinations (there was no known case of immortal
human).
My argument can be structured in two theses:
Thesis 1.
There is principal difference
between two situations:
(S1) when X
belongs to logical content of Y
and
(S2) when believing
(holding, accepting) X is our reason for believing (holding, accepting) Y
(X is one of premises of
argument which we use to justify our acceptance of Y).
Thesis 2.
If (S2) is the case then argument
Y->X is logical fallacy.
If (S1) is the case then argument
Y->X is not logical fallacy.
The simplest
illustration for this is: I deny that all swans are white because I had seen
black swans.
X="Not all swans are
white"; Y="Black swans exist".
X belongs to logical content of Y,
but
accepting X is not our reason for accepting Y.
My reason for believing
that black swans exist is not my preceding belief that not all swans are white,
but my seeing black swans. There is no logical fallacy of begging the question
in my reasoning:
As I had seen black
swans, so, black swans exist.
As black swans exist,
so, not all swans are white.
******
So, Miller’s “simple and
satisfying” “response open to critical rationalist” – that critical rationalim "have no
truck" with reasons, doesn’t need them for rational thought and action;
that they aren’t needed neither to classify hypothesis as false, nor to
classify a course of conduct as foolish); that "all we need is a false
consequence of it" – is quite useless with respect to the problem of
"begging the question". There is much more to criticize it besides.
As for courses of conduct, they have neither
false nor true consequences, they have expected and unexpected consequences,
good and bad (on person’s judgement) consequences.
As for hypotheses,
Miller’s “simple and satisfying” answer is quite unsatisfactory, for it does
nothing but shifts the problem from the question "how are we to judge
truth/falsity of a hypothesis?" to the question "how are we to judge
truth/falsity of its consequences?". If there is a theory X and we judge
some of its consequences as false, then this judgement is a reason to judge X as false. But to judge those consequences as
false we usually need some reasons (which we judge good enough for the
purpose): we don’t attribute to consequences 'true'/'false' values ad arbitrium, by coin-tossing. (I hope
Miller wouldn’t recommend this way of judgement as rational.)
Probably, Miller means
either logically demonstrable falsity of some consequences or their empirical
refutations. But these are far from sufficient "for rational thought and
action". Logical demonstrations of falsity are possible only within formal
disciplines like mathematics which tell nothing about the world (world-1 and
worlds-2, in Popper’s terms), or for self-contradictory hypotheses. Empirical
falsifications are possible only within the domain of empirical sciences, and
it is possible only as far as we consider observations (or consensus of
observers about what is observed) as a good enoug reason for (tentative)
acceptance of observational statements as true. But there is a great many
self-consistent theories and statements outside formal disciplines and mathematics.
What has "rational thought and action" to do with such theories and
statements? Old good positivism declared that all such theories and statements
are metaphysical nonsense; the irony was that positivism itself got into this
category by its own criterion. I am afraid that Miller’s views make a similar
trick with critical rationalism and falsificationism, pushing them outside the
realm of "rational thought and action". Really, falsificationism and
critical rationalism, as well as all their alternatives, are metaphysical (on
Popper’s criterion of demarcation), i.e. empirically unfalsifiable theories.
And there are self-consistent theories among their alternatives. So is the
choice between falsificationism and critical rationalism and their alternatives
beyond "rational thought and action", just arbitrary irrational
commitment?
Even within empirical
sciences (on some criterion of demarcation, whose adequacy is a matter of
rational discussion, but is not a matter of empirical falsification and self-consistency)
the question how to interpret observations and if they falsify a hypothesis is
often controversial and needs critical discussion with arguments (reasons)
which aren’t reducible to unproblematic observational statements and logical
consistency.
Besides, there can be
alternative self-consistent theories with exactly the same testable
consequences! Suppose that there is some accepted and well-corroborated
scientific theory X which has successfully passed all tests up to now. Then
someone proposes an alternative theory Y-as-if-X of the kind: some
superintelligent and superpowerful beings control our experiences and make them
to be exactly the same as they would be if X was true. (This example was
suggested to me by A.Musgrave’s articles about scientific realism and
surrealism. The idea has a long philosophical tradition: Descarte’s "evil
genius, as clever and deceitful as he is powerful",
To summarize, with Popper’s
quotation:
"… observation alone cannot always decide which
of two competing theories is better… In general, more than observation is
needed: also needed is a critical discussion of the merits of the two theories.
Such a discussion must consider whether they solve the problem which they are
supposed to solve: whether they explain what they are supposed to explain;
whether they don’t merely shift the problem, for example, by an untestable ad hoc assumption; whether they are
testable, and how well they are testable." (Popper
2005, 55)
******
Another interesting
example is mathematics. Unlike empirical sciences, it is certainly not the case
with mathematics that hypotheses are proposed and are tentatively accepted as
true until some refuting example is found. In mathematics, hypotheses are
accepted as true if, and only if, they are “positively” proved. If a hypothesis
is neither proved nor disproved as yet, it is neither accepted as true nor
rejected as false. Surely, these mathematicians’ concerns with proofs are not
to be disqualified as “having no job” “as far as rational thought is
concerned”!
Also, it may be said
that all mathematical theorems are, in the logical sense, contained in the
system of basic concepts and axioms. In this sense, all mathematical proofs are
begging the question! Also, in this sense, the discovery of new mathematical
truths and their proofs are “world-3” matters; as a matter of “world-2”, all
these truths and proofs are already there, contained in the system of basic
concepts and axioms. I am afraid that too literal understanding and acceptance
of Popper’s phrase about "epistemology without a knowing subject"
fails to account for importance and non-triviality of these mathematical
discoveries, as well as of discoveries of new (hitherto unknown by any person)
important logical consequences of accepted theories in empirical sciences.
Alan Musgrave on Logomania
“Logomania leads to infinite regress, as
sceptics long ago pointed out. All
inferential beliefs (beliefs obtained by inference from other beliefs) must
rest on non-inferential beliefs. But
logomania entails that non-inferential beliefs are unreasonable – from which it
follows that all beliefs are unreasonable. So, everybody who is not a total
irrationalist must reject logomania. Everybody who is not a total irrationalist
must think there are good reasons for believing P that do not involve inferring
P from other propositions we believe.
Everybody who is not a total irrationalist must think that some
non-inferential beliefs are reasonable beliefs.” (Musgrave 2004, 22)
I quite agree with this.
In other article I have made my point that all our reasoning is necessarily
saturated with 'pre-rational judgements', meaning nearly the same that Musgrave
calls 'non-inferential beliefs'. We can call them also 'extra-logical reasons'.
Necessity of such reasons is closely
relates with what I have written above about begging the question: though, as I
stressed, deductive arguments need not necessarily be logical fallacy of
begging the question, those which aren’t must necessarily begin with some
extra-logical reasons.
To illustrate this with
some examples.
Belief that 1) black
swan were observed (suggested by perceptions) and that 2) there were black swan
(that their perception wasn’t chain of hallucinations) are extra-logical
reasons which start logical chain deductively refuting the theory that all
swans are white.
As well, according with
Popper, reasons for tentative acceptance of scientific theory are:
1) acceptance of
methodological rule that we are to accept tentatively those theories that have
passed successfully severe examinations up to present moment
and
2) judgment of state of
examination: that the theory have passed successfully severe enough
examinations up to present moment.
Both are extra-logical.
And, in case of
acceptance of prediction from scientific theory extra-logical reasons are those
which make us to accept that theory and report of relevant initial conditions.
Popper on 'Positive' and 'Critical' Reasons
Some quotations:
“Our experiences
… may even be described as inconclusive
reasons.” (Popper 1974, 1114)
“…we can
never give positive reasons which
justify the belief that a theory is true.” (Popper 2008, p.310)
“…² can
examine my guess critically, and if it withstands severe criticism, then this
fact may be taken as a good critical
reason in favour of it.” (Popper 2008, p.317)
“…we may
still give preference to t2, even after its falsification, because we have reason to think that it agrees
better with the facts than did t1” (Popper 2008, p.318)
“The principle of rational
discussion: we want to try, as impersonally as possible, to weigh up our reasons for and against a theory…”
(Popper 1992, 199)
These quotations leave
little doubts that Popper acknowledged reasons and considered them as inalienable
from rationality.
The question of Popper’s
views about positive reasons is trickier.
Popper’s formulation of
"the principle of rational discussion" clearly suggests that there
are reasons
for, as well as reasons against a theory. Now if 'positive reasons'
means reasons for, favourable reasons, then Popper surely
did admit existence of positive reasons.
When Popper proposed his
theories, he didn’t just postulate, he argued in their favour, i.e. provided
favourable reasons. And his falsificationist theory of empirical science specifies
relevant favourable reasons. It recommends 1) to admit into science those
hypotheses which are refutable and 2) to accept (tentatively) as true those of
them which have successfully passed diverse and severe examinations. So, it defines
reasons for 1) recognizing a theory as scientific and 2) (tentative) acceptance
of a scientific theory as true. Again, if 'positive reason' means reason for accepting (classifying as true),
then Popper did consider as positive reason for (acceptance of) a scientific
theory its passing diverse and severe examinations. In a sense, it can be even
said that Popper did consider it as a sufficient
reason – not in the meaning of demonstrating truth or high probability, but
in the meaning that it is sufficient for
tentative acception of hypothesis (tentative classifying it as true).
If Popper denied
existence of "positive reasons" (which he did), then his meaning was
different. There is one kind of would-be reasons which can be called
"positive reasons" and whose existence Popper unequivocally and
argumentatively denied – logically valid demonstrations of truth or high
probability, “using this latter term in a sense laid down by the calculus of
probability” (Popper 2008, 321). (And Popper emphasized that probability, in
this meaning, is to be distinguished from degree
of corroboration and verisimilitude,
truthlikeness, i.e. degree of nearing
the truth.)
To achieve clearer view
of the issue, let us consult Popper’s book Realism and the Aim of
Science, chapter 1,
section I. Here Popper explains that he denies possibility “to justify our theories or beliefs rationally: that is to say, by giving
reasons – 'positive reasons' (as I shall call them)…; reasons, that is, for
holding them to be true, or to be at least 'probable' (in the sense of
probability calculus).”
In this sense, “we cannot give any positive
justification or any positive reason for our theories and our beliefs. That is
to say, we cannot give any positive reason for holding our theories to be true.”
(Popper 2005, p.19)
But:
“We can often give reasons for regarding one
theory as preferable to another. They consist in pointing out that, and how,
one theory has hitherto withstood criticism better than another. I will call
such reasons critical reasons…” (Popper
2005, p.20)
Popper makes similar
statements about justification:
“Critical reasons don’t justify a theory…” (Popper
2005, p.20)
But:
“Giving reasons for one’s preference can of
course be called a justification (in
ordinary language). But it is not justification in the sense criticized here.
Our preferences are 'justified' only relative to the present state of
discussion …” (Popper 2005, p.20)
But isn’t this replacement
of the search for truth with some other (instrumental) preferability? Popper
explains that it isn’t:
“My position is this. I assert that
the search for truth – or for a true theory which can solve our problem – is
all-important: all rational criticism is
criticism of the claim of a theory to be true, and to be able to solve the
problems which it was designed to solve. Thus I don’t replace the question whether a theory is true by the question
whether it is better than another. …to describe a theory as better that
another, or superior, or what not, is, I hold, to indicate that it appears to
come nearer to the truth. Truth – absolute truth – remains our aim; and it
remains the implicit standard of our criticism… Thus we are always searching
for a true theory (a true and
relevant theory), even though we can never give reasons (positive reasons) to
show that we have actually found the true theory we have been searching for. At
the same time we can have good reasons – that is, good critical reasons – for thinking that we have learned something
important: that we have progressed toward truth. For first, we may have learnt
that a particular theory is not true according to the present state of the
critical discussion; and secondly, we may have found some tentative reasons to
believe (yes, even to believe) that a new theory comes nearer to the truth than
its predecessors.” (Popper 2005, p.24-25)
My interpretation of
these Popper’s explanations is as follows. What he means – and denies on
weighty arguments – is possibility of 'justification' or 'positive reasons' in
the meaning: logically valid demonstration of truth or high probability. Let us call such reasons 'conclusive
reasons' and 'probabilistic reasons'
accordingly (they both can be united under title 'absolute reasons'; they are absolute
in the sense that they must be not
relative to some assumptions which are themselves 'unjustified' in this
sense). But non-existence of conclusive and probabilistic reasons doesn’t mean that
there are no 'positive
reasons' in the meaning favourable reasons (as opposite to 'negative reasons').
Also, as far as universal
scientific theories are concerned, Popper dis mean that we have no reasons to
believe that a theory, even the best one, is true sensu stricto (≡absolutely true), i.e. have no false content. For example,
On these points I agree
with Popper: there are no 'justification' or 'positive reasons' in his meaning.
But I disagree with
Miller: there are positive reasons in the meaning of favourable reasons (as opposite to 'negative reason'), reasons to
(tentatively) classify theory as true or to believe that it is well nearing the
truth (is highly objectively trithlike, of high verisimilitude). And (as Popper
admits) there is 'justification'
in ordinary-language meaning
congenial with favourable tentative reasons.
Ambiguities of meanings play
here a bad trick: it starts with confusion of 'positive' in the sense 'real',
'exact', 'indubitable', 'certain', 'for sure' with 'positive' in the sense
'good', 'favourable', and proceeds to confusion of 'critical' as 'relative-to-the-present-state-of-critical-discussion'
with 'critical (negative)' as 'unfavourable', 'against', 'refuting'. (To be
true, Popper himself provokes this misunderstanding, for he sometimes used
phrases 'critical
reason' and 'negative reason' as synonimous.)
To emphasize the
difference.
We are to distinguish
'positive reasons' in the meaning of
logically valid demonstration of truth or high probability (conclusive or probabilistic reasons)
from 'positive reasons' in the
meaning of arguments for (favorable reasons);
'critical reasons' in the meaning of
tentative, conjectural, fallible, relative-to-the-present-state-of-critical-discussion reasons (which may be favourable as well as unfavourable
reasons)
from 'critical (negative) reasons'
in the meaning of arguments against (unfavourable reasons).
(What is arguably true
about 'X' in one meaning needn’t be true about 'X' in another meaning. All dogs
are mammals; but hot-dogs aren’t.)
It is also to be
remarked that 'negative
reasons' as logically valid demonstration of falsity or high improbability – conclusive or probabilistic
negative reasons – are, in general (except in cases of self-contradictory
statements), just as impossible as conclusive or probabilistic positive reasons. (One proviso may be needed: as Popper
argued, in a sense, all theories of high content are highly improbable; the
higher is content the lower is probability; 'probability' of empirical scientific
theories tends to zero.)
Critical Rationalism ≠ Falsificationism
I conjecture that the confusion with different meanings of
'positive reasons' and 'negative reasons' stems from a failure to distinguish
clearly two Popper’s theories: 1) critical
rationalism as general non-justificationist conception of rationality which
applies to all sorts of statements, theories, beliefs, attitudes and 2) falsificationism as more specific theory
which applies to empirical sciences.
Falsificationism is proposed as solution for the problem of induction – validity of
inference from singular observational statements to universal statements
(scientific theories). Critical
rationalism (CR; with Bartley’s corrections – comprehensively critical
rationalism, CCR) is proposed as solution for a problem which Bartley calls
'the problem of limits of rationality' (Popper didn’t use that name, but he had
first introduced term 'critical rationalism' in The Open Society and Its Enemies while discussing nearly the same
problem, and developed his conception of critical rationalism in a lecture On the Sources of Knowledge and of Ignorance
while discussing an allied problem – the problem of reliable sources of
knowledge) – it is the problem of infinite justificationist regress arising
from identification of rationality with the demand to accept only rationally justified
statements.
Falsificationism is about inferences "in inductive direction", from singular
to universal statements. (C)CR is about inferences in whatever direction.
Falsificationism says that there is no valid inference from truth/falsity of singular
(basic) statements to truth or probability of universal statements (theories),
but there is valid inference from truth/falsity of singular (basic) statements
to falsity of universal statements (theories). C(CR) says that no position can be justified in compliance with justificationist
demand to accept those and only those
positions which are justified; but every position can be held open to
criticism; and proposes to understand 'rationality' as openness to criticism.
(C)CR is
based on demonstration that justificationist demand is unfeasible because it
leads either to infinite regress or to vicious circle. Falsificationism is based on simple logical truth:
if 1) X contradicts Y
and 2) X is true
then it logically follows that Y is false;
but
if 1) X agrees with Y or logically follows from
Y
and 2) X is true
then it does not logically follow that Y is
true.
(C)CR equally
applies to any position (whether it is singular observational statement or
universal empirical statement-theory or moral attitude or methodological rule
or religious doctrine). Falsificationism
applies only to empirical statements, preferentially universal (empirical
scientific theories).
(C)CR argues
that no position can be justified, conclusively
or probabilistically. In this sense,
justification is impossible. But the same applies to refutation: it is
impossible in the same sense and for the same reasons as justification. Surely,
refutation is possible on some assumptions (as in empirical sciences, refutation-falsification
of a theory on assumption of some basic observational statements). But those
assumtions themselves can’t be justified, whether conclusively or
probabilistically. Suppose we call such refutation tentative refutation, refutation on tentative assumptions. We have
for them also another name – criticisms.
All criticisms are tentative refutations; no criticism is possible without some
'unjustified' assumptions. But tentative justification,
justification on tentative assumptions is just as well possible. To
parallel term 'criticism', we can call it 'commendation'. (We are not to
confuse this with induction: tentative justification-commendation, as well as tentative
refutation-criticism can be deductive. For example, prediction P can be tentatively justified-commended
on assumptions of some theory T and initial conditions C,
if P
is deducable from conjunction of T and C; acceptance of a theory
can be deductively tentatively justified-commended on assumptions of some
methodological rule and judgement of its fulfillment.)
With respect to
'reasons' and 'justification' CR (CCR) and falsificationism differ in
absolute/relative character, comprehensive/limited incidence and symmetry/asymmetry
between positive and negative (in the sense: favourable and unfavourable)
arguments or evidences.
Falsificationism deals with the problem of induction, verification of universal
statements (theories) by singular (observational) statements. In terms of
'justification' it can be described as problem of relative (on assumption
of some basic singular statements) inductive
justification
of universal statements (theories). Falsificationist arguments
show that no such justification-verification (whether conclusive or
probabilistic) is possible; inductive inferences are invalid; while valid
refutation-falsification is possible and indispensable for development of
scientific empirical knowledge. So, situation with 'verification' and
'falsification' of empirical scientific theories relative to observational
(basic) statements is asymmetrical.
CR (CCR) is not concerned with induction. It deals
with the problem of absolute (not relative to some assumptions) justification of any
position, and its arguments show that such justification – whether
conclusive or probabilistic – is impossible. (Besides, its arguments show that it is not ruinous for rationality; rationality
can do well without justification.) But absolute refutation – whether
conclusive or probabilistic – is just as well impossible, for the same reasons.
On the other side, there is nothing in CR (CCR) arguments to deny that we can
often provide tentative reasons good enough for tentative acceptance or
believing some positions (and so, tentatively justify or commend them), as
wells as tentative reasons good enough for tentative rejection (non-believing)
of some positions (and so, tentatively refute or criticize them). So, situation
with refutation and justification, criticism and commendation is symmetrical.
2. Further Criticisms
of D.Miller’s Critical Rationalism
Corroboration, Verisimilitude and Scientific
Status
“… we are interested in theories
with a high degree of corroboration.” (Popper 2008, 77) “So long as theory
withstands detailed and severe tests and is not superseded by another theory in
the course of scientific progress, we may say that it has proved its mettle or
that it is corroborated by past experience.” (Popper 2002, 10) “I introduced the
terms 'corroboration' and 'degree of corroboration' … to describe the degree to
which a hypothesis has stood up to severe tests, and thus "proved iits
mettle".” (Popper 2002, 248-249) “… the degree of corroboration of a
hypothesis depends mainly upon the severity of its tests...” (Popper 2002, 269)
“A theory which is not in fact refuted by testing those new and bold and
improbable predictions to which it gives rise can be said corroborated by these severe tests.” (Popper 2008, 298)
“Confirming
evidence should not count except when it is the result of a genuine test of the
theory; and this means that it can be presented as a serious but unsuccessful
attempt to falsify the theory. (I now speak in such cases of 'corroborating
evidence'.)” (Popper 2008, 48)
In Popper’s theory of
development of scientific knowledge passing diverse and severe examinations
serves as a good reason for accepting scientific theory (classifying it as
true). Popper’s statements such as that “…we can never give positive reasons
which justify the belief that a theory is true” (Popper 2008, p.310) mean by 'positive
reasons' logically valid demonstration of truth or high probability.
Popper’s statements in
“The Logic of Scientific Discovery” and other works make it clear enough that
he did consider passing diverse and severe examinations (the more diverse and severe
the better) as quality mark for scientific theory. He designated it by terms
'corroboration' (to distinguish it from 'verification' and 'confirmation' in
positivistic sense) and 'degree of corroboration'.
Degree of corroboration
(unlike 'probability') is not absolute value, but comparative characteristic:
we can’t assign a numeric value for a degree of corroboration, but in many
cases we have objective reasons (results of experimental observations) to
estimate a degree of corroboration of one theory as higher than a degree of
corroboration of another theory. If theory A has passed all those tests which
theory B has passed plus some of those tests which B has failed to pass, then A
gets a higher degree of corroboration than B. Also, degree of corroboration for
the same theory changes with new tests. A degree of corroboration of A becomes
higher with each new test on its predictions (especially those which doesn’t
follow from alternative theories) which it has successfully passed.
Really, if degree of corroboration wasn’t a
quality mark for scientific theory, if it was not making a difference for its scientific status (relative to present
state of discussions and examinations), there would be no sense in Popper’s
statements such as:
“Instead of discussing the
probability of a hypothesis we should try to assess what tests, what trials, it
has withstood; that is, we should try to assess how far it has been able to
prove its fitness to survive by standing up to tests.” (Popper 2002, 248)
What is also important
about degrees of corroboration is that they can be attributed not only to those
theories which weren’t falsified, have passed successfully all examinations up
to present moment, but also to falsified hypotheses with considerable true content.
This links Popper’s
concept of 'degree of corroboration' with his later concept of objective
truthlikeness, verisimilitude (truth-approximation).[3] Popper remarks on this connection
in “Conjectures and Refutations”:
“…I don’t suggest that the explicit
introduction of the idea of verisimilitude will lead to any changes in the
theory of method. On the contrary, I think that my theory of testability or
corroboration by empirical tests is the proper methodological counterpart to
this new metalogical idea.” (Popper 2008, 318)
Concept of
verisimilitude (or truth-approximation) is introduced to range, with respect to
scientific value, those scientific theories which are, strictly speaking, false
(and falsified), but have much true content (are true or well approximate truth
in wide domains). The best example is Newton’s system of physics which was
falsified by some experiments and superseded by Einstein’s system of physics,
but have retained very much of its scientific status and importance, because it
is much simpler than Einstein’s (both for understanding and use) and is very good
approximation to truth whenever velocities are much less than velocity of light.
If so, then degree of
corroboration can serve as indicator of verisimilitude (relative to present
state of discussions and examinations).
Besides, what is
important about Popper’s theory, is that he acknowledges scientific status for
all empirical (falsifiable) hypotheses, false (falsified) as well as true (or
tentatively held as true).
In the light of all
this, the picture of empirical science, according with Popper, is rich and
diversified. Empirical science incorporates all falsifiable theories. These
theories differ in their scientific status, which may be higher or lower. There
are theories which are tentatively held as true, for they weren’t falsified as
yet. And there are falsified theories. Inside both these groups there are
differences of higher/lower scientific status, depending on universality, degree
of corroboration (severity and diversity of tests which a theory have
successfully passed up to present moment),
estimation of success in solving problems for which a treory was
designed or, in the case of falsified theories, approximation to those which
are held as true…
Miller’s picture of empirical
science, as outlined in Chapter 1 of “Critical Rationalism: A Restatement and
Defense”, is much simpler and poorer. All there is is entrance (admittance) to
science and expulsion out of it. Entrance is free for all falsifiable hypotheses:
“… a hypothesis need submit to tests only after
it has been admitted to science. If it fails any of the tests to which it is
put, then it is expelled, removed from science; if it passes them all, then
nothing happens – that is to say, it is retained.” (Miller 1994, 7)
This picture admits no
gradations of statuses within science, such as between new, untested or poorly
tested theories (low degree of corroboration) and theories which have already
passed successfully diverse and severe testings (high degree of corroboration).
It has no place for falsified theories: falsified theories are “expelled,
removed from science”, – unlike Popper’s view that they belong to science with
status of false theories, and can be of different scientific value, depending
on their verisimilitude. If we accept Miller’s view, then even such theories as
Newton’s physics no longer belong to science (since it was falsified and
superseded by Einstein’s physics)!
Besides, there is a
logical problem, as pointed out by A.Musgrave. For Miller admitting a hypothesis into science and classifying it as true is the same thing. But what if two
alternative, inconsisted hypotheses are proposed and not yet tested? According with Miller they are both admitted into science and
classified as true, though they contradict one another and can’t be both true.
(Musgrave 2007, 194)
Reliability and Probability
Yet one point where I
consider Miller’s arguments and explanations unsatisfactory, concerns the
questions of practical reliability of theories and practical need of estimation
of probabilities.
Popper’s views
In Chapter 1 Miller quotes
Popper who had written in “Objective Knowledge” that “from a rational point of
view we should not "rely" on any theory, for no theory has been shown
to be true, or can be shown to be true”, but “we should prefer as basis for action the best-tested theory”. (Popper 1972,
21-22)
This quotation, on my
view, is rather misleading, for there is important continuation which Miller leaves
out:
“In other words, there is no "absolute
reliance"; but since we have to
choose, it will be "rational" to choose the best-tested theory… Of
course, in choosing the best-tested theory as a basis for action, we
"rely" on it, in some other sense of word. It may therefore even be
described as the most "reliable"
theory available, in some sense of this term. Yet this does not say that it is
"reliable". It is not "reliable" at least in the sense that
we shall always do well, even in practical action, to foresee the possibility
that something may go wrong with our expectations.” (Popper 1972, 22)
It makes clear that Popper
didn’t deny that we can (and often need, for practical matters) to estimate "reliability"
of theories in the sense of relative
reliability: which theory is most reliable among known alternatives, as far
as we can judge in the light of examinations and rational discussions up to
present moment? which theory (among known alternatives) we are better to rely
on in practical decisions? In this meaning "to rely on the theory X"
is the same as "prefer X
as basis for action".
What Popper did deny, is
"absolute reliance".
Also, in fragment quoted
by Miller Popper stated that “'rationality' of choosing the best-tested theory
as a basis for action” is not to be understood “in the sense that it is based
upon good reasons for expecting that
it will in practice be a successful choice: there
can be no good reasons in this sense, and this is precisely Hume’s result.”
(Popper 1972, 22)
What did Popper mean by
this statement? Possibly next fragment will be explanatory:
“…even if our physical theories should be true,
it is perfectly possible that the world as we know it, with all its
pragmatically relevant regularities, may completely disintegrate in the next
second. This should be obvious to anybody today; but I said so before Hirosima:
there are infinitely many possibilities of local, partial, or total disaster.”
(Popper 1972, 22)
I understand Popper’s
meaning as follows. Though acting on the best-tested theories is reasonably the
best we can do, there is no way to estimate absolute probability that we will
succeed in any concrete case (if it is higher than 50%?). Besides the fact that
we have no way to estimate absolute reliability of our theories, there are
infinitely many unforeseeable circumstances which could influence outcome of
the action. So, the best we can do is to
do the best we can, and hope (not expect) that we will succeed. Expectation
is of no profit.
In Chapter 2, before expounding his own views,
Miller discusses how Popper would answer “justificationist question”:
Why should we act on the
best-tested theories if, as Popper states, they can’t be shown to be true and
we have no good reasons “for expecting that it will in practice be a successful
choice”?
Miller remarks that “Popper
rightly offers no direct answer” and continues:
“The most that can be said in response to it is
that there are no reasons for supposing that an unrefuted theory is not true. The
hypothesis that has most successfully survived the critical debate is, we
conjecture, our best source of true information about the world, and if we wish
to act appropriately, there is little sense in ignoring this information; or,
worse still, acting as if it were not true.” (Miller 1994, 39)
Miller calls this
“Popper’s answer”, but give no Popper’s quotation which would support his
suggestion that this was Popper’s answer.
I think it is not “the
most that can be said in response” – and not the best Popper could say.
First, obvious Popper’s
answer is: as far as we can reasonably judge, it is the best choice we have.
And this is what Popper really writes:
“This will be "rational" in the most
obvious sense of the word known to me: the best-tested theory is the one which,
in the light of our critical discussion,
appears to be the best so far…” (Popper 1972, 22)
Also, on my view, the
answer that we should act “as if the best tested of the surviving hypotheses is
true” because “there are no reasons for supposing that an unrefuted theory is
not true” is too weak. Really, the next sentence adds another reason: “The
hypothesis that has most successfully survived the critical debate is, we
conjecture, our best source of true information about the world”. But this reason
makes the first one (that “there are no reasons for supposing that an unrefuted
theory is not true”) superfluous. Really, even if a theory is, sensu stricto, false, it can
nevertheless serve as “our best source of true information about the world”, –
as, for example,
Now let us remark that in
terms of Popper’s theory of knowledge to say that hypothesis is “our best source of true information” is
the same as to say that it is the best approximation to truth (the most
verisimilar, truthlike hypothesis) we have”.
So, instead of answer
that we should act “as if the best tested of the surviving hypotheses is true”
because “there are no reasons for supposing that an unrefuted theory is not
true”, we get rather different answer – in terms of truth-approximation (verisimilitude) and its conjectured
positive correlation with a degree of
corroboration (i.e. severity of examinations which hypothesis have
successfully passed):
We better act on the
best-tested theories because
1) we conjecture that
there is (rather good) positive correlation between a degree of corroboration
and truth/verisimilitude;
2) the more theory nears
the truth, the better, by and large, are its predictions (the more it is
"reliable");
3) we have no better guide
for our actions (nothing better to rely on).
As for probabilities,
Popper strongly argued that there is no useful estimation of probabilities of theories (universal
statements); in a sense, the more "improbable" the theory is the
better (the higher is its content). But Popper never denied practical need to
estimate probability (or likeliness) of
singular events. (Though, this estimation can’t be absolute, it is relative to our conjectural knowledge.) Miller
goes so far as to deny even this.
Miller’s views
Miller’s own views about
practical reliability, as described in Chapter 2, are radically negative. He
argues that estimation of reliability (without any reservations about meanings
of the word, distinction of relative and absolute reliability) is of no
practical use at all.
Miller starts with
“important point … about how scientific theories are used in technology”, which
he proposes to dub "the slavery theory of applied science". He
explain that scientific theories can’t tell us what to do, how to achieve our
purposes. Scientific theories aren’t used “as a guide to action, or as a source
of positive advice” (Miller 1994, 39). For any practical purpose
“…no amount of science will tell us how to
tackle the job – there may be countless ways that will succeed. All that
science can do is to tell us what lines of approach we do well to avoid. That
is, we do not rely on, or even apply, scientific hypotheses when we act; we
exploit them.” (Miller 1994, 39)
There are two points in
this description I disagree with.
First is its conclusion
(the last sentence). It doesn’t follow from "the slavery theory of applied
science". Indeed, science can’t tell us what we are to do to achieve our
purposes, while telling us what options won’t we successful. But science is
used to get predictions: if proposed technological decision will give results
we purpose at, or no (and, sometimes, how great are risks). (Surely, these
predictions are made with proviso: if no unforeseeable events will unmake them.)
And we do rely on those scientific predictions,
– not because science insures success or high probabilities (which it does not),
but because there is nothing better to rely on.
Second, though I agree
that “no amount of science will tell us how to tackle the job” (1), I disagree
with the statement that “All that science can do is to tell us what lines of
approach we do well to avoid” (2). Again, (2) doesn’t follow from (1). Scientific
theories, far from being purely critical instruments for technology, suggest many possibilities about which no one would think without them, and they
suggest directions in which to search
for technological realizations of these possibilities. One of the most salient examples
is the atomic theory. Without development in XX ct. of the atomic theory which
suggests the possibility of nuclear and thermonuclear reactions, no one would
think of technologies to produce and use these reactions.[4]
Now about probability
(likeliness) of singular events. Miller discusses situations when we can’t base
our decisions on scientific predictions:
“Should I be inoculated against typhoid before
my next trip to the tropics? Science … is of little use here; what will matter
is the cleanliness of the food that I eat and of water that I drink, and there
are all kinds of reasons why no scientific prediction of these factors can be
made... It is usually thought that in circumstances such as these … there is no
option but to resort to judgments of subjective probability, or to
considerations of utility, or … both. The unpleasantness of typhoid far
outweighs the unpleasantness of its vaccine; and so if there is more than faint
probability that I shall be exposed to poor drinking water I should be
inoculated. This sounds like good common sense, but I nonetheless demur from
explanation. …I cannot see how recourse to probabilities can help. I want my
action to be appropriate to what is going to happen…, not to what is likely to
happen.” (Miller 1994, 43)
Later,
Miller adduces another
example:
“Which
ticket would you prefer to draw in a sweepstake: the one bearing the
favourite’s name, or the one bearing the winner’s? … what matters is the
conjectural classification of the hypothesis as true…, not its assessment as
probable…” (Miller 1994, 66)
As far as I see it,
there is simple and strong refutation of these Miller’s reasonings.
Especially in the case
of inoculations it is obviously wrong that subjective estimation of probability
(likeliness) is of no avail and can be replaced with "the conjectural
classification of the hypothesis as true". Such replacement would often
result in different decision and would be highly unreasonable. Let us compare,
how we really make such a decision, and how (and with what result) we would
make it if we replaced subjective evaluation of likeliness with conjectural
classification as true/false.
Really (estimation of likeliness). The question is: if, without inoculation, I
will catch typhoid? Suppose, I subjectively estimate (conjecture) that most
likely I won’t, but still there is considerable risk that I will. Suppose also
that, as far as I know, inoculation greatly lessens this risk and the only
inconvenience of it is a slight fleeting pain. Surely, it would be reasonable
for me to take inoculation.
If we replace estimation of likeliness with conjectural classification
as true/false. Conjecture
that without inoculation I most likely won’t catch typhoid is to be replaced
with conjecture that without inoculation I won’t catch typhoid. If I make
decision on this conjecture (without estimations of likeliness-risk), then I
won’t take inoculation. Really, whatever slight are inconveniences of
inoculation, I needn’t undergo them, if I won’t catch typhoid without
inoculation.
I think it is obvious that
the first way of making decision would be reasonable, while the second would be
unreasonable.
It is no so obvious in
the case of sweepstake, but it is well arguable that sometimes decisions based
on informed guesses about probabilities may be more reasonable than decisions
based on informed guesses about winners. For example, it could be reasonable
not to put all eggs in one basket, but to make several stakes on two or more
favourites. Or it may be reasonable even to stake on the option which, on your estimation,
has not the highest but still high enough chance to win, if stakes of other
gamblers are high against it. (For example, you estimate that chances for A are
55% while chances for B are 45%. There are many other gamblers which almost all
stake on A. Let us suppose stakes are
Besides, there is yet another
interesting objection. If we accept indeterministic view of the world and,
especially, of human activity (as Popper did), then it would be mistaken to
classify as true/false statements which say about some future events (especially those where humans are involved, as in
examples with inoculation and sweepstake) that they will or won’t happen. Such
statements are neither true nor false; they have no truth/falsity value until
event happens.
Does Science Need Metaphysical
Presupposition about the Existence of Order in Nature?
Miller’s answer seems to
be unequivocally “No”.
“…inductivism needs to assume that
there is some regularity in the world, whilst falsificationism requires only
that there is some regularity in the world – but it need make no assumption to
this effect.” (Miller 1994, 25)
“…science needs contain no
metaphysical assumption concerning the immutability or order of nature. It need
contain no assumption not explicitly available for testing… Scientific
hypotheses propose order for the world; they don’t presuppose it.” (Miller
1994, 27)
And Miller refers here
to Popper who “argued ... that the principle in question is a
metaphysical one, best replaced by the "methodological rule [that] …
natural laws … are to be invariant with respect to space and time"…”
(Miller, 25)
Popper did actually
argue so. But we need to ask a further question: why such a methodological rule
is needed? Surely, we don’t just accept it “out of the blue”! Doesn’t this methodological rule incorporate metaphysical
presupposition about the existence of some
discoverable genuine order in nature? And doesn’t Popper’s criterion of demarcation between
empirical science and metaphysics make the idea that there are genuine
regularities in the world (laws of nature) constitutive
of science as distinct from metaphysics? We always have two kinds of alternative
possibilities of explanation of some observable events – the one based on some
hypothesis about genuine regularities (laws of nature) and the second admitting
no such regularities (all that happens is just sheer chance). But the second
alternative is excluded from science by Popper’s criterion of demarcation, for
no empirically testable predictions follow from it.
In fact, Popper did
admit science needs metaphysical belief,
or even psychological presupposition,
that there are such genuine regularities in the world. So, in “The Logic of
Scientific Discovery”:
“And our
guesses are guided by the unscientific, the metaphysical (though biologically
explicable) faith in laws, in regularities which we can uncover – discover.” (Popper
2002, 278)
And in “Realism and the Aim of Science”:
“…we may – and indeed, undoubtedly do – believe
this. And perhaps this belief is psychologically important in our search for
true laws. But even this would not make it a methodological presupposition; it would merely make it a psychological one.” (Popper 2005, 75)
So, Popper admits that
the belief in the existence of “laws, …
regularities which we can uncover – discover” is psychologically necessary for scientific activity, but seems to
deny that it “belongs chiefly to methodology or the theory of knowledge” (Popper 2005, 74)
I think that this
distinction is mistaken and that in fact that belief (presupposition) is
incorporated in Popper’s methodology, as remarked above. For more detailed
discussion see my article “Popper on Metaphysical Presuppositions of Science”.
Isn’t This Induction?
“…there may be a "whiff"
of inductivism here. It enters with the vague realist assumption that reality,
though unknown, is in some respects similar to what science tells us or, in
other words, with the assumption that science can progress towards greater
verisimilitude.” (Popper 1992, 1193)
“For a theory which has been well corroborated can
only be superseded by one of a higher level of universality; that is, by a
theory which is better testable and which, in addition, contains the old, well
corroborated theory – or at least a good approximation to it. It may be better,
therefore, to describe that trend – the advance towards theories of an ever
higher level of universality – as ‘quasi-inductive’. The quasi-inductive
process should be envisaged as follows. Theories of some level of universality
are proposed, and deductively tested; after that, theories of a higher level of
universality are proposed, and in their turn tested with the help of those of
the previous levels of universality, and so on. The methods of testing are
invariably based on deductive inferences from the higher to the lower level...” (Popper 2002, 276)
In e-mail discussion about how to interpret Popper’t
theory of falsification S.Law has remarked that “Popper's starting point is that
there is no sound inductive reasoning”, while, on my interpretation, “he
actually ends up defending a version of it – falsification will be a kind of
inductive argument”.
Quite a many Popper’s
critics have seen his theory as veiled inductivism, and quite a many of his
more radical followers complained about remnants of inductivism in his
philosophy. For example, Stephano Gattei qualifies as “Popper’s Later Fall Back
into Inductivism” (Gattei 2004, 461) Popper’s views on the role of
corroboration in science, as expressed in the article “Truth, Rationality and
the Growth of Scientific Knowledge” of “Conjectures and Refutations”.
So, isn’t my interpretation
of critical rationalism a kind of induction-theory? My answer is: “No”, in the meaning in which I see it appropriate
to use the word 'induction', and in which Popper persuasively argued that
induction is neither existent nor needed for knowledge. In some other meaning
of the word 'induction', it surely can be “Yes”. Surely, you can use the word 'induction' in some other meaning, in
which neither I nor Popper would deny induction. No one can forbid anybody to use any word in whatever meaning he likes. I
have no irrational fear of the word. Neither Popper had. You can clearly see it
in quotations above.
It won’t be superfluous
to remind what exactly was that 'induction' which Popper rejected.
Popper argued that there
is no induction in traditional weighty philosophical meaning which engenders
Hume’s problem of induction – induction as valid inference from true singular
statements (of observable facts) to universal statements-theories which ensures
truth or, at least, high probability ("in a sense laid down by the
calculus of probability"). The method which is really used in empirical
sciences – method of conjectures and refutations – is not that induction in two
major features.
First, it has totally
different logical structure. Theory of induction says that acquisition of new
theoretical knowledge begins with observations of individual facts, then those
facts are inductively processed, and, as result, you get universal theories. Popper’s
theory says that it goes quite otherwise: we begin with a problem; then we
conjecture hypotheses as its possible solutions; then we test them with
(experimental) observations of individual events; as result, we tentatively
accept as true those hypotheses which have successfully passed examinations,
and reject those which didn’t.
Second, it doesn’t ensure
neither truth nor high probability.
To
quote Popper:
"What we do use is a method of
trial and of the elimination of error; however misleadingly this method may
look like induction, its logical structure, if we examine it closely, totally
differs from that of induction. Moreover, it is a method which does not give
rise to any of the difficulties connected with the problem of induction."
(Popper 1974, 1015)
Bibliography
Agassi, J. a.
Verisimiluitude. http://www.tau.ac.il/~agass/joseph-papers/ver.pdf
Agassi, J. b.
Karl Raimund Popper (1902-1994).
http://www.tau.ac.il/~agass/joseph-papers/Popperiep.pdf
Bartley, W. 1984. The Retreat to Commitment.
Campbell, D. 1993.
Evolutionary Epistemology. In
Evolutionary Epistemology, Rationality, and the Sociology of Knowledge. Ed.
By G.Radnitzky and W.W.Bartley, III.
Campbell, D. 1993. Blind
Variation and Selective Retention in Creative thought as in Other Knowledge
Processes. In Evolutionary Epistemology,
Rationality, and the Sociology of Knowledge. Ed. By G.Radnitzky and
W.W.Bartley, III.
Gattei, S. 2004. Karl
Popper’s Philosophical Breakthrough. Philosophy
of Science, 71: 448-466.
Law, S. 2004. Philosophy Gym.
Law, S. 2007. The Great Philosophers.
Miller, D. 1994. Critical Rationalism: A Restatement and
Defense.
Miller, D. 2006. Out of Error. Further Essays on Critical
Rationalism.
Musgrave,
A. 2004. How Popper
[Might Have] Solved the Problem of Induction. Philosophy, 79(1): 19-31.
Musgrave,
A. 2007. Critical Rationalism. In E.Suárez-Iñiguez, ed., The
Power of Argumentation (Poznań Studies in the Philosophy of the
Sciences and the Humanities, vol.93).
Musgrave, A. 2009. Realism and Surrealism about Science. In A.Musgrave, Secular Sermons.
Popper, K. 1972. Objective Knowledge: An Evolutionary
Approach.
Popper, K. 1974. Replies to My Critics. In P.A.Schilpp, ed., The Philosophy of Karl Popper.
Popper, K.
Popper, K. 1993.
Popper, K. 2002. The Logic of Scientific Discovery.
Popper, K.
2005. Realism and the Aim of Science.
Popper, K. 2008. Conjectures and Refutations.
Ñàäîâñêèé, Â.Í., 2000. Ýâîëþöèîííàÿ ýïèñòåìîëîãèÿ Êàðëà Ïîïïåðà. Â Ýâîëþöèîííàÿ ýïèñòåìîëîãèÿ è ëîãèêà ñîöèàëüíûõ íàóê. Ðåä. Â.Í.Ñàäîâñêèé. Ì.: Ýäèòîðèàë ÓÐÑÑ.
[1] For, example, V.N.Sadovsky, who has
made the largest work in translating Popper’s books and popularizing his ideas
in Russia, in the introductory article for collection about evolutionary
epistemology and the logics of social sciences, when talking about “the newest
literature” on the problems dealt with in Popper’s philosophy mentions, besides
his own articles, only one book − “the recent book of one of the nearest
Popper’s colleagues David Miller ‘Critical rationalism’ which expounds not only
modern interpretation of what now may be called Popper’s-Miller’s critical
rationalism, but also states interesting ideas about many criticisms of
Popper’s theories” (Ñàäîâñêèé Â.Í. “Ýâîëþöèîííàÿ ýïèñòåìîëîãèÿ Êàðëà Ïîïïåðà íà ðóáåæå ÕÕ
è ÕÕI ñòîëåòèé” // Ýâîëþöèîííàÿ ýìèñòåìîëîãèÿ è ëîãèêà ñîöèàëüíûõ íàóê. Êàðë Ïîïïåð è åãî êðèòèêè, Ìîñêâà: Ýäèòîðèàë ÓÐÑÑ, 2000).
[2] It
can be objected, that all this is subjective (or intersubjective) evaluation
(judgement), which can’t be expressed in objective, formalizable terms. I
readily admit this. My principal point is exactly that no knowledge and no
rationality is possible without subjective (intersubjective) evaluations
(judgements), which can’t be expressed in objective, formalizable terms.
"Epistemology without a Knowing Subject"[2]
cannot be the whole of epistemology: for some purposes it can be useful to
abstract a knowing subject for a while.
But you can’t get adequate understanding of how human beliefs (including that
specific realm we call scientific knowledge) are acquired and changed, what are
necessary constituents and optional possibilities of this process, and which of
those optional possibilities are more reasonable (rational), until we
reintroduce a subject into the picture.
Especially,
if we are talking about rationality,
it is exactly the matter of how a person (subject) acquires and/or changes
his/her beliefs. By beliefs I mean any idea, theory, proposition, attitude
which person accepts, i.e. judges (or, using Miller’s favourite term, classifies) as true (or truthlike,
sound, plausible, commendable etc.). Here I agree with Bartley that rationality
is holding our views open to criticism (or, in other words, pursuit of truth
plus fallibilism).
[3] Joseph Agassi suggests that it was
really Einstein’s theory, which Popper assimilated in his epistemology and
tried (unsuccessfully) to develop into stronger theory:
"The idea of
verisimilitude is implicit in the writings of Albert Einstein ever since 1905…
All his life Einstein presented new ideas as yielding older established ones as
special cases and first approximations. The news has reached the philosophical
community via the writings of Sir Karl Popper half-a-century after Einstein's
trailblazing conception…" (Agassi a) "This
requirement for resemblance to the truth, for verisimilitude, Einstein
characterized as the requirement that the new theory explains the success
(success as explanation and success as having passed severe tests) of its
predecessors by presenting them as special cases and good approximations.
Popper offered a more precise theory of verisimilitude that was devastated by
criticism and he withdrew it." (Agassi b)
[4] I have added this paragraph after
reading Miller’s article "Induction: a Problem Solved" in his later
book Out of Error (2006). There, Miller refers to several
writers who have observed that "for most of their histories science and
technology proceeded independently of each other" and adduces many
examples of prominent “pure” scientists who denied the possibility of some
technological accomplishments which later were realized, or couldn’t envision
technological application of the foremost scientific theories of their times.
For example, Plank, Einstein and Bohr “couldn’t envision applications of atomic
theories”. Miller asks: “Why should this have been so if science is, as so
often thought, the inspiration of technology? The plain answer … is that
science has very little to contribute to the advance of technology except for
criticism; and that this criticism can always in principle be replaced by
empirical rather than theoretical criticism; that is, but practical testing.”
(Miller 2006, 122-123)
On my view, this “plain answer” is mistaken. It is especially obvious if
we apply that answer to the case of “applications of atomic theories”. It is
obviously not the case that all those tremendous “applications”, invented in XX
century, could be as well invented if there was no atomic theory and
theoretical physics at all, that all that these theories do could be replaced
by empirical testing! As well, there would hardly be considerable technological
applications of electricity without theories of electricity. And chemical
technologies would do very poorly without chemical theories. Such examples may
be multiplied in a very long list.
As for the question “Why many prominent “pure” scientists were mistaken
about technological applications of scientific theories or failed to envision
them?”, I think that correct answer is really plain: great scientists are just
as fallible and lacking prescience on a great many questions as we are.
I think that in this context it is relevant to remind of Donald
Campbell’s and Popper’s discussion about evolutionary epistemology. Campbell
have proposed, and Popper have agreed, that variations
in the process of evolution of knowledge or, more widely, of problematic
situation and problem-solving facilities – which encompasses biological
evolution as well as human (conscious, conceptual) knowledge – would be more
correctly described as “blind”, not “chance”. These variations (mutations in
the nature, hypotheses in human knowledge) are “blind” in the sense that they
aren’t predetermined by previous developments. But, nevertheless, they aren’t
purely chance, they are softly
directed by previous developments; previous developments determine dispositions and limitations for new variations. I think that scientific theories
similarly direct technological developments – both by creative suggestion and
critical elimination.