CHAPTER 2
POPPER AND DEWEY'S ACCOUNTS OF SCIENCE AS PROBLEM SOLVING
[--- Unable To Translate Graphic ---]
Any philosophical investigation of the role of problems in science must
begin with a discussion of Popper and Dewey. If we paint with broad strokes
and ignore for the moment philosophical niceties (such as their views on
truth and induction!), there are striking parallels between Dewey's and
Popper's theories of the growth of knowledge. Both are thorough-going fallibilists
- none of our beliefs are immune to criticism and revision. The quest for
certainty is ill-advised. Both emphasize the activity involved in even the
most mundane cases of observation. Information is not simply impressed on
blank wax tablets or accumulated in empty heads; it is always the result
of a process of search, selection, interpretation. Both note the continuities
between common-sense inquiry and scientific reasoning while acknowledging
the human tendency to pay more attention to events which corroborate our
theories than to those which contradict it - critical reasoning must be
learned.
Both stress the non-algorithmic aspects of reasoning - "thinking is
not a sausage machine." Both philosophers view scientific inquiry in
particular and knowledge acquisition in general as problem-solving. Since
it is impossible either to analyze or to justify all our preconceptions,
inquiry begins when we are faced with a gap or inconsistency within our
knowledge system - perhaps something unexpected happens. Somehow a solution
suggests itself to us. If we are reasoning critically, we will suspend judgment
no matter how attractive the proposed solution may be. Instead, we will
design experiments to test the implications of our hypothesis and also investigate
alternative solutions to our problem. Only if the proposed solution survives
all the tests we can muster, can we consider the problem to be solved -
and only for the time being since our knowledge situation is constantly
changing.
Where Dewey and Popper differ - and I believe these differences to be deep
and irreconcilable - is in their philosophical theories of these temporary
end-points of inquiry. For Dewey, knowledge consists of psychologically
compelling beliefs whose assertability is inductively warranted if the beliefs
work satisfactorily. Popper, on the other hand, does not "believe in"
beliefs, nor in inductive support. For him, knowledge is best understood
when completely separated from the minds of the organisms which produced
it. It consists of propositions which have withstood severe criticism. The
only positive credential of knowledge statements is our failure to refute
them - so far. Our aim is to find theories which are true in the correspondence
sense although we can never hope to know they are true.
Much philosophical energy has already been spent on debates about the nature
of truth and knowledge - and rightly so: these are important issues. But
the focus of my analysis is a different one. I want to look more carefully
at what Dewey and Popper say about the starting points of inquiry. What
do they say about the nature of problems? And more importantly, what are
their theories about how problems should be evaluated? However, before plunging
into an analysis of these views we first need to recall what their contemporaries
were saying (or failing to say) about scientific inquiry. Only then can
we appreciate the contributions of these pioneering attempts to understand
science as problem solving.
2.1 The Program of Logical Positivism
In this book we will be discussing a variety of post-positivistic theories
of scientific progress. Since their most salient shared feature is a rejection
of logical positivism, let's quickly sketch what they were reacting against.
[For a fuller account of what he calls "the received view" see
Fred Suppe's The Structure of Scientific Theories. The philosophical background
to the program of the Vienna Circle is analyzed in A. Coffa, Towards the
Vienna Station.]
As the name indicates, logical positivism (or logical empiricism) assumes
that knowledge comes from the senses (reason plays an important processing
role, but does not provide content). One job of the philosopher is to subject
belief systems to logical analysis, separating out positive knowledge claims
from meaningless babble or unverified speculation. When they turned to science,
the logical positivists spent a lot of time axiomatizing scientific theories
or at least debating how they should be axiomatized.
There was general agreement that the structure of science was roughly represented
by a "layer-cake model". (This label was originally introduced
by a critic, Feyerabend, as a term of disparagement, but Feigl, who had
helped coin the phrase "logical positivism" in 1931 [Encycl. of
Phil.] willingly accepted it [Minn.III]). On this model, science consists
of three layers. On the bottom are singular observation reports -- positivists
argued amongst themselves about the exact nature of these: Should they describe
sensations or pointer readings? In the middle are empirical generalizations,
often called "laws", such as the laws named after Kepler, Boyle,
Hooke, and Galileo. Topping off the cake are theories, such as the Kinetic
Theory of Gases, Quantum Theory, Electromagnetic Theory.
Deductive inferences lead from the top of the cake down to the lower level,
thus permitting us both to explain past events (by showing how they are
subsumed under "covering laws") and to predict ones in the future,
events which have not yet been observed. Inductive support percolates up
through the structure - the observation reports support the laws and the
laws in turn support the theories which unify them.
--------------------------
Insert Fig. 2.1 about here[--- Unable To Translate Graphic ---]
--------------------------
The general flavor of this model is quite reminiscent of Whewell's Inductive
Tables where he charts the historical development of sciences such as optics,
astronomy and mechanics or Bacon's description of science as the ascent
of a "ladder of axioms" through induction. However, what was new
about the logical positivists' account was their insistence that the layers
of the cake be characterized in terms of language. According to the logical
positivists, scientific systems should be expressed using two different
languages. The top layer would be expressed in theoretical language, using
terms such as "electron" or "atomic number", while the
lower level would be written in observation language, using terms such as
"red" or "weighs two grams". (Again I will ignore the
debate about whether the observation language should be phenomenalist or
physicalist.)
This two-language account introduced a sharp epistemological hiatus between
the various levels of science - how could one ever hope to get knowledge
about theoretical entities? There had always been some concern about how
one could get knowledge about underlying mechanisms - recall the Atomic
Debates in the mid-nineteenth century or Descartes' discussion about whether
one could discover or infer the mechanical micro-process which would explain
refraction or magnetism - but the positivists both dramatized the problem
and added an entirely new wrinkle: How could we be sure that a hypothesis
which purported to describe micro-processes or unobservable properties was
even meaningful?
The history of the development of the positivists' approach towards the
problem of meaning is a long and complicated one [see Coffa], but many of
their attitudes are encapsulated in the morals they drew from the replacement
of Newtonian mechanics by Relativity Theory. It was perhaps understandable
that a respectable scientific theory might make inaccurate predictions in
unfamiliar domains, which involved unusually high velocities or large gravitational
forces. But what was shocking was that Newtonians had uncritically made
assumptions about the nature of space and time which could not even be tested
and, hence, were without empirical significance. And any claim which could
not be tightly connected to experience, according to the positivists, was
also without cognitive significance. So Newton's talk about space and time
was not just unverified, or even false - it was unverifiable in principle,
and hence meaningless in a strict sense. To prevent future fiascos, scientists
should always operationalize their theoretical claims. No one can guarantee
ahead of time that an untested scientific hypothesis is true, but through
philosophical analysis we should at least be able to guarantee that it has
meaning!
For over thirty years there was broad support for the layer-cake model amongst
philosophers of science as well as wide agreement about what remained to
be done. To complete their program, logical positivists needed to:
1) Give a precise account of how theoretical terms gained their meaning
(was it through operational definitions, reduction sentences, implicit definitions,
or ?)
2) Determine whether theoretical terms refer and if so, to what?
3) Describe an inductive logic which in principle would permit the calculation
of the degree of reasonable belief which should be assigned any meaningful
scientific hypothesis.
4) Provide a model of explanation which did not invoke metaphysical notions
such as causality or
CHAPTER 2
POPPER AND DEWEY'S ACCOUNTS OF SCIENCE AS PROBLEM SOLVING
[--- Unable To Translate Graphic ---]
Any philosophical investigation of the role of problems in science must
begin with a discussion of Popper and Dewey. If we paint with broad strokes
and ignore for the moment philosophical niceties (such as their views on
truth and induction!), there are striking parallels between Dewey's and
Popper's theories of the growth of knowledge. Both are thorough-going fallibilists
- none of our beliefs are immune to criticism and revision. The quest for
certainty is ill-advised. Both emphasize the activity involved in even the
most mundane cases of observation. Information is not simply impressed on
blank wax tablets or accumulated in empty heads; it is always the result
of a process of search, selection, interpretation. Both note the continuities
between common-sense inquiry and scientific reasoning while acknowledging
the human tendency to pay more attention to events which corroborate our
theories than to those which contradict it - critical reasoning must be
learned.
Both stress the non-algorithmic aspects of reasoning - "thinking is
not a sausage machine." Both philosophers view scientific inquiry in
particular and knowledge acquisition in general as problem-solving. Since
it is impossible either to analyze or to justify all our preconceptions,
inquiry begins when we are faced with a gap or inconsistency within our
knowledge system - perhaps something unexpected happens. Somehow a solution
suggests itself to us. If we are reasoning critically, we will suspend judgment
no matter how attractive the proposed solution may be. Instead, we will
design experiments to test the implications of our hypothesis and also investigate
alternative solutions to our problem. Only if the proposed solution survives
all the tests we can muster, can we consider the problem to be solved -
and only for the time being since our knowledge situation is constantly
changing.
Where Dewey and Popper differ - and I believe these differences to be deep
and irreconcilable - is in their philosophical theories of these temporary
end-points of inquiry. For Dewey, knowledge consists of psychologically
compelling beliefs whose assertability is inductively warranted if the beliefs
work satisfactorily. Popper, on the other hand, does not "believe in"
beliefs, nor in inductive support. For him, knowledge is best understood
when completely separated from the minds of the organisms which produced
it. It consists of propositions which have withstood severe criticism. The
only positive credential of knowledge statements is our failure to refute
them - so far. Our aim is to find theories which are true in the correspondence
sense although we can never hope to know they are true.
Much philosophical energy has already been spent on debates about the nature
of truth and knowledge - and rightly so: these are important issues. But
the focus of my analysis is a different one. I want to look more carefully
at what Dewey and Popper say about the starting points of inquiry. What
do they say about the nature of problems? And more importantly, what are
their theories about how problems should be evaluated? However, before plunging
into an analysis of these views we first need to recall what their contemporaries
were saying (or failing to say) about scientific inquiry. Only then can
we appreciate the contributions of these pioneering attempts to understand
science as problem solving.
2.1 The Program of Logical Positivism
In this book we will be discussing a variety of post-positivistic theories
of scientific progress. Since their most salient shared feature is a rejection
of logical positivism, let's quickly sketch what they were reacting against.
[For a fuller account of what he calls "the received view" see
Fred Suppe's The Structure of Scientific Theories. The philosophical background
to the program of the Vienna Circle is analyzed in A. Coffa, Towards the
Vienna Station.]
As the name indicates, logical positivism (or logical empiricism) assumes
that knowledge comes from the senses (reason plays an important processing
role, but does not provide content). One job of the philosopher is to subject
belief systems to logical analysis, separating out positive knowledge claims
from meaningless babble or unverified speculation. When they turned to science,
the logical positivists spent a lot of time axiomatizing scientific theories
or at least debating how they should be axiomatized.
There was general agreement that the structure of science was roughly represented
by a "layer-cake model". (This label was originally introduced
by a critic, Feyerabend, as a term of disparagement, but Feigl, who had
helped coin the phrase "logical positivism" in 1931 [Encycl. of
Phil.] willingly accepted it [Minn.III]). On this model, science consists
of three layers. On the bottom are singular observation reports -- positivists
argued amongst themselves about the exact nature of these: Should they describe
sensations or pointer readings? In the middle are empirical generalizations,
often called "laws", such as the laws named after Kepler, Boyle,
Hooke, and Galileo. Topping off the cake are theories, such as the Kinetic
Theory of Gases, Quantum Theory, Electromagnetic Theory.
Deductive inferences lead from the top of the cake down to the lower level,
thus permitting us both to explain past events (by showing how they are
subsumed under "covering laws") and to predict ones in the future,
events which have not yet been observed. Inductive support percolates up
through the structure - the observation reports support the laws and the
laws in turn support the theories which unify them.
--------------------------
Insert Fig. 2.1 about here[--- Unable To Translate Graphic ---]
--------------------------
The general flavor of this model is quite reminiscent of Whewell's Inductive
Tables where he charts the historical development of sciences such as optics,
astronomy and mechanics or Bacon's description of science as the ascent
of a "ladder of axioms" through induction. However, what was new
about the logical positivists' account was their insistence that the layers
of the cake be characterized in terms of language. According to the logical
positivists, scientific systems should be expressed using two different
languages. The top layer would be expressed in theoretical language, using
terms such as "electron" or "atomic number", while the
lower level would be written in observation language, using terms such as
"red" or "weighs two grams". (Again I will ignore the
debate about whether the observation language should be phenomenalist or
physicalist.)
This two-language account introduced a sharp epistemological hiatus between
the various levels of science - how could one ever hope to get knowledge
about theoretical entities? There had always been some concern about how
one could get knowledge about underlying mechanisms - recall the Atomic
Debates in the mid-nineteenth century or Descartes' discussion about whether
one could discover or infer the mechanical micro-process which would explain
refraction or magnetism - but the positivists both dramatized the problem
and added an entirely new wrinkle: How could we be sure that a hypothesis
which purported to describe micro-processes or unobservable properties was
even meaningful?
The history of the development of the positivists' approach towards the
problem of meaning is a long and complicated one [see Coffa], but many of
their attitudes are encapsulated in the morals they drew from the replacement
of Newtonian mechanics by Relativity Theory. It was perhaps understandable
that a respectable scientific theory might make inaccurate predictions in
unfamiliar domains, which involved unusually high velocities or large gravitational
forces. But what was shocking was that Newtonians had uncritically made
assumptions about the nature of space and time which could not even be tested
and, hence, were without empirical significance. And any claim which could
not be tightly connected to experience, according to the positivists, was
also without cognitive significance. So Newton's talk about space and time
was not just unverified, or even false - it was unverifiable in principle,
and hence meaningless in a strict sense. To prevent future fiascos, scientists
should always operationalize their theoretical claims. No one can guarantee
ahead of time that an untested scientific hypothesis is true, but through
philosophical analysis we should at least be able to guarantee that it has
meaning!
For over thirty years there was broad support for the layer-cake model amongst
philosophers of science as well as wide agreement about what remained to
be done. To complete their program, logical positivists needed to:
1) Give a precise account of how theoretical terms gained their meaning
(was it through operational definitions, reduction sentences, implicit definitions,
or ?)
2) Determine whether theoretical terms refer and if so, to what?
3) Describe an inductive logic which in principle would permit the calculation
of the degree of reasonable belief which should be assigned any meaningful
scientific hypothesis.
4) Provide a model of explanation which did not invoke metaphysical notions
such as causality or
intelligibility.
There was also broad agreement on the types of solutions which would be
acceptable. The philosophical accounts of science should be logical and
normative, not psychological or descriptive. Thus, the problem of how science
develops never was on the agenda. To the extent that the growth of science
was rational, it was assumed to occur by a sort of sedimentation process
which laid down successive layers of statements of increasing generality.
There might be temporary debates about items on the very top of the cake
or in new domains out at the side, but the growth of science should be a
rather orderly process of accumulation. The question of where new concepts
or hypotheses came from was an issue for psychology, not for philosophy
- hence, Reichenbach's much quoted distinction between the context of discovery
(of interest only to the psychologist or historian of science) and the much
more important context of justification (of interest to the philosopher,
working scientist and any one who wishes to apply science).
2.2 The Decline of Logical Positivism
A detailed history of the decline of logical positivism has yet to be written.
(In Stoppard's play Jumpers, the character George Moore says that at one
moment in history the onus of proof passed from the atheist to those who
wished to defend theism, "...quite suddenly, secretly, the noes had
it".) Perhaps something similar happened with logical positivism. There
are still many philosophers of science (perhaps even a majority) who work
very much in a soft positivist tradition. However the central problems of
the positivist program have been quietly de-coupled and the constraints
on their solutions slowly relaxed. Thus people today who work on explanation
may invoke causes; some confirmation theorists deal with subjective degrees
of belief; people who theorize about meanings may invoke prototypes, metaphors
or Gestalts. The rigid theory/observation distinction has largely been dropped,
at least as a tool of analysis. This has opened the way for the so-called
'semantic view' of theories which posits abstract set-theoretic structures
which are then instantiated by various physical systems.
But the most exciting philosophical developments have arisen out of conscious
attempts to provide radical alternatives to the whole positivist program.
In the course of this book I will be summarizing some of these alternative
philosophical theories and describing the special strengths and weaknesses
of each. These theories of scientific progress vary so enormously that the
philosophers cannot be considered a school, or even a tradition (though
labels such as "historicist" or "naturalist" have been
proposed). However, I think they do share certain assumptions.
First of all, their views are anti-foundationist. Science does not and cannot
grow through monotonic accumulation of positive knowledge claims. Whereas
the logical positivists viewed the Einsteinian Revolution as a methodological
indictment of Newtonian scientists - a sort of "never again" cautionary
tale - Popper saw it as just the latest example of how critical debate could
cause us to revise our most fundamental assumptions and called for "Revolution
in Permanence". Kuhn took scientific revolutions to be a crucial feature
of the development of science and emphasized how extensively they shook
up our pictures of the world. (No layer cake could hope to survive the tremors
of a Kuhnian Gestalt switch.) So science was fallible through and through.
And no amount of positivistic epistemological puritanism would help because
observations were necessarily theory-laden and theoretical systems were
in turn inevitably influenced by metaphysical and ontological assumptions
of a non-empirical nature.
As a consequence any useful picture of a scientific system had to be much
richer and more interactive than the austere layer cake model. And since
no elements in the system were privileged and could in principle be revised,
it was also important to develop a sophisticated account of the development
of the system over time. What has emerged is a general conception of science
which I will call the "tapestry model" (or perhaps I should say
models because, as we will see later, there are important differences amongst
these past-positivists).
--------------------------
Insert Fig. 2.2 about here
--------------------------
Surrogates of the elements of the old layer cake model still reside in the
center (although the language-based theory/law distinction is no longer
important). But they are surrounded by other items which interact with them
in all sorts of ways. Thus theories may gain meaning from favored models
and analogies not just through their connection to observations. The 'empirical'
testing and confirmation of the central theories is a complex process which
depends not just on observation reports but also on all sorts of other theories,
including theories about how instruments work and theories about other causally
relevant phenomena. In short, the theories and observation reports of any
science reside in the center of a complicated fabric which is constrained
on all sides by a 'framework' which is itself subject to revision. There
is no foundation and the paths of transmission of meaning, evidential support,
and critical appraisal run horizontally as well as vertically.
The general problem for post-positivist philosophers will be to describe
in more detail patterns within the tapestry and the processes by which it
is woven and re-woven. (The problem of describing the development of science.)
Some will specialize in the role of heuristics - no longer will it be an
oxymoron to speak of a "logic of discovery". Others will argue
that even the above model is too austere and study effects of scientific
institutions on the progress of science.
2.3 Dewey's Theory of Problems
Although logical positivism was the dominant view of science during
the first half of this century, it was not the only one. And whatever one
might think of its account of mature, finished science, positivism provided
no account of the process of learning about the world, whether that be the
discovery of new scientific knowledge or the imparting of past discoveries
to children. It was to just these issues that John Dewey directed his attention.
Dewey describes and analyzes the starting points of inquiry in many places.
The discussion below is my synthesis (and reconstruction) based primarily
on How We Think (1910) and Logic, The Theory of Inquiry (1938). Juxtaposing
passages written over several decades will obviously blur any subtle changes
in Dewey's ideas on the subject. But, since the result is a rich and interesting
theory of the origin and structure of problems, I believe the enterprise
to be worthwhile. As is typical of Dewey's discussion of any subject, he
draws a variety of useful analytical distinctions while emphasizing the
continuities and developmental processes which actually occur in nature.
In a similar spirit, I will distinguish between the biological, psychological
and intellectual aspects of human problems while recognizing the overlaps.
A prerequisite for any sort of deliberate inquiry is curiosity - Dewey calls
it a "natural resource" for thinking (HWT, p. 29). In animals
and very young children, curiosity takes the form of excess "organic
energy" (p.31), a sort of "physiological uneasiness" which
causes a profusion of more or less random exploratory behavior - poking,
prying, sucking, fingering and thumping (p.31).
Dewey implies that the major function of this sort of behavior is the accumulation
of information which may come in handy later on. He views the two-year-old's
endless strings of "why" questions as being primarily an attempt
to "eke out his store of experience" (p.32). It is only later
that the "whys" are genuine requests for explanations. Yet, the
more directed activities of the exploring child are akin to the experimental
processes which will predominate later. And there is a developmental continuity
between the almost purely physiological curiosity of a baby in a highchair
and the sense of intellectual wonder of an Einstein contemplating travels
on a light ray. In certain situations (exactly which ones we will describe
later), there is superimposed onto the restless biological energy of curiosity
the more-or-less irritating and frustrating psychological state of doubt.
Dewey concurs with Pierce and James in finding the adoxastic state of active
agnosticism painful, but necessary for effective inquiry. Pierce is especially
emphatic on both of these points: "Doubt is an uneasy and dissatisfied
state from which we struggle to free ourselves...(R., p. 8). (He compares
doubt to both hunger and the irritation of a nerve.) "...the mere putting
of a proposition into the interrogative form does not stimulate the mind
to any struggle after belief. There must be a real and living doubt, and
without this, all discussion is idle" (R., p. 9).
James describes the psychological condition which motivates inquiry as "an
inward trouble to which his mind till then had been a stranger, and from
which he seeks to escape..." (J., p.31). In a similar vein Dewey views
the state as one of "perplexity, hesitation, doubt" (p. 9). It
is "a condition of mental unrest and disturbance" (p. 13). It
is tempting to adopt any solution to our problem immediately because the
"suspense is likely to be somewhat painful" (p. 13). Dewey, like
Pierce, stresses that the difficulty must be felt before effective inquiry
can proceed. In applying his theory to education, he says:
"Instruction in subject matter that does not fit into any problem already
stirring in the student's own experience, or that is not presented in such
a way as to arouse a problem is worse than useless for intellectual purposes."
(p. 199)
And in Logic we read: "...to set up a problem that does not grow out
of an actual [indeterminate] situation is to start on a course of dead work..."
(p. 108). Although feeling doubt is a necessary antecedent condition for
inquiry, Dewey points out that such feelings can be pathological in origin
and have nothing to do with the agent's actual situation (Logic, p. 106)
so the feeling is not sufficient.
Curiosity supplies the general energy for exploration. The irritation of
a specific doubt stimulates us to seek the relief of belief. But which situations
provoke doubt? Where do problems come from? Dewey says that a problem arises
whenever anything - "no matter how slight and commonplace in character
- perplexes and challenges the mind so that it makes belief at all uncertain..."
(HWT, p. 9). One type of problem arises in what Dewey calls a "forked-road
situation" (p. 11). In such cases, our destination (goal) is clear,
but we are uncertain as to which road (means) leads to it. Here, the main
purpose of inquiry is to fill in gaps in our knowledge in order that we
may achieve some practical end. We are resolving dissonances between what
we have and what we want.
A second sort of problem described by Dewey is the case in which something
unexpected happens. (In his example, the weather suddenly turns cool.) Here
the problem, as Dewey describes it, is to find the significance of the perplexing
observation. Does the cooler air indicate rain? Although we may have immediate
practical reasons for wanting to forecast the weather, we need not. Humans
seem to need to have their ideas fit together in a unified, orderly manner.
Animals also seek out suitable methods for achieving their goals and may
be surprised by unusual events, but they do not have full-fledged problems
in Dewey's sense. He stresses that an indeterminate situation does not lead
to inquiry until it is judged to be problematic (Logic, p. 107). So we see
there is an important intellectual or cognitive component in the most mundane
of human problems. Scientific problems differ from commonsense problems
in that they are less immediately tied to the practical goals of "establish[ing]
objects of use and enjoyment..." (Logic, p. 115). The subject matter
of science is also of a higher level of abstraction.
To summarize, according to Dewey knowledge is the outcome of inquiry, and
inquiry begins with problems. For a problem to occur to X, it would appear
that for Dewey at least four conditions must be fulfilled (this is my list,
not Dewey's):
(1) X must really be in a situation which is "open" or "indeterminate,"
(2) that situation must induce in X a feeling of specific doubt,
(3) X must judge the situation to be problematic by at least vaguely articulating
what is doubtful about it (later analysis may sharpen the problem), and
(4) X must have the intention of resolving the indeterminacy.
2.4 Discussion of Dewey's Theory
To those modern readers who have been raised on a regimen of antipsychologistic
philosophy, the insistence on the necessity of feeling doubt may be puzzling.
If I cognitively appraise my situation as problematic and set off to investigate
it, why must I also experience the irritation of doubt in order for successful
inquiry to ensue?
In Pierce's case, I think he required real, living doubt in order to rule
out certain sorts of abstruse scholastic (in the bad sense) philosophizing
which he found to be completely sterile. Requiring the feeling is also a
barrier to Pyrrhonian skepticism - most of us are unable to muster up sincere
doubt about the number of fingers on our hands.
In Dewey's case, the requirement on emotional state fits in well with his
general psycho-epistemological approach and his extension of the biological
concept of homeostasis to the growth of knowledge. When our belief system
is unsettled, we struggle to bring it back into equilibrium. The hungry
organism restlessly explores its environment until satisfied by food. The
doubting mind engages in inquiry until it finds satisfactory beliefs. The
theory also explains why poking unwanted information down kids does not
result in effective learning.
But is the theory true? Is the feeling of specific doubt necessary for successful
inquiry? And is instruction in material which satisfies no problem in the
child's life "worse than useless?" Emotional states are notoriously
difficult to disambiguate and identify, and I know of no relevant research
on this question, so I'll just rely on informal examples drawn from my own
experience.
Let's begin by looking at paradigm cases of intense doubt. Especially excruciating
is the state of romantic uncertainty where one is reduced to pulling off
daisy petals: "She loves me - she loves me not." A more serious
example is in an operating room: "Will they find cancer? Will they
be able to remove it all?" The intensity of doubt in these cases seems
to accrue from two sources: First of all a lot hinges on the outcome of
our query. Secondly, there is really very little we can do in the way of
appropriate inquiry - pulling petals or reading the patient's horoscope
in the newspaper are clearly idle and ineffective substitutes for real inquiry
so we are faced with a second problem, namely what methods can we use to
solve the first problem? So far so good for Dewey's theory - these deeply
problematic situations do induce intense doubt which builds up unless discharged
by inquiry.
Let us now consider some examples of intense inquiry and ask what feelings
seem to be motivating the activity. Recently I watched two children play
Ripper, a computerized adventure game. The object of the game is to find
and destroy Jack-the-Ripper, before too many maidens are killed. To do so,
there are many sub-problems to be solved - codes to decipher, secret doors
to be opened, rats and cobras barring the way - and there are a variety
of resource people available to help - Sherlock Holmes, Florence Nightengale,
Roentgen, Madame Curie. The children exhibited classic problem-solving techniques.
Some sub-problems could only be solved by brute-force trial and error elimination,
others required insight and creativity. At certain junctures, they needed
to consult a dictionary or encyclopedia. (At one point they even telephoned
the local reference librarian!) They soon found out how important it was
to keep records of both their successful and unsuccessful conjectures.
All of these activities were conducted with the utmost seriousness over
a period of nearly six hours. (Suppertime finally intervened.) By the end
of the afternoon, although the game was not over, they had learned a good
deal - how to open the secret door, which wing of the building the Ripper
was in, how to drive off the rats, etc. An adult might add that they had
also learned some historical facts (e.g., that Roentgen invented x-rays)
and perhaps even some methodological maxims (e.g., the importance of keeping
records).
But was the children's activity stimulated by the irritation or tension
of specific doubt? Was it really the need to have specific knowledge of
where the Ripper was or how to open the door that kept them going? I think
not: the major satisfaction they seemed to get out of the game was the sheer
joy of problem-solving. They didn't really care about the specific content
of the beliefs which they formed (it didn't matter where the Ripper was
as long as they caught him) and, hence, it is misleading to say that it
was specific doubts which impelled them. Their primary motive was not to
find out about the Ripper or the door or the rats; what impelled them was
not the desire to know, but the desire to achieve as problem-solvers. Yet
their efforts yielded knowledge.
Another example is that of the paid consultant or information-gatherer -
I am thinking of jobs ranging from census-taker to laboratory technician
to reference librarian to scientists working on federal contracts. Such
people may feel no personal need whatsoever to know how many homeless people
live in Bloomington or the percentage FSH in my blood serum or Karl Popper's
horoscope sign or the diameter of the exit hole of a bullet fired into a
goat. Yet they engage in more or less complicated forms of inquiry in order
to find these things out. They are not motivated by the need to resolve
doubt, but by the need to make a living. (They may also derive enjoyment
from the act of problem-solving.)
My conclusions are as follows: Much successful inquiry is motivated by doubt,
but not all is. If the doubt is too intense, bad methods of inquiry may
be adopted (especially if there is no better method immediately available).
In other cases inquiry is motivated by the pure joy of problem-solving,
not out of overwhelming interest in the subject matter itself. This emotional
mode probably predominates in games, but it also plays an important role
in mathematics and "pure" science. For example, many mathematicians
didn't care much either way how the 4-color map problem turned out - they
just wanted to solve it. And in ordinary life, people may also derive pleasure
from the elegance of a solution to a very pressing, practical problem. People
can also be induced with money to learn or make new discoveries. Knowledge
producers may be more efficient if they are also motivated by epistemological
hunger, but it is not a pre-condition for inquiry.
If my conclusions above are correct, then the set of motivating factors
available to the teacher becomes richer (although the moral decision as
to how to use them becomes more complicated). If it is socially mandatory
that certain information be mastered (not everyone would accept that there
is any such material), and if it is not possible to get every student in
a class to feel interested and intrigued by the problematic subject matter,
there are still at least two motivational techniques available - one is
to introduce the student to the pleasures of problem-solving in general.
(I think Dewey would have no objection whatsoever to this strategy and perhaps
even had something similar in mind when he said...) The second one is much
cruder - bribe the recalcitrant learner! There are obvious ethical restrictions
on the circumstances in which this technique should be used, but it at times
may be more honest than trying to make the student pretend to be interested.
By "bribing," I include any reward which is extrinsic both to
the knowledge gained and to the process of obtaining it. Common bribes include
bonuses added to the weekly allowance for good grades, gold stars, smiles
from the teacher, early recess, etc.
An interesting problem now arises as to the optimum balance between internally-induced
motivation to solve a problem and that which comes from outside. We will
return to this issue much later when we discuss the appraisal of problems,
especially in a science policy context.
2.5 Popper and Problems - An Overview
We have seen how Dewey's theory of knowledge acquisition centers on
problems, but Popper was the first to give a detailed and comprehensive
account of science as problem-solving. While a postgraduate student in London
(Chelsea College) in the sixties, I was privileged to hear Sir Karl's lectures
and attend his seminar on an almost weekly basis. During this time, I also
had frequent talks with Bill Berkson and Peggy Marchi about Joseph Agassi's
problem-centered methodology. As a result, my own "common-sense"
ideas about science and problems are probably a mishmash of first-hand and
second-hand versions, both oral and written, of a Popper/Agassi approach,
and these have undoubtedly been further modified by two decades of my own
teaching. In the discussion which follows, I will try to rely on written
sources in order to sort out which ideas came from where. However since
this is not primarily an historical essay, I have not pursued questions
of chronology or priority very far.
From the very beginning, Popper adopted an explicitly problem-centered approach
when he did philosophy. His earliest long work is entitled "Die beiden
Grundprobleme der Erkenntnis Theorie." And his 1934 Logik der Forschung
(I will refer to the 1959 English translation) is explicitly organized around
a series of philosophical problems -- e.g., the problems of induction, demarcation,
the `empirical basis,' comparing degrees of falsifiability, simplicity,
and the interpretation of probability statements.
In the Preface, Popper begins with an aphorism from Kant: "...whenever
a dispute has raged for any length of time, especially in philosophy, there
was, at the bottom of it, never a problem about mere words, but always a
genuine problem about things." (L.S.D., p. 13). He goes on to contrast
philosophy with the situation in science where there is, he claims, "a
generally accepted problem-situation" so a physicist can "attack
his problem straight away" leaving it to others to "fit his contribution
into the framework of scientific knowledge." (p. 13). In philosophy,
by contrast, there is not even agreement on whether there are genuine philosophical
discussions, so all that one can do is to "begin afresh from the beginning."
(p. 13).
As one reads Popper's classic essays in philosophy of science, especially
those in Conjectures and Refutations, a clear pattern of philosophical inquiry
emerges (one that might well be described as dialectical in the pre-Hegelian
sense of the term): He starts with a problem and shows why it is interesting
and important. He presents one or more somewhat sensible sounding solutions
to the problem, but then criticizes them in turn. (Often these solutions
comes from the work of earlier philosophers and sometimes Popper presents
revised, strengthened versions of the positions before criticizing them.)
Finally, he presents the best solution he can think of to date and shows
how it solves the original problem. What Popper generally does not do (and
here he is not alone) is to also summarize any unattractive features of
the last solution -- that job he leaves for his critics, who are presumably
better able to see any flaws.
Now one would think that philosophers would find such a pattern attractive
and easy to follow -- it's one which Aristotle frequently adopts -- but
it has on occasion been misunderstood. Sometimes commentators confuse the
positions which Popper is criticizing (which are, after all, called "solutions")
with his own positive proposal! More understandable is the difficulty that
some readers' experience with Popper's strengthened or rationally reconstructed
versions of historical texts.
The importance of beginning philosophical discussions with a clear statement
of a problem was also stressed in Popper's seminar, so much so that we students
often greeted each other in the Common Room with a cheery "Hello, what's
your problem today? And why is it interesting?" That this is not merely
a rhetorical device quickly emerges when one attempts to detect unstated
problems. Consider, for example, Hempel's famous essays on explanation.
What is his problem? Well, to explicate the concept of scientific explanation.
But why is this problem important? Here Hempel's own writings are remarkably
silent. Others have used his covering-law model as a standard of merit and
faulted accounts which do not fit it. But is this Hempel's intention? Given
his rather neutral discussion of partial, elliptical, and approximate explanations
and his shelving of the problem of deep vs. minimal explanations, it is
not clear that he was primarily interested in demarcating good explanations
from bad.
There are two other possibilities I can think of. Perhaps, Hempel was interested
in debates about the unity of science and his root problem was: Do good
explanations in all fields satisfy the same formal requirements? Or perhaps
he was interested in articulating an aim of scientific inquiry which would
not utilize the obscure notion of verae causae but yet would not lapse into
instrumentalism. Without a problem context, it is difficult to keep the
discussion about Hemplian explanation from degenerating into one about how
we're going to use words. Suppose we were to decide that paradigm cases
of functional analysis didn't fit either of Hempel's models. So what? Maybe
there are D-N, I-S and functional explanations. Let's even go on to admit
that they aren't really explanations, but only analyses. Now what? Are they
still useful, illuminating, worth looking for? Without having a theory built
around the explicated concept, we can't use it to make claims. And it's
hard to propose a theory without having it be the solution to one or more
problems.
An even more perplexing example of a hidden or missing problem is provided
by Popper himself in his later essays on World-3 where his style of exposition
is basically one of listing theses instead of solving problems. (For a persuasive
attempt to uncover the problems which World-3 might solve, see Irzik's forthcoming
"....".) Although Popper's own early philosophical writings about
science are themselves explicitly problem-centered, problems do not seem
to figure quite as importantly in his early accounts of science. When he
presents science as a series of conjecture and refutations, he points out
that the conjectures which we test are "tentative solutions to our
problems" (C & R, p. vii), but his primary concern there is to
describe how conjectures are given a preliminary appraisal in terms of their
empirical content -- other things being equal, we are to prefer universal
generalizations which make precise, testable assertions about a wide domain
of objects. The best tests of our conjectures are severe ones -- we are
to draw out the most outlandish, most improbable consequences of our conjecture
and check on their truth. If the prediction is true, our conjecture receives
a high post-test appraisal and is said to be corroborated. If the conjecture
fails the test, then it's back to the drawing board and we must dream up
another. It is the situation in which we have a refuted conjecture on our
hands that Popper characterizes as "the general problem situation in
which the scientist finds himself" (p. 241). He also mentions other
sources of contradictions, such as those arising within a theory or between
two different theories (p. 222) and lists theoretical difficulties such
as getting rid of ad hoc elements or uniting two disparate theories
(p. 241). He also refers to methodological problems, such as that of designing
severe tests (p. 222) or the Duhemian problem of deciding which component
of a falsified system should be rejected (p. 239).
Nevertheless, I think it is only fair to say that although there are certainly
hints of Popper's later views, the main emphasis in both the L.S.D. and
C. & R. is on the evaluation of conjectures and experimental tests,
not on the discussion of problems. There are probably two reasons for this.
First, as we saw above, Popper seems to have thought that the problem-situation
in any given science at any given time was generally pretty clear, so if
there was no problem about scientific problems (unlike the situation in
philosophy), perhaps he thought there was no need for him to talk about
them.
Secondly, Popper was evidently drawn into philosophy of science by the demarcation
problem; in particular he wondered what the relevant differences were between
Einstein's bold, speculative theory, on the one hand, and those of Freud
or Marx on the other. The answer which Popper arrived at had to do with
the falsifiability of theories and seemed to be independent of the nature
of the problems they were designed to solve. In fact, Popper himself stressed
that science, religion, myth, and metaphysics often dealt with the same
questions as did science (what are things made of, where did they come from,
what holds the earth up, why does the sun rise every morning, what is life?);
it is only the quality of their answers which are different. Here Popper
differed dramatically from positivists who would declare many such questions
to be pseudoproblems or meaningless by restricting the language of scientific
discourse to operational terms.
I have said that Popper's falsificationist methodology does not talk much
about problems, except for the one arising out of refuted conjectures or
violated common sense expectations, but that in his writing about the growth
of science he talks a lot about the role of metaphysical theories. For example,
in the L.S.D. he says that ideas may exist in an untestable form for a long
time before they make contact with experimental science. (pp. 278-9) As
examples, he lists atomism, the theory of terrestrial motion, the corpuscular
theory of light and the fluid theory of electricity (p. 278).
But this leads to a curious situation. Popper clearly thinks traditional
philosophical problems are important and can receive progressively better
answers. And he also says that metaphysical questions such as "How
can we understand change?" or "What is/are the ultimate elements
from which all others derive?" eventually receive scientific (i.e.,
testable) answers. Yet he stops just short of treating such questions as
genuine/typical scientific problems. Rather the stress is on the conflicts
in which observations play a crucial role and he is seemingly left in the
curious position of having to say that since atomic theory did not become
testable until the late 19th Century, the problem of atomism did not become
part of science until then! My intention is not primarily to saddle even
the early Popper with this position. Rather it is to support my claim that
at this time Popper did not articulate a detailed account of scientific
problems other than those arising out of empirical anomalies.
Let us now turn to what Popper says about scientific problems in his early
works on the philosophy of social science. Possibly as a reaction to the
well-entrenched tradition of unbridled speculations about human nature and
human societies, Popper here stresses the importance of practical and technological
problems. While admitting that he inclines towards the view that science
is most significant as a "spiritual adventure," he recognizes
the importance of practical problems and practical tests which serve both
as "a spur and as a bridle" (P. of H., p. 56). He quotes with
approval Kant's claim that "... it is wisdom that has the merit of
selecting, from among the innumerable problems which present themselves,
those whose solution is important to mankind" (p. 56).
He immediately applies this principle to philosophy itself, saying methodologists
should work on problems encountered by the practicing scientist instead
of floating off into "that atmosphere of futile subtlety which has
brought methodology into disrepute with the practical research worker"
(p. 57). Popper goes on to develop a methodology of "piece-meal engineering"
for the social sciences, which is intended to be both epistemologically
and morally preferable to utopian theorizing.
A detailed discussion of this methodology need not concern us here (but
see Irzik, BJPS). What is significant for the present project is Popper's
proposal that the inquiry be based on a different sort of problem than any
of those discussed so far, namely the problem of finding an appropriate
means for attaining some desired end. Instead of arising out of inconsistencies
within our knowledge, these practical problems arise out of a gap between
what we have and what we want. Popper adds some interesting fine structure
to his account, recommending that we design societies in such a way that
problems of removing pain take priority over those of insuring pleasure
(his doctrine of neg-utilitarianism) partly because it is so much easier
to agree on what is evil than on what is good (O.S., pp. 158-159). Again
the details need not concern us, but we should note that there are clearly
two components to his evaluation of problems, one methodological and one
moral.
In these writings he also introduces a methodological tool for the historian
-- that of "situational logic" combined with the Rationality Principle
(RP). Popper's RP says roughly that agents always act appropriately to the
situation in which they find themselves. The problem then for the historian
who would understand an agent's action is to discover exactly what that
situation was. This is a third type of scientific problem, one of building
testable models of "situations," i.e., complex initial conditions.
The law which "animates" the model is already known -- Popper
compares the role of the RP in situational logic to that of Newton's laws
in models of the solar system.
This account has many extremely problematic features some of which have
been discussed in the literature (see NK): Is the RP falsifiable? Is it
the agent's objective situation, his culturally defined situation, or the
situation as he idiosyncratically perceives it which is relevant? If the
latter, how testable can our models be? But the whole debate is predicated
on the assumption that the problem of modelling complex initial conditions
is a central one in some branches of inquiry; furthermore, it is also a
philosophically challenging one. I conclude that Popper, Hempel and other
"unitarians" are technically correct when they stress the similarities
in the form of explanations throughout the sciences. However, it is also
true, and equally important to stress, that the types of problems which
scientists struggle can vary enormously, ranging from the search for a unified
field theory all the way through to an investigation of the special environmental
conditions which led to the extinction of the dinosaurs or to the old chestnut
of why Napoleon invaded Russia. By focusing on problems, we better understand
the differences of intellectual style which mark different disciplines.
Let us now turn to what I am calling the late phase of Popper's philosophy,
namely the essays collected in Objective Knowledge. Now problems have moved
squarely onto center stage. (As a crude indicator, it's amusing to note
that the index of O.K. contains four inches of entries under the word "problem,"
compared with at most four lines in any of his preceding books. Of course,
this may partially reflect the tastes of the indexer!) It will be more convenient
to postpone detailed discussion of the presented views until the next chapter,
but let me hit some of the high points. Now Popper summarizes the growth
of science with the following scheme:
P1 ---> TS ---> EE ---> P2, where P1 is the problem we begin with,
TS is a tentative solution (or conjecture), EE stands for what may be a
multi-step process of error elimination (presumably through refutations
and then a replacement with refined conjectures) and eventually we then
move on to a new problem, P2.
This new mode of presentation is not inconsistent with any of his earlier
ideas, but the emphasis is dramatically different. In his earlier writings,
progress was most often characterized in terms of successive conjectures
and the extent to which they
appear to be getting nearer the truth. For example, in C & R, he lists
three requirements for saying that a new theory is superior to its predecessors:
it should be based on a simple, new unifying idea, it should not only explain
what the old theory explained but also have new empirical consequences,
and it should pass more tests than did the old theory (pp. 241-42). Now,
however, Popper suggests we measure progress by comparing P2 to P1 in terms
of the relative depth of the new problem.
Nowhere does Popper give an analysis of the concept of depth (which he also
applies to theories), but intuitively he may have something like the following
in mind: Whereas Kepler's problem was to find the regularities of planetary
motion, Newton's problem was to find the laws of all mechanical motions,
terrestrial and celestial. (We will explore this issue further when we discuss
Lakatos theory of problem-shifts.) Neither is it quite clear whether the
appraisal of progress based on problems supercedes, complements, or is equivalent
to the appraisal of conjectures. Suffice it to say, the schema certainly
dramatizes the fact that science both begins with problems and generates
new problems. Furthermore Popper claims that "the history of science
should be treated not as a history of theories, but as a history of problem
situations" (p. 177). He goes on to say that we can neither understand
a theory nor discuss it rationally without knowing the problems it was meant
to solve. (The same view is found in C & R, pp. 198-99.) This, too,
is a very curious position. I can well imagine an earlier Popper saying
that we must know what states of affairs a scientific theory rules out before
we can comprehend it or evaluate it, or that a philosophical theory must
be understood within the context of a debate, but his claim is explicitly
about scientific theories: "A theory is comprehensible and reasonable
only in its relation to a given problem situation, and it can be rationally
discussed only be discussing this relation." (C & R, p. 199) If
true, this thesis would make problems an integral aspect of finished science.
It would also have enormous implications for science education, ones which
differ significantly from Dewey's views. We will return to it when we look
at Hattiangadi's theory of the historical structure of problems and van
Fraassen's analysis of the context of requests for explanation.
In Objective Knowledge Popper continues his program of de-psychologizing
science. So-called "observation" statements certainly do not describe
sensory experiences and in sophisticated sciences they do not even involve
humans as instruments. Knowledge, he argues, is not a species of belief
and corroboration is not a measure of the degree of rational belief. Problems
exist whether or not we are puzzled by them. What is most important for
an understanding of science is the public problem situation, not the scientist's
private psychological motives or bewilderment. Thus to understand Galileo's
defense of his seriously flawed theory of the tides, we should not look
to psychological explanations in terms of irresistable fixations to the
archetypal idea of circular motion (Dilthey) nor an overwhelming desire
to prove Copernicus true and defeat the Pope at any cost (Koestler); rather
we should look at the problem he was trying to solve (namely to show that
the earth moved) within a certain theoretical framework of theory plus non-controversial
background assumptions.
Just as in his earlier discussions of situational logic Popper does not
make clear exactly how the lines between objective and subjective, or logical
and psychological, or agent and situation should be drawn. His intention
seems to be to focus on the external or public situation and to introduce
if necessary the contents of agents' cognitive states, but not their feelings
and desires -- at least this seems to be his program for understanding the
history of science. This attempt to divorce objective problem situations
from subjective conscious states reaches an extreme when Popper describes
biological evolution as a process of problem solving in which both endosomatic
organs and spider's webs are compared to theories (p. 145). He does, however,
leave one important function to be filled by subjective experience, namely
the role of emphasis in scientific inquiry, "the picking out of a problem
or a theory as important... or the dismissal of some theory as irrelevant...
the proposal that [it] ... be relegated to the `background' of the discussion."
(p. 167). Our task in this book, of course, is to see whether it is possible
to give reasons for these judgments of importance, relevance and salience.
This concludes our brief overview of Popper's major claims about problems.
His most frequent example of a scientific problem is that of explaining
a violated expectation (which means accounting for both the positive instances
of the refuted theory as well as the newly discovered counterexample), but
he also mentions the problems of unifying theories, making them less ad
hoc, removing inter-theoretic conflicts, as well as the more methodological
problems of designing severe tests and analyzing the relationship between
refuting evidence and the system as a whole. In the social sciences he stresses
both practical problems, e.g., the problem of finding an acceptable and
effective institutional means for achieving some societal end (such as relieving
pain), and the theoretical problem of analyzing or reconstructing the detailed
problem situations of agents' whose behavior we wish to explain. Throughout,
he treats problems as objective components of processes in the public domain.
In contrast to Dewey, private subjective feelings of puzzlement play a very
minor role in Popper's account.
Chapter II: Bibliography
Irzik, Gurol. .
Koertge, Noretta. .
Popper, Karl R. (1963). Conjectures and Refutations: The Growth of Scientific
Knowledge. London: Routledge and Kegan Paul.
Popper, Karl R. (1972). Objective Knowledge: An Evolutionary Approach. Oxford:
Oxford University Press.
Popper, Karl R. (1959). The Logic of Scientific Discovery. New York: Basic
Books, Inc.
Popper, Karl R. (1945). The Open Society and Its Enemies: Volume 1 - The
Spell of Plato. London: Routledge and Kegan Paul.
Popper, Karl R. (1964). The Poverty of Historicism. New York: Harper &
Row, Publishers.