The "EMPIRICIST CONSENSUS" in Philosophy of Science

In order to appreciate the great intellectual turmoil stirred up by the philosophical debates in which Thomas Kuhn's Structure of Scientific Revolutions was centrally involved, it is necessary to characterize the situation in philosophy of science prior to the 1960's when this debate began.  While there were of course sharp philosophical disagreements and controversies within philosophy of science throughout the first half of the twentieth century, there was -at least at a general level- also widespread agreement on certain basic assumptions about science and the philosophy of science.  This consensus began fall apart about mid-century and the consequences of this intellectual shift -which are still very much being played out today- have generated the debates with which this course is concenred.  Though it has no "official" name, for present purposes we will refer to this view as "the empiricisit consensus."
 

I. History and Constitution of the Empiricist Consensus:

The term "empiricist consensus" is an artificial label, which I have coined for teaching purposes, to designate a very general position with respect to scientific knowledge.  This label is approximately synonymous with the expression "the received view" of scientific theories, which appears frequently in the debates following the publication of Kuhn's SSR to designate what Kuhn is attacking.  It can also be called the "empiricist model" of scientific explanation.  These names refer to the view of science that tended to dominate in most (but not all) philosophy of science for roughly the first half of this century.  Insofar as those who held it tended to equate scientific knowledge with all genuine knowledge of the world, we may also say it was a very broad outlook on human knowledge in general, in other words it was a general epistemological point of view.

 It is important to stress that within this general consensus, there was considerable room for diversity among philosophers.  Thus one large group of philosophers within this consensus would be labeled "positivists," or more specifically "logical positivists" or "logical empiricists."  This group tended to dominate, but other philosophers, for example, Bertrand Russell, Karl Popper, and many American pragmatists could be placed within this consensus, but disagreed sharply with the positivists.  Thus in using this label I am creating an artificial "stereotype" which can be characterized in terms of a number of philosophical views, each one of which could be considered an element of this consensus.  All of the elements listed below does not characterize any one real philosopher.  However virtually all of those philosophers who clearly belong to this consensus would accept enough of these elements such that they all could be considered as bearing a very strong philosophical "family resemblance."

    Insofar as they tended to share certain assumptions, debates among philosophers within
the consensus could be considered "internal" to that consensus.  Such debates could take the
general assumptions of the consensus for granted and thus allowed philosophers of science to
focus on the specific points over which they disagreed.  Such issues tended to dominate in
philosophy of science until the sixties.

    There were always critics who stood outside the dominant consensus (e.g. Michael Polanyi,
Pierre Duhem, Alfred North Whitehead or Alexandre Koyre).  However, their disagreement
with the dominant consensus could be considered "external" to the main field of debate within
philosophy of science, and was thus largely ignored by those within the consensus.
 
 

    But ultimately disagreements and internal problems within the consensus weakened it to the point where the "dam broke" and the once dominant consensus became flooded with both internal and external problems.  This happened to coincide rather closely with the publication of Thomas Kuhn's Structure of Scientific Revolutions, which therefore often has been seen as a primary cause for the dissolution of the consensus.  It is certainly an exaggeration to regard Kuhn's book as having "caused" the break up of the empiricist consensus, but as events happened, its time of publication was such that debates in philosophy of science during the breakup of the consensus often took Kuhn's work as a point of departure.

II: The elements of the consensus:

    The question of authority challenges the philosopher of science to account for why
anyone ought to accept the scientific account of the world.  As a response to this question, in
the most broad sense the consensus is a position about what one ought to believe regarding
the world as experienced by human beings.  Since philosophers, virtually by definition of
"philosophy," will describe what one ought to believe as "rational," the empiricist consensus
is a view about what it is rational to believe with respect to the world we experience.
Thus one element of the consensus is that it is a theory of "scientific rationality," i.e. the reasoning
by which scientists come to formulate and accept -to "justify"-  the beliefs they hold, which comprise "scientific knowledge."
[If one rejects the view that what it is "rational to believe" is "what one ought to
believe," or in other words if one is what we might call an "irrationalist,"
then one is in effect rejecting a philosophical approach to belief formation
altogether, presumably in favor of some other method, say, flipping a coin,
inducing a "mystic" state, or believing everything consistent with some
          "authority," whether in human or written form.]
 
Because the empiricist consensus is a response to showing why one is rationally
justified in holding scientific beliefs, the consensus treats philosophy of science as exclusively
concerned with epistemological matters.  To say that a certain view is an "epistemological"
view is the philosopher's way of saying that it is concerned with the criteria, sources and
extent of human knowledge.  Another way of making this same point is to say that for
philosophers in the consensus, science is treated as purely a cognitive activity, i.e. a mode of
acquiring knowledge.

Of course members of the consensus realized that scientific knowledge
can be prized for a variety of different human reasons, not the least of which is the power
such knowledge may give us for improving or otherwise affecting human well-being.  But
that was not their concern; they tended to regard scientific knowledge as an end in-itself,
rather than for its utility (or disutility) in furthering other goals.  Scientists and philosophers,
could, and perhaps should, consider the uses to which scientific knowledge is or might be put,
but any such consideration was, so to speak, "extracurricular" to their function qua scientist,
or qua philosopher of science.
 

Thus another prominent element of the consensus was that its account of scientific rationality was one which anchored that rationality in purely cognitive goals; it is in its pursuit of knowledge, pure and unaffected by practical utilities, that the
authority of science is to be justified.
 
     Talking about "scientific rationality" as an account which justifies accepting scientific
"beliefs" can be misleading, for the term "beliefs" may refer to the private psychological
states of individual human beings, perhaps even allowing the possibility of "beliefs" which
cannot be expressed in language.


For those who held the empiricist consensus, "scientific beliefs" are restricted to what can in some sense be made public (and thus open to analysis) by being communicable in the meaningful expressions, or "statements" of some ideal or "artificial" language.

Insofar as a sentence in a "natural language" (e.g. English of German) can be "ambiguous" or can be given different meanings, we may be justified in believing it under one possible meaning and not justified in believing it under another; we would have in effect two different beliefs.  Thus to members of the consensus, those beliefs which one can be rationally justified in believing must be statements expressed by unambiguous statements of an specially constructed "ideal language" in which all meaningful expressions can be given precise meanings.

     The move away from psychological notions like "cognitive subject" and "holding a
belief" to linguistic notions such as "asserting (or denying) a statement" in an "ideal
language" is known as "the linguistic turn." This method of approach was characteristic of a
wide range of philosophy in the first part of this century, especially in areas affected by the
development of modern symbolic logic.  For those who made this turn, since the expression
"scientific knowledge" referred to what may be considered the entire body of rationally
justified beliefs about the world which people experience, the consensus endorsed the
"linguistic" element that "science" in effect refers to a very large group of "statements"
expressed in a special artificial "language" of precisely defined scientific terms.
 

   The statements of an ideal language can be divided into two kinds with respect to what makes them true or false:

Philosophers call statements whose truth is really a consequence of the meanings of their terms  "analytic statements," and distinguish them from those whose truth does not depend on their meaning, but rather depends on the facts of the world, which are called "synthetic statements."
 

Some philosophers, typically those called "rationalists," have held that it is possible to be justified in believing synthetic statements which are about the world, even though experience itself never provides adequate evidence for the truth of such statements.  Such philosophers argued that "Reason," unaided by any experience, could justify such statements.  Statements which can be thus justified independently of experience are called "a priori" and distinguished from those that can be justified only by appeal to experience, "a posteriori" or "empirical" statements.

"Rationalists" thus hold to the possibility of justifying believing "synthetic a priori"statements.

     Those who belong to the empiricist consensus commonly reject this possibility of synthetic a priori knowledge.  For them all justified statements which are not "analytic statements," i.e. those whose truth does not rest on the meanings of terms, but refers to "the world," are only justified by appeal to experience.  Expressed in the language of "scientific rationality," we may say that our rational warrant for holding scientific beliefs must always be a certain body of empirical evidence expressed in certain "observation statements" or "basic statements" which are considered to be justified or "directly verified" by extra-linguistic experience.

Exactly how the alleged "direct justification" was achieved was a matter of considerable disagreement among consensus philosophers, but there was widespread agreement that such basic statements form the foundation on which all of the cognitive authority of science rests.

 
 Thus another element of the consensus was its commitment to empiricism against the rationalist view that some things about the world could be known by pure reasoning.

[This does not mean that the empiricist consensus has no place for reason in determining scientific belief; far from it, as we shall see when we examine the hypothetico-deductivist model of justification which it came to endorse, the empiricist consensus strongly holds that scientific knowledge is made possible through deductive reasoning from both general laws and particular observation statements expressing "initial conditions."]

     A critical problem arises when we reflect on the fact that scientific knowledge most
obviously is not limited to what human beings have observed, in other words scientific beliefs
are not limited to what are often called "observation statements."  Scientific "explanations"
are often couched in terms of appeals to certain "laws of nature" which are universal
statements not about particular observations actually made by human beings but about all or
any possible observation which would be made under specified conditions.  Such "law-like"
statements may even be about what would be the case under conditions which in fact never
occur and possibly never could occur ("subjunctive conditionals" or "counterfactuals"). Thus
their authority cannot come directly from experience, but must appeal to some process of
reasoning from directly verifiable observation statements to such "law-like" universal
statements.  Just how such "reasoning" worked and what was the nature of the justification it
provided for the law-like statements was another considerable problem for those who worked
within the consensus.
 

     According to consensus, the essential importance of such law-like statements lies in
the fact that the aim of science is to provide explanations of particular observed phenomena
by showing that the occurrence of such phenomena can be validly deduced from a collection
of several statements of universal laws and other observation statements describing observable
"initial" or "antecedent" conditions.  When such a deduction is performed prior to the
occurrence of the phenomenon, it is called a "prediction," after the phenomenon has already
occurred, it is called an "explanation."
However the valid deduction of a single observable phenomenon -or even a very large number of phenomena- from a law does not establish the truth of the law. To say a deduction is "valid" means only that if those laws are true then that phenomenon would occur.  This is an uncontroversial point of formal, deductive logic: the deductive relationship is one-way. The truth of the universal law implies, given certain initial conditions, the truth of  the particual observation statements deduced from it; but the truth of any number of  particualr observation statements does not and cannot imply the truth of the universal law-like statements from which they can be deduced.  "Lower level" laws themselves can also be "explained" by deduction from "higher level" laws, but that, of course, does not establish the truth of those higher level laws.
This model of explanation by valid deduction from law-like statements is known as (from the Greek word "nomos" for "law") the "deductive-nomological model of explanation."  Whether or not all scientific explanations could be cast in this pattern was a bone of contention for many members of the consensus, but there was widespread agreement that this model of explanation did account for at least some forms of explanation provided by scientific knowledge.
     Many sciences are deeply committed to explaining observable phenomena through
appeal to general statements about entities, states, or processes, which are postulated to exist
but are not in fact observed, and perhaps not even in principle "observable."  Whether
honored with the label of "law of nature" or not, all such statements which transcend what is,
or could ever be, observed, are  called "theoretical statements."  Theoretical statements are
marked by the fact that they contain one or more conceptual terms appearing to refer to
entities, states, or processes which are not observed.  Obviously, then, no theoretical statement
can be directly justified by empirical means.
Again there was considerable disagreement as to the nature of whatever justification could be provided for theoretical statements, but insofar as it had to be different from the direct justification by sensory experience of pure observation statements, one of the core elements of theempiricist consensus is that it is always possible to distinguish between observational and theoretical statements.  Some members of the consensus may have admitted that the observational/theoretical distinction could not be sharply drawn, but all held it to be a real and
important distinction.
     The observational/theoretical distinction marks the difference between why we are
justified in believing the two kinds of statements.  Observation statements (also called "basic
statements" or "protocol Statements") are justified by a "direct" appeal to experience; they are
the statements of science which "reach outside" the body of language and connect scientific
beliefs directly to the world we experience.  However, theoretical statements are different; we
are not justified in believing them on such direct grounds.  These statements are treated as
"hypotheses" which we are justified in believing only on the basis of the "evidence" provided
by the "observation statements."  The observational/theoretical distinction thus marks the
distinction between evidence and that which the evidence "supports," what are usually called
"theories."  Thus theoretical statements are admitted into science only when they can, under
specific circumstances, be used to deduce directly verifiable "observation statements."  A
statement about unobservable entities or processes which could not be used to predict and
observable consequences is said to have no "empirical content" or "empirical significance"
and cannot form an acceptable statement in scientific knowledge.

     People -and textbooks- often talk about the laws of science as having been "proved."  The word
"prove" in philosophy and logic has a very strong meaning; it indicates that the truth of the
evidence, if it is true, establishes necessarily the truth of those statements which such
evidence is said to support.  Can the observational evidence ever prove the general law-like or
theoretical statements for which it is presented as the evidence?

    It has been widely admitted since the time of the great British empiricist, David Hume that in this strong sense no empirical evidence about particular observations can ever prove the truth of a universal statement which extends beyond what is already present in the empirical evidence. (This is often called "Hume's problem" or "the problem of induction.")  Holders of the empiricist consensus are all "descendants of Hume" in this respect.

     Traditional epistemology tended to regard the justification of true beliefs as
"knowledge" to require that such beliefs be proved to be necessarily true, or as was
commonly said, to be "certain." Under such a definition of knowledge, it would seem that
neither general laws nor theoretical statements could be known.

Therefore, philosophers in the consensus were led to relax the classical demand for certainty (necessarily true) and to hold instead that we are rational to believe general laws and theoretical statements which are "well-confirmed" or "corroborated" by the empirical evidence.  We can express this by saying that another element of the consensus is a commitment to "fallibilism," i.e., the view that no statement in science is ever immune to refutation by some as yet unknown evidence.
              Fallibilists reject the view that any theory is ever proved.  According to fallibilism any theory -even one which is well confirmed by present evidence-  may in the future be rejected on the basis of new evidence
     The relationship between empirical evidence and theory was the topic over which
defenders of the consensus spent most of their energies.  The aim of the philosophy of
science, so they argued, is to exhibit and analyze this relation, but they disagreed among
themselves radically as to the nature of this relation.  Originally the hope was that one could
"reduce" or translate law-like or theoretical statements into open ended sets of observation
statements, but gradually this ideal was seen as unattainable.  Later on, some, like the
positivists Carnap and Reichenbach, were most anxious to develop a "logic of confirmation"
according to which one could assess just how probably true any universal statement would be
in the light of any particular body of empirical evidence.  Others, like Popper, argued that
empirical evidence could serve only a negative role, "falsifying" those hypotheses we ought to
reject ("falsificationism").  According to this outlook what we ought to believe is those
hypotheses subject to rigorous empirical testing which have not been refuted.  Nevertheless,
both the positivist "verificationist" program and its arch rival within the dominant consensus,
the Popperian falsificationist program, both accepted some form of a "hypothetico-deductive
model" of the relation between evidence and theoretical statements.
     According to the "hypothetico-deductive model of justification" which came to be an accepted element of the empiricist consensus, we are justified in accepting such scientific laws and theories by proposing them hypothetically and then deducing from these laws and theories -together with statements of initial conditions established directly by observation-  certain other observation statements.
When observation reveals that a phenomenon occurs as in a way which such observation statements say they occur, then the law from which it is deduced is "confirmed," or "corroborated"; when observation reveals that phenomena do not occur as deduced, the law is "refuted" or "falsified."  Prior to any successful predictions, a theoretical statement is purely "hypothetical."  Only observation statements may be regarded as "directly verified."  Positivists came to accept that all hypotheses are only "confirmed" by the evidence. Falsificationists emphasized that attempts to refute the hypothesis have failed and spoke of the hypothesis as "corroborated."  After it has been confirmed by a wide range of successful predictions, and in the absence of any falsifying failed predictions, theoretical statements move from being of a purely "hypothetical" status to being rationally justified and are often called "laws of nature."  The difference between a "law" and a "theoretical hypothesis" (or "conjecture") is thus merely the degree of confirmation or corroboration.

     Many who belonged to the empiricist consensus held that, at least ideally, lower level
theoretical statements (whether regarded as mere "hypotheses" or as "laws") could be
explained by showing that they could be deduced from higher level more embracing
theoretical statements, until in an ideal perfected science, all phenomena could be seen as
deducible from set of relatively few fundamental "laws of nature."

The epistemic justification for such theoretical laws, then, is seen as based upon the "foundation"
of all actualobservation statements which confirm the predictions deduced from the theory; this
foundation"supports" or "rationally justifies" the theoretical structure of scientific laws which is erected
upon it.  Hence this view is said to be a form of "foundationalistic epistemology."  Another
element of the consensus was its commitment to a strong "empiricistic foundationalism."
     One could imagine that in an ideal case the body of rationally justified scientific
beliefs, a "total unified science," could be arranged as a hierarchy of statements with the most
general all-embracing fundamental laws of nature at the top, and lower level, more specific
laws being deduced from them under restricted conditions, in much the same fashion that the
theorems of geometry follow from its fundamental axioms and postulates.  However, such a
system of deductively interrelated theoretical statements, becomes "science" only because the
rational warrant for belief in these statements comes not from the "top down," i.e., from
axioms (laws) to theorems (as appears to be the case in geometry), but from the "bottom up,"
i.e., from the empirical observation statements which serve to confirm directly the lowest
level most particular statements deduced from the system of theoretical laws.  The result was
an image of science as a pyramidal hierarchy of statements, with observation statements
at the foundation and universal general laws at the top.
     According to the empiricist consensus what is normally taught as the "scientific
method" is in effect those procedures which warrant the rational acceptance of theoretical
laws, i.e., the process of deducing empirical consequences from theoretical hypotheses and
statements of observable conditions, and the confirmation or refutation of those hypotheses.
This procedure is seen as a process of "testing" hypotheses and was generally regarded as the primary occupation of scientists.
While the methods for devising the specific tests of individual hypotheses may vary from science to science, typically most defenders of the consensus held that all sciences share in common an allegiance to this method of testing general laws by empirical evidence as the only method for establishing the rational warrant for accepting theoretical laws and the scientific explanations and predictions deduced from them.

Thus defenders of the consensus accepted as another element a commitment to the "unity of science" with respect to its methodology.


     The hypothetico-deductivist account of justification leads philosophers to distinguish two distinct "contexts" in which scientific research might take place.  The first refers to the process by which the hypothesis is formulated. How do we first arrive at the hypotheses which, after suitable testing, come to be accepted as laws of nature?  How do scientists "discover" the hypotheses which are subsequently tested?
This aspect of scientific inquiry in which hypotheses are generated in the first place was called the "context of discovery."

For a hypothesis to become an accepted part of scientific belief, it must be justified. That which justifies a hypothesis, thereby turning it into an accepted scientific belief, is the process of testing by the hypothetico-deductive procedure indicated above.  This part of the scientists' activity was referred to as the "context of justification" and it was sharply distinguished from the process by which the scientist formulates the hypothesis in the firstplace, which was called the "context of discovery."

The important consequence of this distinction is that it is only the context of justification which makes a hypothesis scientific.  Thus only a logical analysis of the context of justification is relevant to the philosophical account of the rational warrant for scientific belief, i.e. answering the question of authority.

Generally members of the consensus tended to relegate the "context of discovery" to the personal, and often idiosyncratic psychological factors which characterized particular historical human scientists, accounts of which may be psychologically or historically interesting, but which are irrelevant to the philosophical account of how a hypothesis becomes a part of science, i.e. to scientific rationality.  The philosopher is interested in the "logic" by which a scientific belief comes to be warranted, not the
psychological or historical process which led to its formulation in the first place.

Thus the distinction between "context of justification" and "context of discovery" was also an essential
element of the empiricist consensus in philosophy of science.


     It seems that one uncontroversial aspect of science is that what today's scientists
believe differs markedly from the past, as the science of each period has differed from that of
its predecessors.  Scientific belief is not static, but continuously changing, perhaps often in
extensive or revolutionary ways. How can change in scientific belief be explained by the
consensus?

One possibility, of course, is that it can occur that laws which were formerly well
confirmed by repeated tests, in the future get refuted by new tests, thus changing the body of
empirical evidence, which may have been once favorable but becomes unfavorable as human
experience of the world grows.  This is one kind of change.
But often change was understood as a case of "reducing" more restrictive narrow hypotheses to broader more
general hypotheses.  According to the consensus it was ideally possible to reconstruct the history of any science showing that what were originally conceived as general laws turn out to be more restricted cases of yet more general laws.  (This was especially held in the case where Galileo's mechanics was seen as a restricted case of Newtonian mechanics and Newtonian mechanics subsequently seen as a restricted case of Einstein's relativistic mechanics.)

In this way, change in science could be explained in terms of progress moving
towards the ever more general all-embracing fundamental laws of nature encompassing an
ever greater range of observable phenomena, i.e. extending the height and increasing the base
of the pyramid of statements which is scientific knowledge.

Thus another element of the consensus was its belief that progress in science could be seen as "cumulative."
 
     If laws are regarded as "true," this progress could be seen as progress towards the truth
about the world we experience.  But have we any reason to hold that the fundamental laws
are true?  The word "truth" is generally taken to imply that what is true must be about
something which exists or is "real."  This definition of "truth" is called the "correspondence theory of truth."

According to the correspondence theory of truth, a true statement is true because it represents things as they really are, the "facts" correspond to what the law says, whereas the false statement is false because it fails to do so.  According to this notion of "truth" as referring to a representational relation between statement and reality, since all theoretical statements go beyond the purely
observable and make statements about unobservable processes or properties or objects, perfected scientific theories are really true descriptions of this "unobservable" world which is often said to "lie behind the observable phenomena."  This reality is often believed to stand in some causal relation to what is observed.  In short it was held by the consensus that if one holds that such statements are true in this correspondence sense, then one must hold that such unobservable entities are the real causes of observable phenomena.  This position is known as scientific realism. Some philosophers within the consensus were scientific realists in this sense.

     However, one could accept all the elements of the consensus and argue that only
observation statements are true or false; theoretical statements are neither true nor false, but
are accepted because of their use as "instruments" enabling the scientist to infer from one set
of observation statements, those stating the antecedent conditions, to another set of
observation statements, those describing the observed phenomenon predicted or explained.
This view, often called instrumentalism, is a specific form of "anti-realism," and was defended
by some members of the consensus.  Thus in one sense the consensus was neutral on realism
versus anti-realism, but in another sense, namely that realism demands a commitment to a
particular account of reality, i.e. to a metaphysics beyond what the consensus agreed upon,
and inasmuch as many members of the consensus disdained metaphysics, it is fair to regard
the empiricist consensus as anti-realist, at least in spirit, though many of its defenders would
have tried to find a way to retain a commitment to truth that allowed them to call themselves realists.

     One can naturally ask whether the consensus image of a pyramid of statements
justified by resting on observation statements describes actual science.  Perhaps surprisingly
defenders of the consensus freely admitted that it did not.  Their justification for this view
was that the philosopher of science was concerned with developing a "model" or account of
an ideal perfected science.  They fully well realized that actual real sciences always fell short,
often severely, of this ideal picture of what a science should be.  But this fact merely
indicates that real science is incomplete and part of a human, historical process; thus it is heir
to all the errors of which humans are capable.

     What, then, we may well ask, is the purpose of a philosophy of science which ignores
real science and concentrates on a perhaps unattainable ideal?

In response, the consensus held that he function of the philosopher of science is normative; the ideal picture of science painted by the philosopher serves as a standard by which one can assess the degree to which
any body of beliefs approaches what it should be to gain rational acceptability, i.e. to be "scientific knowledge."
At the same time this standard enables one to distinguish the bona fide true science from pseudo-science, and thus distinguishes what the scientifically rational person ought to believe from what he or she ought to reject.  The problem of drawing the boundary between science and pseudo-science was known as the "problem of demarcation" and was a central concern of many who held the empiricist consensus.