Content uploaded by Elliott Sober
Author content
All content in this area was uploaded by Elliott Sober on Nov 14, 2017
Content may be subject to copyright.
POPPER’S SHIFTING APPRAISAL
OF EVOLUTIONARY THEORY
Mehmet Elgin and Elliott Sober
Karl Popper argued in 1974 that evolutionary theory contains no testable laws and is
therefore a metaphysical research program. Four years later, he said that he had changed
his mind. Here we seek to understand Popper’s initial position and his subsequent re-
traction. We argue, contrary to Popper’s own assessment, that he did not change his
mind at all about the substance of his original claim. We also explore how Popper’s
views have ramifications for contemporary discussion of the nature of laws and the
structure of evolutionary theory.
1. Popper’s Evolving Interest in Evolution
Although our main focus is on Popper’s 1974 and 1978 papers, his interest in
evolutionary theory (ET) started considerably earlier, and a review of this his-
tory will be useful in locating Popper’s two papers within his larger intellectual
trajectory. Popper’s book The Poverty of Historicism was published in 1957, but
much of it (including the material on evolution) was given as lectures in the
1930s, and the book was first published in three parts in Economica,in1944–
45. Although Popper does not assert in The Poverty of Historicism that ET con-
tains no empirical laws, and this is the central issue in his 1974 and 1978 papers,
he advances philosophical theses about ET in that earlier book that helped shape
his subsequent thought. In fact, he there defends what seems like a logically stron-
ger thesis—that there are no laws of evolution, period.
Contact Mehmet Elgin at Muğla SıtkıKoçman Üniversitesi, Edebiyat Fakültesi, Felsefe Bölümü,
48170 Kötekli, Muğla, Turkey (melgin70@gmail.com). Contact Elliott Sober at the Philosophy De-
partment, University of Wisconsin–Madison, 53706 (ersober@wisc.edu).
We thank Dylan Beschoner, Richmond Campbell, Jonathan Hodge, Gregory Mayer, Alan Sidelle,
and the anonymous referees of this journal for useful comments. Mehmet Elgin acknowledges the sup-
port of TÜBİTAK BİDEB 2219. Elliott Sober acknowledges the support of the William F. Vilas Trust.
HOPOS: The Journal of the International Society for the History of Philosophy of Science, vol. 7 (Spring 2017).
2152-5188/2017/0701-0001$10.00. © 2017 by the International Society for the History of Philosophy of Science.
All rights reserved. Electronically published February 16, 2017.
31
In section 27 of The Poverty of Historicism (called “Is There a Law of Evo-
lution? Laws and Trends”), Popper argues that there are no general laws gov-
erning the evolution of societies, the evolution of organisms, or the evolution
of the whole universe. He claims this despite granting that many laws are used
in explaining the phenomena discussed in the social sciences, in evolutionary
biology, and in cosmology:
The crucial point is this: although we may assume that any actual suc-
cession of phenomena proceeds according to the laws of nature, it is im-
portant to realize that practically no sequence of, say, three or more causally
connected concrete events proceeds according to any single law of nature.If
the wind shakes a tree and Newton’s apple falls to the ground, nobody
will deny that these events can be described in terms of causal laws.
But there is no single law, such as that of gravity, nor even a single def-
inite set of laws, to describe the actual or concrete succession of causally
connected events. ...The idea that any concrete sequence or succession
of events (apart from such examples as the movement of a pendulum or a
solar system) can be described or explained by any one law, or by any one
definite set of laws, is simply mistaken. There are neither laws of succes-
sion nor laws of evolution. (Popper 1957, 117)
In the same section of the book, Popper faults Comte and Mill for thinking
that there are laws of succession, each of which describes the sequence of states
through which a system of given type must move. Popper admits that there are
historical trends, but he says that trends cannot be characterized by universal
generalizations and so are not laws.
We have a few criticisms of these claims from The Poverty of Historicism.
Popper shifts between saying that there is no “single”law of evolution and say-
ing that there is no “one set of definite laws.”The first of these requires a prin-
ciple for individuating laws. If L
1
,L
2
,...,L
n
are each laws, why is their con-
junction not a single law? And conversely, a sentence that seems to express a
single law can easily be rewritten as a conjunction, each conjunct of which ex-
presses a law. For this reason, Popper’s point is better put by using the second
formulation that “there is no one set of definite laws.”In any event, we do not
think that Popper gives a good reason for accepting this thesis. He grants that
there are laws of change for pendulums and solar systems but never makes it
clear why there are not laws of evolution for other objects. We also see a prob-
lem in Popper’s expressing his thesis that there are no laws of evolution by as-
suming that laws must be deterministic. Why can a law not be probabilistic?
Popper is right to separate laws from mere trends. However, even if statements
HOPOS |Popper’s Shifting Appraisal of Evolutionary Theory
32
like “events of type T
1
are often followed by events of type T
2
”are not law-like,
this leaves it open that probability statements sometimes are. After all, there are
objective interpretations of probability statements that do not equate probabil-
ities with actual frequencies. This is a point that Popper later took to heart when
he proposed a propensity interpretation of probability (Popper 1959b). Quan-
tum theory does more than describe de facto trends. Trends depend on initial
conditions, as Popper says, but this does not rule out the possibility that there
are probabilistic laws that are independent of initial conditions.
Popper’s interest in ET broadened in the 1960s (Simkin 1993; Watkins 1995;
Munz 2006). There were three reasons why. Popper came to believe that there is a
close relationship between his method of trial-and-error elimination in science
and the process of natural selection (Ruse 1977; Nanay 2011). He also saw
new ways in which ET is relevant to his views on historicism. And finally, he be-
came convinced that ET is the only significant scientific theory that bears on the
mind-body problem and on the problem of freedom in a physical universe.
1
As Popper delved deeper into ET, he came to believe that the principle of
natural selection (NS) is a priori. This was a conclusion that puzzled him. On
the one hand, he maintained that scientific theories are empirical, not a priori,
but on the other hand, he was convinced that ET is the most successful theory
in its domain and has no serious competitors. Popper (1974, 1978) attempted
to reconcile these ideas, although he reached different conclusions in the two
papers. His solution in the 1974 paper was to suggest that ET is not a theory
of forces like Newtonian theory but rather involves what he calls “situational
logic.”This formed part of his thesis that ET provides in-principle explanations
without supplying actual explanations of particular phenomena, since actual
explanations of a given event must cite empirical laws. In the 1978 paper, Pop-
per additionally suggested that there is an alternative formulation of the prin-
ciple of NS that is not a priori. The 1974 a priori principle and the 1978 em-
pirical reformulation are these:
(PNS) If traits T
1
and T
2
are found in a population, T
1
is fitter than T
2
,
the two traits are heritable, and no other evolutionary causes impinge on
the population, then T
1
will increase in frequency.
2
1. Clear statements of his interest in ETregarding the relationship between ET and his evolutionary
epistemology can be found in Popper (1957, 133; 1959a, 91; 1972, 261; 1974, 133; 1994a, 12). For
Popper’s view about how ET bears on the mind-body problem and on freedom in a physical universe,
see Popper (1972, 1978, 1994a).
2. Popper (1974) does not explicitly indicate which statement about NS he thinks is a priori, but he
does say that given the statistical definition of fitness, the principle he has in mind is a priori. This is our
justification for saying that Popper thinks that PNS is a priori.
Elgin and Sober |SPRING 2017
33
(PNSR) All traits in all organisms evolved because of natural selection.
3
Popper (1978) says that the distinguished evolutionary biologists C. H. Wad-
dington, G. G. Simpson, J. B. S. Haldane, and R. A. Fisher all attribute great
explanatory power to PNS. Popper replies to this pantheon by pointing out
that PNS is a tautology and that the explanatory power of a tautology is zero.
PNS’s tautological status stems from the fact an organism’sfitness is defined as
its degree of actual reproductive success. By proposing the PNSR reformula-
tion, Popper hoped to increase the empirical content of ET by turning its main
principle into an empirically testable statement. As we explain in what follows,
Popper (1978) thought that PNSR is false. Given this, it is questionable how
much ET is improved by replacing the trivial claim PNS with the false claim
PNSR.
2. Difficulties concerning A Priori Formulations of ET
In his 1961 Herbert Spencer Lecture (which appeared in print as Popper
[1972]), Popper says that it is a mistake to compare ET to Newtonian physics:
“Darwin’s discovery of the theory of natural selection has often been compared
to Newton’s discovery of the theory of gravitation. This is a mistake. Newton
formulated a set of universal laws intended to describe the interaction, and
consequent behavior, of the physical universe. Darwin’s theory of evolution
proposed no such universal laws”(267). Popper claims that Darwinism’s suc-
cess resides in its providing in-principle explanations: “Nevertheless, Darwin’s
revolutionary influence upon our picture of the world around us was at least as
great, though not as deep, as Newton’s. For Darwin’s theory of natural selection
showed that it is in principle possible to reduce teleology to causation by explaining,
in purely physical terms, the existence of design and purpose in the world ”(267).
4
Two paragraphs later, Popper says: “Although this was a great achievement, we
have to add that the phrase in principle is a very important restriction. Neither
Darwin nor any Darwinian has so far given an actual causal explanation of the
adaptive evolution of any single organism or any single organ. All that has been
shown ...is that such explanations might exist (that is to say, they are not log-
ically impossible).”Thus, the reason Popper thinks it is a mistake to liken ET
to Newtonian theory is that ET does not postulate empirical laws; it offers a
3. This is how we interpret Popper’s (1978, 345) claim that “all organisms and especially all those
highly complex organs whose existence might be interpreted as evidence of design and, in addition, all
forms of animal behavior have evolved as the result of natural selection.”
4. In this quotation, the use of italics is Popper’s. The same holds for italics in subsequent quota-
tions, unless otherwise noted.
HOPOS |Popper’s Shifting Appraisal of Evolutionary Theory
34
priori mathematical models, which at best can show how it is possible in prin-
ciple to explain adaptations in causal terms.
How, according to Popper, did evolutionary biology lapse into tautology? In
his 1961 Herbert Spencer Lecture, he alludes to the source of this problem:
“What I regard as Darwin’scentralidea—his attempt to explain genetic changes
which led to better adaptation in the sense of better chances for the individual an-
imal or plant to survive—has recently suffered an eclipse. This is due very largely
to the fashionable pursuit of mathematical exactness, and to the attempt to de-
fine survival value statistically, in terms of actual survival (of a gene, or some
other genetic unit, in a population)”(Popper 1972, 271). In the next two par-
agraphs, Popper argues that such mathematically exact theories cannot explain
complex adaptations. Popper addresses this topic in his “Science: Problems,
Aims, Responsibilities,”initially presented at the Plenary Session of the 47th An-
nual Meeting of the Federation of American Societies for Experimental Biol-
ogy and published in the Federation Proceedings in 1963 but subsequently re-
published as chapter 4 of his The Myth of the Framework:“There is a difficulty
with Darwinism. While Lamarckism appears to be not only refutable but ac-
tually refuted (because the kind of acquired adaptations which Lamarck envis-
aged do not appear to be hereditary), it is far from clear what we should con-
sider a possible refutation of the theory of natural selection. If, more especially,
we accept the statistical definition of fitness which defines fitness by actual sur-
vival, then the theory of the survival of the fittest becomes tautological, and
irrefutable”(1994b, 94).
Although Popper talks about the difficulty of formulating empirical laws
about NS that use a statistical definition of fitness in his 1961 Herbert Spencer
Lecture, it is only in his 1963 talk that he claims that this way of describing
NS is tautological. However, Popper does not give a full treatment of the is-
sue in that talk. He just adds to the above quoted remarks that we should avoid
the mistake of thinking that NS is the only mechanism that can lead to appar-
ently goal-directed adaptations. The way forward, he suggests, is to conceptu-
alize alternative mechanisms and then design crucial experiments to decide be-
tween them and NS. We note that this is and has been common practice in
evolutionary biology, where drift is often taken to be an important competitor
to NS.
The issue of tautology reappears in Popper’s 1965 Arthur Compton Memo-
rial Lecture: “Quite apart from evolutionary philosophies, the trouble about
evolutionary theory is its tautological, or almost tautological character: the dif-
ficulty is that Darwinism and natural selection, though extremely important,
explain evolution by ‘the survival of the fittest’(a term due to Herbert Spen-
cer). Yet there does not seem to be much difference, if any, between the asser-
Elgin and Sober |SPRING 2017
35
tion ‘those that survive are the fittest’and the tautology ‘those that survive are
those that survive’” (Popper 1972, 241–42). Popper raises the same issue in his
1969 Emory University Lectures:
Biologists (especially Fisher) felt compelled to define as “more fit”those
which more often survive. Thus, what once looked like a promising ex-
planatory theory becomes quite empty. The statement “Evolution tends
to produce higher forms because only the fittest survive”may sound like
an explanation. But if we substitute here for “the fittest”its defining phrase,
we get: “Evolution tends to produce higher forms because those forms
which more often survive more often survive.”So our “because”phrase
has degenerated into a tautology. But tautology cannot explain anything.
All tautologies are equivalent to “All tables are tables”or “Those who live long
are those who live long.”(Popper 1994a, 54, emphasis added)
5
The tautology issue comes up again in chapter 2 of Popper’sObjective Knowl-
edge, but this time he goes a little further in his treatment of the subject: “A
central problem of evolutionary theory is the following: according to this the-
ory, animals which are not well adapted to their changing environment per-
ish; consequently those which survive (up to a certain moment) must be well
adapted. This formula is little short of tautological, because ‘for the moment
well adapted’means much the same as ‘has those qualities which made it sur-
vive so far.’In other words, a considerable part of Darwinism is not of the na-
ture of an empirical theory, but it is a logical truism”(Popper 1972, 69).
After this paragraph, Popper attempts to separate, to our knowledge for the
first time, the empirical and the a priori parts of the theory. He says that the
empirical claims are these: the environmental conditions in which organisms
live change slowly, organisms are sensitive to such changes, the only way organ-
isms can survive such changes is by producing mutations some of which re-
spond to these changes, and useful mutations sometimes occur. Popper then
describes what he thinks the a priori part of this theory is: “If the process of
adjustment has gone on long enough, then the speed, finesse, and complexity
of the adjustment may strike us as miraculous. And yet, the method of trial and
of the elimination of errors, which leads to all this, can be said not to be an
empirical method but to belong to the logic of the situation. This, I think, ex-
plains (perhaps a little too briefly) the logical or a priori components in Dar-
winism”(1972, 70). Popper is here constructing a conditional statement whose
5. We see things differently; although all mathematical truths are mathematically equivalent, some
help explain empirical phenomena while others do not (Sober 1984, 2011a).
HOPOS |Popper’s Shifting Appraisal of Evolutionary Theory
36
antecedent describes properties of organisms and environment and whose con-
sequent is about the emergence of novel adaptations. The conditional, he says,
is a priori, although the antecedent and the consequent are both empirical.
Logically speaking, Popper’s conditional resembles “if someone is a bachelor,
then he is unmarried.”Popper does not take the discussion any further in this
work.
Popper’sfirst thorough treatment of the tautology issue and ET was in sec-
tion 37 of his 1974 autobiography (entitled “Darwinism as a Metaphysical Re-
search Program”), which appeared in the Schilpp volume on Popper. Here Pop-
per announces that ET “is not a testable scientific theory, but a metaphysical
research program—a possible framework for testable scientific theories.”He
clarifies what he means by a metaphysical research program in footnote 242:
“The term ‘metaphysical research programme’was in my lectures from about
1949 on, if not earlier; but it did not get into print until 1958, though clearly
in evidence in the last chapter of the Postscript (in galley proofs since 1957). I
made the Postscript available to my colleagues, and Professor Lakatos acknowl-
edges that what he calls ‘scientific research programmes’are in the tradition of
what I described as ‘metaphysical research programmes’(‘metaphysical’be-
cause nonfalsifiable)”(Popper 1974, 175).
This remark requires clarification. Although Popper thought that ET is a
“metaphysical research program,”he did not think that Newton’s theory or
Einstein’s theory falls in this category. Lakatos, in contrast, held that all three
are “scientific research programs”because all three contain hard-core assump-
tions that are unfalsifiable because of the Duhem-Quine problem. For this rea-
son, Lakatos proposed that such programs should be evaluated by their overall
ability to solve theoretical and empirical problems. The latter part of what
Lakatos says in fact coincides with what Popper says about metaphysical re-
search programs. Popper decides whether a theory is metaphysical on the basis
of whether its purported laws are empirical (Stamos 2007); he does not deny
that genuine physical theories may also deploy metaphysical assumptions. It is
in this connection that Popper sees a difference between ET on the one hand
and Newton’s and Einstein’
s theories on the other.
It is important to recognize that just because Popper labeled ET as meta-
physical, we should not jump to the conclusion that he took ET to be a pseu-
doscience (Hull 1999). For Popper (1978, 344), “the doctrine of natural selec-
tion is a most successful metaphysical research programme.”Psychoanalysis
and astrology are not in that league; Popper thinks they are pseudosciences.
Popper then explains Darwinism’s success as a metaphysical research program
in terms of “situational logic”:“I think that there is more to say for Darwinism
than that it is just one metaphysical research programme among others. In-
Elgin and Sober |SPRING 2017
37
deed, its close resemblance to situational logic may account for its great success,
in spite of the almost tautological character inherent in the Darwinian formu-
lation of it, and for the fact that so far no serious competitor has come forward”
(1974, 135). The a priori character of ET’s main principle does not lead Pop-
per to dismiss the theory; rather, he tries to provide a philosophical explanation
of how ET could be scientifically successful.
Popper then clearly states that by Darwinian theory or Darwinism he is re-
ferring to the modern synthesis theory of evolution. He says that this theory
involves two main theses:
(1) The great variety of the forms of life on earth originate from very few
forms, perhaps even from a single organism: there is an evolutionary
tree, an evolutionary history.
(2) There is an evolutionary theory which explains this. It consists in the
main of the following hypotheses. (Popper 1974, 135)
Popper then lists four hypotheses that flesh out this second component. The
first says that heredity is faithful. The second says that small variations arise
in populations caused by accidental mutations. The third says that NS is a
mechanism that eliminates unfit variants. The fourth says that the variability
(what Popper calls the ‘scope of variation’) found in a population is subject
to NS. This last hypothesis basically says that the hereditary mechanisms that
give rise both to mutations and to mostly faithful reproduction are themselves
the products of NS. It is in this sense that NS is said to determine the scope of
variation.
It should be clear that the two main theses of ET are not tautologies, and
neither are the first, second, and fourth hypotheses that Popper mentions. Pop-
per thought that unrestricted existential claims are not falsifiable and therefore
not empirically testable; however, he also thought that existential statements
are falsifiable if they are restricted to specific space-time regions (Popper 1963,
195–96).
6
This is how common ancestry and the three hypotheses should be
understood—they each pertain to life on planet Earth.
7
It follows that there
are important parts of ET that are empirically testable. The only element that
is a candidate for being a tautology is the third hypothesis. If this analysis of
what Popper says is correct, then the only reason Popper has for thinking that
ET is not a testable scientific theory is that the third hypothesis is a tautology.
6. Others do not require much in the way of background theories; “there are five apples in the bas-
ket now”can be refuted just by counting them apples.
7. The idea that common ancestry is a historical hypothesis, not a general law, is already in The
Poverty of Historicism; see Popper (1957, 106–7).
HOPOS |Popper’s Shifting Appraisal of Evolutionary Theory
38
The problem, he thinks, is engendered by the definition of fitness that biolo-
gists use: “Adaptation or fitness is defined by modern evolutionists as survival
value, and can be measured by actual success in survival: there is hardly any
possibility of testing a theory as feeble as this”(Popper 1974, 137).
Although Popper speaks of the “synthetic theory of evolution,”he evidently
is not thinking of all the propositions in that “theory”when he says that the
theory is feeble or tautologous. We suggest that when he here considers what
“evolutionary theory”is, he is looking for laws, so all the singular empirical
statements in the “theory”are set to one side. What remains, he says, is a lonely
tautology. We emphasize that the fact that PNS is a tautology does not show
that there are no laws in ET. Otherwise, Newton’s theory would fall by the
wayside if “F5ma”turned out to be a definition of force. Popper thinks that
PNS is the main principle of ETand the only candidate law that ET offers. For
Popper, PNS is the common thread of ET that connects all the other evolu-
tionary claims he lists; if it fails to be empirical, what we are left with is a bunch
of unconnected claims.
As already noted, Popper thinks that the way to understand how ET can be
successful despite its lacking empirical laws is by understanding its deployment
of situational logic. To get a grip on what Popper has in mind here, consider
the following remark: “Models, as here understood, may be called ‘theories,’
or be said to incorporate theories, since they are attempts to solve problems—
problems of explanation. But the opposite is far from true. Not all theories are
models. Models represent typical initial conditions rather than universal laws.
And they therefore need to be supplemented by ‘animating’universal laws of in-
teraction—by theories which are not modelsin the sense here indicated”(Popper
1994b, 165).
8
Popper thinks that in the natural sciences there are two kinds of
explananda—token events and types of events. Token events are explained or pre-
dicted by the conjunction of universal laws and initial conditions. However, types
of events are explained by constructing models that “represent typical initial con-
ditions,”which are always idealizations (164). Popper’s example of a token event
is the occurrence of a particular lunar eclipse, and his example of a type of event is
the periodic recurrence of lunar eclipses. The idea is that if you want to either
explain or predict a single occurrence of a lunar eclipse, you just write down some
laws and relevant initial conditions and then show that the event in question fol-
lows from them. However, in explaining why lunar eclipses have a periodic cycle,
you need to build a model representing typical initial conditions and show that
under these typical initial conditions lunar eclipses periodically recur. In the sec-
8. This quotation is from The Myth of the Framework. The chapter is based on a talk Popper gave in
1963.
Elgin and Sober |SPRING 2017
39
ond case, laws enter into explanations only indirectly in the sense that the models
are built or constructed in accordance with those laws.
Although Popper says in the passage previously quoted that a model-based
explanation must always have universal laws in the background, he says a page
later that model-based explanations can go forward in the social sciences even
though they are not backed up by universal laws: “I wish to propose the thesis
that what I have said about the significance of models in the natural sciences
also holds for models in the social sciences. In fact, models are even more im-
portant here because the Newtonian method of explaining and predicting sin-
gular events by universal laws and initial conditions is hardly ever applicable in
the theoretical social sciences. They operate almost always by the method of
constructing typical situations or conditions—that is, by the method of con-
structing models”(1994b, 166). Constructing a model in the social sciences
proceeds by describing a typical situation, the problem in that situation, the
parts of the typical situation that are relevant to the problem at hand, and
the aims of an agent. Popper thinks that these models do not require the spec-
ification of any psychological laws. All you need to assume, according to Pop-
per, is a so-called rationality principle. This principle says that agents act in ac-
cordance with the requirements of the situation for achieving their aims.
Although Popper seems to think that the only candidate for a law in the so-
cial sciences that resembles Newtonian laws is the rationality principle, he says
that this principle is not a law; he thinks it is almost empty. This is because he
holds that the rationality principle is just an assumption about how a typical
agent would behave in a given typical situation, and in this sense it does not
necessitate an outcome in the way that Newton’s laws together with initial con-
ditions have entailments about Uranus’s orbit.
Right after he asserts that there are no testable laws in ET, Popper (1974,
135) says that Darwinism is an application of situational logic. His point is this:
First you describe a population of objects that is somewhat stable, but not ex-
actly, and that contains small variations where some of those variations help
their possessors to better cope with the conditions in which they find them-
selves. Then you “add to this the assumption of the existence of a special frame-
work—a set of perhaps rare and highly individual conditions—in which there
can be life, or more especially, self-reproducing but nevertheless variable bodies.
Then a situation is given in which the idea of trial and error-elimination, or of
Darwinism, becomes not merely applicable, but almost logically necessary”
(134). Here again, Popper is formulating a conditional in which the assump-
tions of the model are the antecedent and statements about resulting adapta-
tions are the consequent; the whole conditional is a priori. In accordance with
this general framework, you can construct specific models for a specific situa-
HOPOS |Popper’s Shifting Appraisal of Evolutionary Theory
40
tion; the specific case is an instance of a general type of event. The models, then,
are testable but not the framework itself. All you can say about this framework is
how often it is realized in nature.
9
In contrast, Newtonian theory as a whole is
empirically testable and so are its applications in specific models.
For Popper, what holds things together in a model in the social sciences is
the so-called rationality principle, while in ET what holds things together is
PNS. However, in the case of the social sciences, the rationality principle is ob-
viously false if you interpret it as an empirical claim; therefore, this principle is
better interpreted as a regulative principle in the construction of models.
10
In
the case of ET, the situation is different, in that the PNS “law”is true a priori.
Even so, there is something that ET and the social sciences have in common;
for Popper, the point of building models in both is not to test these general
“laws”but to test singular models constructed in accordance with those “laws.”
Popper’s demand that a genuine scientific theory must include empirical
laws shaped his view of what a scientific explanation is. For him, a causal ex-
planation of a token event must cite at least one empirical law: “To give a causal
explanation of an event means to deduce a statement which describes it, using
as premises of the deduction one or more universal laws, together with certain
singular statements, the initial conditions”(Popper 1959a, 38). Popper was
therefore driven to conclude, not just that ET contains no empirical laws but
that it cannot provide causal explanations of the adaptive features that organ-
isms actually possess.
It is interesting that Popper thinks that a model-based explanation is about
the “typical situations”that a kind of system occupies. Does this mean that the
system usually obeys the model’s postulates? If so, Popper’s usage of “model”is
at variance with what scientists and philosophers now often mean by that term.
They have in mind the idea that models contain idealizations that render them
false. Far being true of what a system is like in the kinds of situations it usually
occupies, the model is true only in situations that the system never occupies.
3. A Change of Heart or the Old View in Disguise?
We now return to Popper’s famous talk delivered in 1977 at Darwin College,
Cambridge, which was published in Dialectica in 1978. Although Popper says
9. Brandon (1990) endorses this position.
10. Popper (1994b, 169) says that the rationality principle is a consequence of a methodological
rule: “The adoption of the rationality principle can therefore be regarded as a by-product of a method-
ological postulate. It does not play the role of an empirical postulate.”PNS in ET, however, cannot be
interpreted as a methodological or regulative rule since it is an a priori truth. Yet, according to Popper,
their functions in the construction of models are the same.
Elgin and Sober |SPRING 2017
41
that he has changed his mind about the absence of testable laws in ET, we will
argue that Popper in fact did not change his mind at all (see also Hull 1999,
485). He begins by describing an evolutionary claim that is empirical: “In its most
daring and sweeping form, the theory of natural selection would assert that all
organisms, and especially all those highly complex organs whose existence might
be interpreted as evidence of design and, in addition, all forms of animal be-
haviour, have evolved as the result of natural selection. ...If formulated in this
sweeping way, the theory is not only refutable but actually refuted”(Popper
1978, 345). Popper mentions two reasons for thinking that this PNSR formu-
lation is falsifiable—sexual selection and drift. To discuss sexual selection, Pop-
per uses Darwin’s example of the peacock’s tail. Peacocks evolved gaudy tails
because peahens preferred to mate with gaudy males. Popper sees this as a coun-
terexample to PNSR because he accepts Darwin’s assumption that sexual se-
lection is not a species of NS. Many biologists would now disagree. They would
say that when males evolve gaudy tails in response to the mating preferences of
females, this is just like polar bears evolving thicker fur in response to the weath-
er’s getting colder. Sexual selection is a special kind of NS wherein the selective
environment experienced by one sex is created by properties of the other. As for
Popper’s idea that drift is a counterexample to PNSR, he does not mention any
actual examples of drift; rather, he notes that an organ may look adaptive and
that this may lead us to mistakenly conclude that it was produced by NS. He
says this can be a mistake because it is possible for an organ to be a product
of drift and become adaptive only later. However, the mere possibility of this sce-
nario is not enough to show that PNSR is false; one needs an actual case. Given
the assumption that there are such cases, Popper can conclude that some organs
and behaviors have evolved without NS’s being the cause.
Popper’s point about PNSR is fair enough; it is falsifiable. However, that
does not show that PNS is empirical or that ET contains an empirical law. Pop-
per has not changed his mind at all; he has merely changed the subject, from
PNS to PNSR. This is also evident from another remark in the 1977 Darwin
lecture: “The fact that the theory of natural selection is difficult to test has led
some people, anti-Darwinists and even some great Darwinists, to claim that it
is a tautology. ...I mention this because I too belong among the culprits. ...
My solution was that the doctrine of natural selection is a most successful
metaphysical research programme. ...I still believe that natural selection works
in this way as a research programme. Nevertheless I have changed my mind
about the testability and the logical status of the theory of natural selection”
(Popper 1978, 344–45). When Popper says that “the theory of natural selec-
tion”is testable, he presumably has PNSR in mind. He grants that this principle
is not true. If laws must be true, then PNSR is not a law. But is it law-like? That
HOPOS |Popper’s Shifting Appraisal of Evolutionary Theory
42
is, if it were true, would it be a law? Maybe not; perhaps PNSR resembles the
claim that all cancers are caused by cigarette smoke. If this statement were true,
would it be a law? We think not. It could be an accidental generalization. It
violates no law of nature to suppose that some cancers have other causes. As
noted, Popper thinks there actually are other causes of trait evolution apart from
NS. But what is actual is also nomologically possible. This suggests that even if
PNSR were true, it would not be a law.
Although Popper (1974) is grappling with a genuine philosophical puzzle
concerning the conceptual structure of ET, Popper (1978) achieves something
that is rather trivial; he simply constructs a statement about NS that is testable
and (let us grant) false. Similar constructions can be carried out in phrenology
and psychoanalysis, but Popper would never dream of viewing those construc-
tions as vindicating those two pseudosciences. Popper steadfastly maintained
that a scientific theory must contain one or more empirical laws; his picture
of ET, both in 1974 and in 1978, is that the theory lacks this crucial element.
The thought occurs to us that Popper may have seen that point quite clearly.
4. Evaluating Previous Interpretations of Popper’s Views
about Laws in ET
Hull (1999), citing Popper (1978), says that Popper was misled by Fisher, Hal-
dane, Simpson, and other evolutionary biologists. Hull claims that the defini-
tion of fitness as actual survival is an idealization introduced to make the math-
ematics work well; for this reason, it should not be considered an essential part
of ET. Hull further submits that biologists are aware that such idealizations are
needed for the theory to be operationalized. We do not think that this reply to
Popper is successful. For one thing, the mathematics of ET does not require
that one equate an organism’sfitness with its actual degree of reproductive suc-
cess. This is something that the propensity interpretation of fitness (Brandon
1978; Mills and Beatty 1979) made clear. Second, “operationalizing”a theory
that contains probabilistic parameters requires estimates of those parameters’
values; it does not require that one define probability to mean actual frequency.
And third, even if we reject the statistical definition of fitness, the question re-
mains of what should replace it. Defenders of the propensity interpretation of
fitness have suggested the following replacement:
(PNSP) If traits T
1
and T
2
are found in a population, T
1
is fitter than T
2
,
the two traits are heritable, and no other evolutionary causes impinge on
the population, then T
1
will probably increase in frequency.
Elgin and Sober |SPRING 2017
43
PNSP is just like PNS, except that “probably”has been inserted. Popper’s prob-
lem remains, since PNSP appears to be a priori, given the definition of fitness
as a probabilistic propensity.
11
Nanay (2011) claims that Popper changed his account of the scientific pro-
cess of conjectures and refutations after the 1960s under the influence of Laka-
tos, Zahar, and Worrall and that he came to think that new scientific conjec-
tures are not generated at random; rather, they are influenced by scientists’
beliefs concerning which earlier conjectures have been refuted and which have
been left standing. According to Nanay, since Popper thought that there was a
strong analogy between ET and his theory of the growth of knowledge, apply-
ing his new views to ET led Popper to offer an alternative theory of evolution.
Ruse (1977) also claims that Popper was misled by his attempt to understand
ET from the perspective of his theory of the growth of knowledge, since there
are important differences between the two. We claim that there is a deeper
philosophical reason behind Popper’s proposing an alternative theory of evolu-
tion: he believed that ET provides in-principle explanations for typical events
but can never provide actual explanations for particular phenomena, and he
thought this because he believed that the main principle of ET is a priori.
Stamos (1996) argues that it is a consequence of Popper’s philosophy that
ET is not a genuine empirical science. Central to Stamos’s argument is the idea
that Popper held that a genuine scientific theory must postulate universal laws
and that they and only they are falsifiable and empirical in the strict sense; see
also Stamos (2007). Starting with Popper’s discussion of whether there are laws
of evolution in The Poverty of Historicism, Stamos claims that it is a conse-
quence of Popper’s position that the hypothesis of common ancestry is not a
genuinely scientific claim. However, as we explained in section 1, Popper does
not make any claim about the scientific status of ET in The Poverty of Histor-
icism (Ruse 1977; Hull 1999). Popper rightly denies that the Darwinian hy-
pothesis of common ancestry is a law, since that hypothesis is about the living
things on Earth; it is a singular statement about that token biota. Popper does
not deny that such spatiotemporally restricted existence claims are falsifiable
(Hull 1999).
In his later treatments of ET, Popper clearly states that there are testable em-
pirical parts of ET. And even Popper (1959a) is clear that only whole theoret-
ical systems are falsifiable, not the parts thereof.
12
Therefore, we object to
11. Given that Popper (1959b) proposed a propensity interpretation of probability, it is curious that
he never saw that there could be a propensity interpretation of fitness.
12. This holistic stance on testability encounters the well-known “tacking problem.”If U is untest-
able and T is testable, then U&T will often be testable.
HOPOS |Popper’s Shifting Appraisal of Evolutionary Theory
44
Stamos’s (1996) claim that Popper thinks ET is not a genuine scientific theory
because it contains untestable existential statements. Popper’s problem with
ET is that PNS is a priori. As Hull (1999) pointed out, this sort of misunder-
standing of Popper’s views stems from the false dichotomy in which the only
alternatives for Popper are genuine science and pseudoscience. Popper thought
that some theories in the social sciences and in psychology are pseudoscientific;
however, he did not think that all sciences that fail to postulate universal laws
of nature are thereby pseudoscientific.
In Popper’s view, there are three categories into which you can place a the-
oretical system: sciences that generate empirical knowledge and understanding
through the postulation of universal laws of nature, sciences that do not pos-
tulate suchlaws but nonetheless generate knowledge and understanding through
the proposal of methodological rules and the construction of models, and pseu-
dosciences that do not generate empirical knowledge and understanding at all.
Popper is placing ET in the second category, not in the third. His position co-
incides with the positions taken by several other subsequent philosophers who
claim that evolutionary biology or biology in general lacks universal empirical
laws of nature in the strict sense; this position is defended in different ways by
Smart (1963), Rosenberg (1994), and Beatty (1995).
13
For criticisms of Ro-
senberg’s and Beatty’s arguments, see Sober (1997).
With respect to the issue of tautology, Stamos cites Gould as an authority
and claims that Popper failed to see that fitness does not need to be defined in
terms of actual survival. Consequently, when Popper thought that NS is a tau-
tology, he was being misled by biologists who failed to understand how fitness
should be defined. Our reply, noted before, is that defining fitness as a prob-
abilistic propensity gives rise to tautologies of its own. In addition, it is well
understood today that Fisher’s fundamental theorem of NS really is an a priori
mathematical truth (Price 1972; Ewens 1989; Lessard 1997). Thus, Popper
did not make a mistake in thinking that there is a tautology problem that needs
to be addressed. His question was how a science like this can be valuable de-
spite the fact that it does not postulate universal empirical laws. This in itself is
an interesting philosophical problem regardless of the issue of what the best
understanding of fitness is.
Ruse (1977) argues that what Popper said in The Poverty of Historicism
agrees with the received view of evolution, rather than being a criticism of
it. Ruse thinks that Popper’s mistake lies elsewhere, especially in his Intellectual
Autobiography, where Popper claims that Darwinism is not a testable scientific
13. Mitchell (1997, 2000) and Woodward (2000) argue that biology can do without strict universal
laws. This, of course, involves no commitment as to whether there are such laws.
Elgin and Sober |SPRING 2017
45
theory. After citing Popper’s claim that Darwinism does not predict the variety
of life forms, Ruse explains in detail under what sort of initial conditions ET
would issue in this type of prediction. Ruse lists the following conditions: there
is a mainland with large populations and there are isolated areas with varying
conditions, populations vary genetically and phenotypically, and organisms in
the mainland sometimes get to those isolated areas. Ruse then writes: “Had one
reason to believe that life on the planet was fairly old (e.g., through the fossil
record or general complexity of structure), yet were one to find that absolutely
no speciation at all had occurred, then I suggest that, contra Popper, modern
evolutionists would be worried. Their theory, parts of it at least, would have been
falsified. The claims that they make about speciation would seem not to hold”
(1977, 645, emphasis ours).
The first question to ask about Ruse’s argument is whether “speciation has
never occurred”is an observation. If it is not, the fact that it is incompatible
with ET does not show that ET is falsifiable. But which part of ET conflicts
with “speciation has never occurred”? It conflicts with the hypothesis of com-
mon ancestry, but that hypothesis is not law-like, as it describes the history of
life on Earth. Popper grants that there are empirical elements in evolutionary
biology. What he denies is that there are empirical laws. Similar comments
apply to Hodge’s (1987, 233) point that Popper grants that spatiotemporally
restricted statements are both falsifiable and verifiable, and so are scientific.
Hodge correctly notes that such statements are to be found in evolutionary bi-
ology. However, this point does not touch Popper’s contention that ET lacks
universal empirical laws.
Evolutionary biologist John Maynard Smith (1986, 5) says that “the theory of
evolution is not falsifiable in the sense required by Popper.”He says that the the-
ory is a “logical deduction”and that Popper is demanding something more—that
“a scientific theory must say something about the world, and not merely about
logical necessity.”Maynard Smith goes on to say that “Darwinism as a testable
scientific theory can take various forms. I will give it first in the form in which
Darwin himself proposed it, and then in a ‘neo-Darwinist’form in which most
biologists hold it today.”Maynard Smith then says that the hypothesis of com-
mon ancestry is “clearly falsifiable ...as a single fossil rabbit in Cambrian rocks
would be sufficient.”This is a double mistake. The discovery that MaynardSmith
describes would require an important reworking of when various taxa first ap-
peared and how different taxa are genealogically related, but neither of these
changes shows that the hypothesis of common ancestry is false. And even if it
did, that does nothing to establish that there are falsifiable universal laws of evo-
lution. Maynard Smithnext turns to whathe regards as the second component of
Darwinian theory, that NS was the major, but not the only, cause of evolution.
HOPOS |Popper’s Shifting Appraisal of Evolutionary Theory
46
He says “it would be falsified if one could show that organisms lack one of the
three necessary properties of multiplication, variation, and heredity.”Here again,
what is falsified is not a law. When Maynard Smith then turns to the modernthe-
ory of evolution (6–7), he notes three innovations, but says that they are “changes
of emphasis rather than substance.”He does not say whether the modern theory
satisfies Popper’srequirements.
More than a decade later, the biologist Ernst Mayr reacted to Popper in a
lecture Mayr delivered in Stockholm when Mayr received the Crafoord Prize
from the Royal Swedish Academy of Science: “Despite the initial resistance by
physicists and philosophers, the role of contingency and chance in natural pro-
cesses is now almost universally acknowledged. Many biologists and philoso-
phers deny the existence of universal laws in biology and suggest that all regu-
larities be stated in probabilistic terms, as nearly all so-called biological laws have
exceptions. Philosopher of science Karl Popper’s famous test of falsification
therefore cannot be applied in these cases”(Mayr 2000, 82). It is ironic that
Mayr saw physicists and philosophers as devotees of determinism. By 1999,
quantum mechanics had been around for a long time. It is a probabilistic the-
ory, and many physicists and philosophers regarded it as making a strong case
for indeterminism. However, Mayr is right that probability claims are unfalsi-
fiable under Popper’s (1959a) strict definition of that concept; if a coin lands
heads in each of a thousand tosses, this does not falsify the claim that the coin’s
probability of heads on each toss is 1/2. Even so, Popper (190–97) supple-
mented his strict definition of falsifiability with something more liberal. This
relaxed definition of falsifiability says that the hypothesis that the coin is fair is
refuted by the thousand heads, since the hypothesis says that that outcome is
“sufficiently”improbable. It should be noted, however, that Popper thinks that
it is a matter of convention how improbable an outcome needs to be if it is
to “falsify”a hypothesis. In any event, the deeper question raised by Popper
(1974, 1978) lies elsewhere. The question is not whether laws can be proba-
bilistic but whether they must be empirical. Quantum theory is empirical,
but PNSP is not.
14
5. Popperian Themes in Current Philosophy of Biology
In this section, we relate Popper’s views on whether ET contains empirical laws
to some more recent literature. Popper claims that ET cannot be interpreted as
14. Although we have said something about the reception of Popper (1974, 1978) among philos-
ophers and scientists, there is another story to be told concerning the papers’reception among cre-
ationists. For some discussion of this, see Numbers (2006).
Elgin and Sober |SPRING 2017
47
a theory of forces, but we disagree. The different evolutionary causes that im-
pinge on a population (e.g., selection, mutation, migration) are represented in
the theory as vectors that combine to produce a resultant. Indeed, the Hardy-
Weinberg law functions as a zero force law in population genetics, and, as such,
it resembles the law of inertia in Newtonian physics. And PNS (or better,
PNSP) functions like the law of gravitation in that it describes how a system
will behave when a single force acts on it (Sober 1984).
15
Even if Popper had
known about these analogies, we think he would have dismissed them on the
grounds that these evolutionary principles are a priori statements rather than
empirical laws. Our response (to be fleshed out more in what follows) is that
laws need not be empirical. Matthen and Ariew (2002) agree with Popper that
ETshould not be interpreted as a theory of forces, although their reasons differ
from Popper’s.
As discussed earlier, Popper claims that ET can provide in-principle causal
explanations but cannot provide actual causal explanations of single events. Al-
though the distinction between how-possibly and how-actually explanations is
now standard in philosophy of biology (Brandon 1990; Plutynski 2004, 2005;
Lange and Rosenberg 2011),
16
it is a further step to claim that a priori models
furnish only the former and that empirical laws are needed to furnish the latter.
We think that a priori models can do both (Sober 2011a). We further suggest
that a priori dynamical models deserve to be called laws, since they do the same
explanatory and predictive work that empirical laws perform (Elgin 2003).
And while empirical causal laws are hard to find in evolutionary biology, they
obviously exist in physics. This marks an interesting difference between the
two sciences.
The package of ideas just described has come in for criticism. Lange and
Rosenberg (2011) criticize Sober (2011a) by arguing that explanations built
by using a priori models of NS are not causal explanations. Our response is that
Lange and Rosenberg’s thesis about explanation is perfectly compatible with
the claim that there are a priori causal models of NS, a claim that Lange and
Rosenberg do not dispute. We also think that Lange and Rosenberg fail to pro-
vide a convincing criterion for when an explanation is causal (Elgin and Sober
2015). Díez and Lorenzano (2015) develop a different line of criticism; they do
not contest the claim that there are a priori causal models in evolutionary biol-
ogy but argue that the same is true in physics. Our reply is that Sober (2011a)
already noted that a priori models can be constructed in any science that con-
15. See Hitchcock and Velasco (2014) for further discussion of whether the force idea applies to
ET.
16. Ewens (2004) thinks that there is a difference between models in physics and models in evo-
lutionary biology in that the former are more realistic and less idealized.
HOPOS |Popper’s Shifting Appraisal of Evolutionary Theory
48
tains empirical laws. The fact remains that there are few if any empirical laws in
evolutionary biology but that such laws obviously exist in physics.
17
Popper’s idea of situational logic resembles two ideas that continue to be
well received in philosophy of science. The first is Giere’s (1979, 1999) ap-
proach to the structure of scientific theories. According to Giere, a theory (re-
gardless of whether it is physical or biological or social) can be divided into two
components. First, there is an a priori definition of a kind of theoretical system;
for example, Newton’s three laws of motion plus his law of gravitation define
what it means to be a Newtonian system. The second component is empirical.
It involves the claim that this or that token physical system is a Newtonian sys-
tem. Notice that neither of Giere’s components is both general and empirical.
Popper’s analysis of ET as a situational logic is similar. For Popper, ET provides
a collection of statements that define a certain kind of object; let’s call it a Dar-
winian system. We then apply this definition to an actual case by saying that it
is an instance of a Darwinian system. Although the definition of a Darwinian
system is not empirically testable, specific instances of models constructed by
using this definition are. When some of these specific instances are refuted, we
learn something about the world—namely, that the assumptions that define a
Darwinian system are not satisfied in those cases. However, this does not show
that the definition of a Darwinian system is wrong. For Giere, this pattern ap-
plies both to physics and to biology, but for Popper there is an important dif-
ference because in physics at least sometimes the background laws of a theory
can be empirically refuted. Popper can grant that Giere’s formulation applies to
both sciences and still maintain that there is an important philosophical differ-
ence between them.
The second point of contact between Popper’s“situational logic”and more
recent ideas in philosophy of science is the semantic view of theories. Beatty
(1981), Thompson (1989), and Lloyd (1994) argue that ET should be inter-
preted by using this framework; Ereshefsky (1991) disagrees. These authors are
responding to the same problem that exercised Popper; they all are trying to
provide a philosophical explanation of ET’s success despite the apparent fact
that ET’s only “law”is a priori. Popper was against “conventionalism”(his term
for the view that so-called laws are mere definitions) in the context of physical
theories, although he thought that conventionalism is a coherent and consis-
tent philosophical view. His position on the social and biological sciences is
close to the semantic view of theories, and he embraces conventionalism about
the status of social and biological “laws.”
17. We do not claim that there cannot be empirical laws in evolutionary biology; our thesis con-
cerns the science as it now is constituted.
Elgin and Sober |SPRING 2017
49
6. Are There Any Empirical Laws in ET?
Popper argues that PNS is a priori, but he claims something more; he addition-
ally claims that there are no empirical laws at all in ET. What is his argument
for this stronger thesis? The same question arises if you think that PNSP is a
better rendering of the probabilistic concept of fitness. Popper is neglecting the
possibility that ET contains both source laws and consequence laws (Sober
1984).
18
PNS and PNSP are consequence laws; they purport to describe the
changes that will occur (deterministically, or with some degree of probability)
if certain initial conditions obtain. It is a separate task to describe the circum-
stances that make one trait fitter than another. These sources of fitness differ-
ence can be found in the physical environment that a population inhabits, in
the biotic environment outside of the population (e.g., the prey, predators, par-
asites, and diseases that the population has to deal with), and in the population
itself (as described by theories of sexual selection and of sex ratio evolution).
Source laws specify the synchronic supervenience bases for variation in fitness;
they have the form
If a population contains traits T
1
and T
2
at time t, and its environment
has properties E, and the organisms in the population have other traits
O
1
,O
2
,...,O
n
with frequencies p
1
,p
2
,...,p
n
at time t, then T
1
will
be fitter than T
2
at time t.
Some propositions of this form, like the models of sex ratio evolution con-
structed by Carl Dū
̈sing, R. A. Fisher, and W. D. Hamilton, are mathematical
truths (Sober 2011b), but are there any empirical laws that have this form?
There is another way to pursue the question of whether there are empirical
laws of evolution while granting that PNS and PNSP are not examples. Instead
of trying to determine whether there are empirical source laws, one might con-
sider something like the converse of PNS and PNSP. PNS and PNSP are both
“forward-directed”conditionals; they say (roughly) that if there is variation in
fitness at a given time, then you should expect change in trait frequency some-
time later. Campbell and Robert (2005) propose a “backward-directed”condi-
tional that approximates the converse of PNS and PNSP, namely,
(D*) For almost every pair of variant organisms aand bthat occupy the
same environment, if ais complex and is generally more reproductively
successful than b, then there exist heritable traits of aand bwhich, in
18. Popper is not the only philosopher who ignores the possibility that the theory of NS contains
source laws; Fodor and Piattelli-Palmarini (2010) make the same mistake.
HOPOS |Popper’s Shifting Appraisal of Evolutionary Theory
50
interaction with aand b’s environment, cause ato be more reproductively
successful than b.
This is an English paraphrase of the D* statement that Campbell and Robert
put in logical notation. The word “generally”in D* leads us to think that aand
bare types of organisms, not tokens. The same word also indicates that D* is
not about some brief period of time in which ahappens to out-reproduce bbut
rather concerns a sufficiently long stretch of time. Campbell and Robert (2005,
673) say that they prefer to call D* a model rather than a law.
19
The existential quantifier in D* is meant to mark the fact that the selection
process that makes trait T
1
fitter than trait T
2
in environment E
1
need not be
the same as the selection process that makes T
3
fitter than T
4
in environment
E
2
. Perhaps T
1
and T
2
are traits of a butterfly species and T
1
is fitter than T
2
because of predators, whereas T
3
and T
4
are traits of an elm species and T
3
is
fitter than T
4
because of diseases. Here Campbell and Robert are taking on
board the fact that the property of fitness is multiply realizable (Brandon
1978; Rosenberg 1978; Mills and Beatty 1979; Sober 1984); this has the con-
sequence that the property a population has of undergoing a selection process is
also multiply realizable. Campbell and Robert emphasize that their D* has the
quantifier order “for each pair of traits, there exists a story about natural selec-
tion”not the quantifier order “there exists a single story about natural selection
that applies to all traits.”We hope that our English rendering of D* captures this
point.
Richmond and Robert argue that D* is testable, and we agree. Our question
is whether D* would be a law of nature if it were true. Richmond and Robert
do not consider this question, but the question is worth asking in the context
of the current article. What we said earlier about Popper’s PNSR reformulation
of PNS is relevant here. Popper said that PNSR is false because of sexual selec-
tion and drift. We likened PNSR to the claim that all cancers are caused by
cigarette smoke. Our point was that PNSR is not law-like; even if it were true,
it would not be a law. We suggest that the same point applies to D*. For D* to
be a law, it must be nomologically impossible for there to be a process other
than NS that could often bring it about that one trait generally is more repro-
ductively successful than another. Here is a reason to doubt that this is so: for
any finite temporal interval, if the two variants aand bare equally fit and thus
19. Campbell and Robert formulate their D* in reply to Brandon’s (1978) contention that expla-
nations of specific evolutionary outcomes that appeal to NS are empirical, but that a general “principle
of natural selection”will be a tautology or will be so vague that it cannot be tested or will be empirically
false.
Elgin and Sober |SPRING 2017
51
evolve by the nonselective process of pure drift, there is a nonzero probability
that awill generally out-reproduce bduring that interval. This claim about
a single pair of traits applies to nsuch pairs, for any finite n.
20
Even if D* is not
an empirical law of nature, the question remains whether there are other
backward-directed propositions about evolution that are.
7. Concluding Comments
We have argued that Popper’s (1974) reason for holding that ET is not a test-
able scientific theory is that he thinks that ET lacks empirical laws. He thinks
this even though he grants that there are testable statements in ET. As Stamos
(2007) points out, postulating testable empirical laws is, for Popper, a necessary
condition for a theory to be scientifically testable. This is why Popper con-
cludes that ET is a metaphysical research program. Popper does not conclude
that ET is a pseudoscience; instead, he unsuccessfully attempts to reformulate
PNS and also tries to explain ET’s success in terms of his idea of situational
logic.
Those who agree with Popper that laws of nature must be empirical may
want to resist Popper’s conclusion by insisting that a theory can be testable
without its containing testable laws. Indeed, this suggestion finds a happy
home in Popper’s own concept of falsifiability; if S is a falsifiable statement,
and S and A are logically compatible, then S&A is falsifiable. The fact that
S is not a law and A is a tautology does not change that fact. We suspect that
Popper was misled by the examples he found in physics; Newton’s theory, rel-
ativity theory, and quantum mechanics are testable theories, and the laws they
propose are testable. As noted earlier, Popper does not hesitate to speak of evo-
lutionary “theory,”but he does demand that it conform to the pattern he found
in physics. Finding that it does not, Popper concludes that ET is a “metaphys-
ical research program,”not a testable scientific theory. However, if testable the-
ories need not contain testable laws, there is no need to draw Popper’s conclu-
sion.
This is a good reply to Popper, but we take issue with the assumption that
laws of nature must be empirical. Modern evolutionary biology is replete with
mathematical models that are mathematical truths. They describe the proba-
20. There is an additional point: even if D* is an empirical law, it does not do much to explain why
one trait generally out-reproduces another. In this respect, D* resembles “adaptationism,”at least on
some interpretations of that ism. Both D* and adaptationism are generalizations that can be tested
by case-by-case analyses of different traits in different populations, and if you believe these generaliza-
tions, they will play a heuristic role, guiding your search for specific explanations (Orzack and Sober
1994).
HOPOS |Popper’s Shifting Appraisal of Evolutionary Theory
52
bilities that populations have of changing their trait frequencies, given the pop-
ulation’s initial and boundary conditions. These dynamical models explain and
predict in the same way that empirical laws do; this is why we want to drop the
requirement that scientific laws must be empirical. We are not suggesting that
“bachelors are unmarried”is a scientific law; to say that some laws are a priori is
not to say that all a priori statements are laws. Popper was right that ET is dif-
ferent from theories in physics, but that does not mean that ET is untestable or
that it is not a genuine scientific theory.
REFERENCES
Beatty, J. 1981. “What’s Wrong with the Received View of Evolutionary Biology?”In Pro-
ceedings of the 1980 Biennial Meetings of the Philosophy of Science Association. East Lan-
sing, MI: Philosophy of Science Association.
———. 1995. “The Evolutionary Contingency Thesis.”In The Biological Sciences: The Sec-
ond Pittsburgh-Konstanz Colloquium in the Philosophy of Science, ed. G. Wolters and
J. Lennox, 45–81. Pittsburgh: University of Pittsburgh Press.
Brandon, R. 1978. “Adaptation and Evolutionary Theory.”Studies in History and Philosophy
of Science 9:181–206.
———. 1990. Adaptation and Environment. Princeton, NJ: Princeton University Press.
Campbell, R., and J. Robert. 2005. “The Structure of Evolution by Natural Selection.”Bi-
ology and Philosophy 20:673–96.
Díez, J., and P. Lorenzano. 2015. “Are Natural Selection Explanatory Models A Priori?”Bi-
ology and Philosophy 30 (6): 787–809.
Elgin, M. 2003. “Biology and A Priori Laws.”Philosophy of Science 70:1380–89.
Elgin, M., and E. Sober. 2015. “Causal, A Priori True, and Explanatory.”Australasian Jour-
nal of Philosophy 93 (1): 167–71.
Ereshefsky, M. 1991. “The Semantic Approach to Evolutionary Theory.”Biology and Phi-
losophy 6 (1): 59–80.
Ewens, W. J. 1989. “An Interpretation and Proof of the Fundamental Theorem of Natural
Selection.”Theoretical Population Biology 36:167–80.
———. 2004. Mathematical Population Genetics. 2nd ed. Interdisciplinary Applied Math-
ematics. New York: Springer.
Fodor, J., and M. Piattelli-Palmarini. 2010. What Darwin Got Wrong. New York: Farrar,
Straus & Giroux.
Giere, R. 1979. Understanding Scientific Reasoning. New York: Holt, Rinehart & Winston.
———. 1999. Science without Laws: Science and Its Conceptual Foundations. Chicago: Uni-
versity of Chicago Press.
Hitchcock, C., and J. Velasco. 2014. “Evolutionary and Newtonian Forces.”Ergo 1 (2): 39–
77.
Hodge, M. J. S. 1987. “Natural Selection as a Causal, Empirical, and Probabilistic Theory.”
In The Probabilistic Revolution, ed. L. Kruger, 233–70. Cambridge, MA: MIT Press.
Elgin and Sober |SPRING 2017
53
Hull, D. 1999. “The Use and Abuse of Sir Karl Popper.”Biology and Philosophy 14 (4): 481–
504.
Lange, M., and A. Rosenberg. 2011. “Can There Be A Priori Causal Models of Natural Se-
lection?”Australasian Journal of Philosophy 89 (4): 591–99. doi:10.1080/00048402.2011
.598175.
Lessard, S. 1997. “Fisher’s Fundamental Theorem of Natural Selection Revisited.”Theoret-
ical Population Biology 52:119–36.
Lloyd, E. 1994. The Structure and Confirmation of Evolutionary Theory. Princeton, NJ:
Princeton University Press.
Matthen, M., and A. Ariew. 2002. “Two Ways of Thinking about Fitness and Natural Se-
lection.”Journal of Philosophy 99 (2): 55–83.
Maynard Smith, J. 1986. The Problems of Biology. Oxford: Oxford University Press.
Mayr, E. 2000. “Darwin’sInfluence on Modern Thought.”Scientific American 283:79–83.
Mills, S. K., and J. H. Beatty. 1979. “The Propensity Interpretation of Fitness.”Philosophy of
Science 46:263–86.
Mitchell, S. 1997. “Pragmatic Laws.”Philosophy of Science 64:468–79.
———. 2000. “Dimensions of Scientific Law.”Philosophy of Science 67:242–65.
Munz, P. 2006. “Popper’s Darwinism.”In Karl Popper: A Centenary Assessment, ed. I. C.
Jarvie, K. Milford, and D. Miller, 131–41. Aldershot: Ashgate.
Nanay, B. 2011. “Popper’s Darwinian Analogy.”Perspectives on Science 19 (3): 337–54.
Numbers, R. 2006. The Creationists: From Scientific Creationism to Intelligent Design. Rev.
ed. Cambridge, MA: Harvard University Press.
Orzack, S., and E. Sober. 1994. “Optimality Models and the Test of Adaptationism.”Amer-
ican Naturalist 143:361–80.
Plutynski, A. 2004. “Explanation in Classical Population Genetics.”Philosophy of Science
71:1201–15.
———. 2005. “Explanatory Unification and the Early Synthesis.”British Journal for the
Philosophy of Science 56 (3): 595–609.
Popper, K. 1957. The Poverty of Historicism. London: Routledge & Kegan Paul.
———. 1959a. The Logic of Scientific Discovery. London: Hutchinson.
———. 1959b. “The Propensity Interpretation of Probability.”British Journal for the Phi-
losophy of Science 10:25–42.
———. 1963. Conjectures and Refutations: The Growth of Scientific Knowledge. London:
Routledge & Kegan Paul.
———. 1972. Objective Knowledge: An Evolutionary Approach. Oxford: Clarendon.
———. 1974. “Intellectual Autobiography.”In The Philosophy of Karl Popper, ed. Paul Ar-
thur Schillp, 3–181. La Salle, IL: Open Court.
———. 1978. “Natural Selection and the Emergence of Mind.”Dialectica 32 (3–4): 339–
55.
———. 1994a. Knowledge and the Body-Mind Problem: In Defence of Interaction. Ed. Mark
Amadeus Notturno. London: Routledge.
———. 1994b. The Myth of the Framework: In Defence of Science and Rationality. Ed. Mark
Amadeus Notturno. London: Routledge.
HOPOS |Popper’s Shifting Appraisal of Evolutionary Theory
54
Price, G. R. 1972. “Fisher’s‘Fundamental Theorem’Made Clear.”Annals of Human Genet-
ics 36:129–40.
Rosenberg, A. 1978. “The Supervenience of Biological Concepts.”Philosophy of Science
45:368–86.
———. 1994. Instrumental Biology; or, The Disunity of Science: Science and Its Conceptual
Foundations. Chicago: University of Chicago Press.
Ruse, M. 1977. “Karl Popper’s Philosophy of Biology.”Philosophy of Science 44 (4): 638–61.
Simkin, C. G. F. 1993. Popper’s Views on Natural and Social Science. Brill’s Studies in Epis-
temology, Psychology, and Psychiatry. Leiden: Brill.
Smart, J. J. C. 1963. Philosophy and Scientific Realism. London: Routledge & Kegan Paul.
Sober, E. 1984. The Nature of Selection: Evolutionary Theory in Philosophical Focus. Cam-
bridge, MA: MIT Press.
———. 1997. “Two Outbreaks of Lawlessness in Recent Philosophy of Biology.”Philoso-
phy of Science 64: S458–S467.
———. 2011a. “A Priori Causal Models of Natural Selection.”Australasian Journal of Phi-
losophy 89 (4): 571–89.
———. 2011b. Did Darwin Write the Origin Backwards? Amherst, NY: Prometheus.
Stamos, D. 1996. “Popper, Falsifiability, and Evolutionary Biology.”Biology and Philosophy
11 (2): 161–91.
———. 2007. “Popper, Laws, and the Exclusion of Biology from Genuine Science.”Acta
Biotheoretica 55 (4): 357–75.
Thompson, P. 1989. The Structure of Biological Theories. New York: SUNY Press.
Watkins, J. 1995. “Popper and Darwinism.”In Karl Popper: Philosophy and Problems, ed.
Anthony O’Hear, 191–206. Cambridge: Cambridge University Press.
Woodward, J. 2000. “Law and Explanation in Biology: Invariance Is the Kind of Stability
That Matters.”Philosophy of Science 68 (1): 1–20.
Elgin and Sober |SPRING 2017
55