Metascience
DOI 10.1007/s11016-014-9928-1
ESSAY REVIEW
A defence of string theory
Richard Dawid: String theory and the scientific method. Cambridge: Cambridge University Press, 2013, x+202pp, $90.25,
£52.81 HB
Keizo Matsubara
Ó Springer Science+Business Media Dordrecht 2014
String theory has been very influential within theoretical physics for the past few
decades. It is the most popular attempt to solve the problem of formulating a viable
theory of quantum gravity. Furthermore, it also unifies the other fundamental forces
within one theoretical framework. In string theory, it is assumed that what before
was thought to be point particles should really be seen as one-dimensional extended
entities, i.e. strings. Different particles are supposed to correspond to various
vibrational patterns of the strings.
In recent years, much critique has been directed towards string theory, and the
fact that it inhabits such a strong position within the theoretical physics community.
The main complaint is the complete lack of empirical confirmation of new
predictions derived from string theory. The books, criticizing string theory, that
have received most attention are the ones written by Smolin (2006), Woit (2006),
but there are many more critics some of which express much harsher criticism
towards string theory than the above-mentioned authors.
Richard Dawid opposes this negative evaluation of string theory in his book
String Theory and the Scientific Method. In this book, it is argued that string theory
in important ways has contributed to a change in how many theoretical physicists
view scientific method. And Dawid argues that this new attitude towards scientific
method is rational and should not be discounted. He points out that the attitude
towards string theory differs wildly between its critics and defenders. He argues that
this is due to an important difference in attitude towards what constitutes good
scientific method. Dawid goes as far as using the term ‘‘meta-paradigmatic shift’’ to
describe the situation. Critics of string theory are supposed to be embracing what
Dawid calls ‘‘the classical paradigm of theory assessment’’. These critics think that
the only way a scientific theory rightfully deserves to be trusted, and become part of
K. Matsubara (&)
Rotman Institute of Philosophy and Department of Philosophy, Western University, Stevenson Hall,
London, ON N6A 5B8, Canada
e-mail: [email protected]
123
Metascience
established science, is by receiving empirical confirmation for new predictions. The
defenders of string theory on the other hand take non-empirical reasons to trust
theories prior to empirical confirmation much more seriously.
The book can be seen as a defence of the rationality of performing research in
string theory due to strong non-empirical reasons to trust the theory. Dawid does not
deny that empirical confirmation still is the best reason to put trust in a theory.
However, he argues that it is not the only acceptable reason for accepting a theory.
Dawid provides a rather detailed description of how non-empirical theory
assessment is supposed to work. The most important part of this is to evaluate what
Dawid calls ‘‘limitations of scientific underdetermination’’. Of course, underdetermination of theory by data is a much-discussed problem within philosophy of
science. Dawid specifically uses the term ‘‘scientific underdetermination’’ to
distinguish the kind of underdetermination that he focuses on. Scientific underdetermination is underdetermination with respect to all presently accessible data. It
also presupposes a certain number of background assumptions about scientific
theory building and method. This distinguishes it from underdetermination with the
respect to all possible data. It also rules out possibilities that logically are allowed
but would never be taken seriously by the scientific community. It is for instance
taken for granted that we trust in induction in some sense.
To evaluate limitations of scientific underdetermination, Dawid presents three
different reasons for trusting theories prior to empirical confirmation. The first one
is called ‘‘the plain no alternatives argument’’ (NAA). Here, the reason is that given
a number of well-established background assumptions, it seems to be the only one
way to formulate a consistent theory. Arguments of this kind have been put forth to
defend string theory as the only way to both provide a theory of quantum gravity
and unify all fundamental forces. This is something that differentiates string theory
from other approaches that only attempt to formulate a theory of quantum gravity
and do not attempt any complete unification. Of course, one possible way to counter
this argument, which Dawid himself acknowledges, is by assuming that some of the
established principles need to be abandoned in a successor theory. The principles
could be found to be only approximately correct.
The second reason is called ‘‘the argument of unexpected explanatory
coherence’’ (UEA). This is when unexpectedly a number of earlier accepted
concepts and ideas naturally follow from the basic assumptions of a theory. String
theory was originally conceived as a theory of strong interactions, but later it was
realized that it could provide us with a working theory of quantum gravity and at the
same time incorporate the other fundamental interactions. This is probably one of
the main reasons for why string theory has been taken so seriously. It is important to
note how string theory demands an inclusion of gravity; this was not something that
was aimed for when the theory was first formulated. The empirical success of the
earlier theories, which conceptually are unified and incorporated in string theory, is
important for why this is taken to be good evidence. It is basically a non-empirical
no miracles argument. Sure, there are also a number of new unexpected ideas such
as extra dimensions of spacetime. These have, however, been put to use to give
explanations for how different spectra of particles can arise and why there are a
specific number of generations of particles. Unfortunately, no one has been able to
123
Metascience
formulate a solution of string theory that directly implies the specific predictions of
the standard model. At present, it seems that string theory allows for a huge number
of solutions, an issue we will return to later in this review, at least one of which
might very well be in perfect agreement with the observed phenomena.
The third reason is ‘‘the meta-inductive argument from the success of other
theories in the research program’’ (MIA). This is an induction based on previous
successful predictions in the research program. If a research program manages to
predict a number of previously unexpected phenomena that are later confirmed
empirically, we are then permitted to give strong credence that further predictions
are true already before their empirical confirmation. Here, the point is that string
theory is supposed to be seen as a being a part of the research program of particle
physics, which, as we know, has made many successful predictions, most recently
the discovery of the Higgs boson. I see one big problem with this claim, namely to
what extent can string theory be said to part of the same research program as that
which gave us the standard model? While there are strong ties, there are also big
conceptual differences and plenty of people in the particle physics community do
not endorse string theory.
So these are the three main reasons to be used in the non-empirical theory
assessment. They are supposed to guide us in our choice of theories to trust and
work further on. Dawid also points out that non-empirical theory assessment is not
something completely new; instead, it is argued to have been a part of science all
along. Without assuming that there is a certain amount of limitations on scientific
underdetermination, trust in new predictions based on earlier successful theories
would not make sense. There are many situations where we trust new predictions of
our theories. Take the Higgs boson or the top quark for instance, even before their
discoveries it is fair to say that particle physicists strongly believed that they existed.
It is also argued in the book that the development where we more and more rely
on theoretical arguments has been going on for quite some time. This is called ‘‘the
marginalization of the phenomena’’. Dawid describes how physics has distanced
itself further and further from our concrete everyday experience. In classical
physics, the objects described are directly observable, whereas in microphysics, we
can no longer directly observe the objects but only observe certain predictions. It is
also described how observable phenomena become less and less important in
concept formation and how previously established theories become more and more
important compared to phenomena in how we formulate new theories.
Dawid argues that with string theory, the importance of non-empirical theory
assessment has become even more important. It is claimed that while non-empirical
theory assessment has been part of scientific method all along, there are a number of
new factors introduced in string theory.
One of the new things is that string theory is often claimed by its supporters to be
a final theory in a certain sense. Critics may discard such claims as expressing the
hubris of string theorists. Dawid, on the other hand, explains why he thinks that we
can take these claims more seriously. That all fundamental interactions are unified is
one obvious reason. Also string theory suggests that there is a minimum length; if
there are no lengths shorter than this, we do not need to worry that string theory will
be replaced by another theory at shorter length scales.
123
Metascience
The aspect of string theory that Dawid finds most important though is the lack of
free parameters, he uses the expression ‘‘structural uniqueness’’ to describe this. In
other theories or theoretical frameworks, we are free to adjust parameters. The
parameters are then not supposed to be derived from first principles but be adjusted
to fit experiment. In string theory, this is not the case. A specific solution of string
theory will also give us all the values that were previously given by adjustable
parameters. If, as the hope with string theory used to be, there exists very few stable
solutions of string theory then the theory would furthermore be what Dawid calls a
‘‘highly predictive structurally unique theory’’. Dawid argues that the structural
uniqueness is a good reason to take final theory claims seriously.
I think it is quite plausible that if string theory would turn out to be a highly
predictive structurally unique theory, and one of the few allowed solutions implies
the precise predictions of the standard model, then that would be very impressive. It
would give us very strong reasons to trust string theory even in the absence of new
confirmed empirical predictions. Unfortunately, this does not seem to be the case;
instead string theory appears to be very far from highly predictive. It is estimated
that there exists a huge number of sufficiently stable solutions; the number 10500 is
often mentioned. This has led to further scepticism and controversy regarding string
theory. Some string theorists hope that this conclusion is wrong and that string
theory will turn out to be highly predictive after all. Other defenders of string theory
have accepted the conclusion and believe in the existence of a multiverse and
invoke the controversial anthropic argument. I will not here discuss this issue any
further.
It is good that Dawid discusses and in some detail evaluates the rationality and
importance of non-empirical factors when evaluating theories. This is something I
think it would be quite valuable to develop further and be discussed much more.
Having said that, it should be stressed that Dawid still thinks that it would be even
better to have empirical confirmation. On page 118, he writes,
A position along these lines does not deny the important difference between
theoretical reasoning and empirical testing. Neither does it deny that empirical
testing must be the ultimate goal of natural science nor that empirical
confirmation substantially enhances the trustworthiness of an individual
scientific theory.
While I agree with much of what Dawid claims in the book there are aspects that I
find quite troublesome and I do think that his view on string theory is a bit too
optimistic. I find the claim that string theory has led to a meta-paradigmatic shift to
be quite hyperbolic. A change in scientific method could be part of the traditional
paradigm shifts already described in Kuhn (1962). Also, neither Dawid nor I
endorse any radical view regarding incommensurability, which is often associated
with Kuhn. Another reason why I think this talk of a meta-paradigmatic shift to be
an exaggeration is that if, as Dawid argues, non-empirical theory assessment has
been part of science all along, the change in method is not really that radical.
Instead, the story can be told such that there is much more continuity in the
development of scientific method. Sure, there are differences in attitude towards
scientific method among different scientists, but I do not think they are as extreme
123
Metascience
as described in this book. Using the expression ‘‘meta-paradigmatic’’ shift to capture
this conflict is inappropriate.
In (Johansson and Matsubara 2011), my co-author and I already criticized
Dawid’s claim that there has been a meta-paradigmatic shift within theoretical
physics. In the book, Dawid comments on this critique. Unfortunately, his
description of our views is quite misleading. In our paper, we discuss how string
theory would be evaluated from the point of view of four different views in
philosophy of science, namely those of the logical positivists, Popper, Kuhn and
Lakatos. Given Dawid’s description of our view, we endorse a Lakatosian view and
conclude that string theory is degenerative. It is true that given Lakatos’ own criteria
string theory is degenerative. This does not in itself mean that research in string
theory must be abandoned, even according to Lakatos, especially since there are no
progressive rivals in the area of quantum gravity. While my co-author and I are
quite sympathetic to the views of Lakatos, we do not wholeheartedly defend his
views. Given Dawid’s description of our paper, we seem to be much more critical
towards string theory than we actually are. He also claims that we do not take into
account that criteria for evaluating theories may change. However, we actually
explicitly states that scientific method can change, and we also discuss internal
theoretical reasons for trusting string theory, albeit in less detail and with slightly
less optimism than Dawid. Our conclusion is rather that it is perfectly reasonable to
perform research in string theory, we only advice that in the absence of empirical
confirmation it would be good with more pluralism in the research on quantum
gravity. This is because we agree with what Dawid stated in the quotation above,
namely that empirical testing must be the ultimate goal of natural science. It seems
that Dawid has missed the point about why we argued against the claim that there
has been a meta-paradigmatic shift.
In the final chapter of the book, Dawid enters into the discussion on scientific
realism. The limitations of scientific underdetermination could be seen as making
most sense assuming scientific realism. However, in the book, it is argued that it
could instead be seen as an alternative to traditional scientific realism. Dawid also
emphasizes that given the various dualities in string theory, it is difficult to defend a
traditional form of scientific realism in terms of individual objects. Dualities of
string theory are such that physically equivalent theoretical formulations may
describe reality in wildly different ways that prima facie would describe completely
different world. For instance, by treating different entities as fundamental and
disagreeing about the geometry and topology of spacetime. Physicists do not treat
the dual descriptions as being in competition, instead they think of them as
representing the same underlying reality. For this reason, Dawid advocates that
string theory is best understood in terms of a form of structural realism. The
structure that Dawid urges us to be realists about, if we accept string theory, is the
total mathematical structure describing all dualities and how they are connected. I
am not quite sure what that is supposed to mean though. I think there are much more
of interest to say about how to understand dualities and what relevance they may
have for the debate on scientific realism. Also, I find it important to carefully reflect
upon the relation between a mathematical formalism and its physical interpretation.
I am largely sympathetic to Dawid’s conclusion that the dualities lead to trouble for
123
Metascience
a traditional object based ontology. What I think requires much more development
and clarification is what his positive thesis really amounts to.
To summarize: I think that the book makes an important contribution to
philosophy of science by discussing and arguing for the role that non-empirical
theory assessments play in science. I also find it pleasing to see a book in philosophy
of science arguing in favour of string theory. As described above, I do, however,
unfortunately find some of the claims to be a bit too optimistic and exaggerated.
References
Johansson, L.G., and K. Matsubara. 2011. String theory and general methodology: A mutual evaluation.
Studies in History and Philosophy of Modern Physics 42: 199–210.
Kuhn, T.S. 1962. The structure of scientific revolutions. Chicago, IL: Chicago University Press.
Smolin, L. 2006. The trouble with physics. Boston, MA: Houghton Mifflin.
Woit, P. 2006. Not even wrong: The failure of string theory and the continuing challenge to unify the laws
of physics. London: Jonathan Cape.
123