Content uploaded by Bendik Hellem Aaby
Author content
All content in this area was uploaded by Bendik Hellem Aaby on Sep 27, 2016
Content may be subject to copyright.
12 DIFFErEnT aPPrOachES TO ThE PhIlOSOPhy OF ScIEncE
In the philosophy of science we are often concerned with
scientic concepts such as theory, explanation, laws of
nature and so on. Philosophers readily oer dierent theo-
ries or conceptions of such concepts, but there are many
dierent approaches to understanding and theorizing abo-
ut these concepts. In this essay I sketch what I take to be
two central approaches to the philosophy of science. I call
these the global and the local approach to philosophy of
science. To illustrate, I discuss a well-known case from the
philosophy of science, namely philosophical theories of
scientic explanation. e deductive-nomological model
of the logical empiricists (e.g. Hempel and Oppenheim
1949, Hempel 1965, Nagel 1961) will illustrate the global
approach to scientic explanation, while (new) mechanis-
tic explanation (e.g. Machamer et. al. 2000, Craver 2007,
Bechtel 2006) will illustrate the local approach.2
I am mainly concerned with meta-philosophical qu-
estions in this essay. An important fact about science is
that its body of knowledge, disciplines, elds, theories and
methods is larger than ever; and the philosophy of science
has likewise expanded with numerous new elds and sub-
classes of the philosophy of the special sciences, such as
the philosophy of biology, neuroscience, psychology, social
science, etc. In philosophy in general, meta-philosophy
has been thriving in the last decades, with philosophers
taking a keen look at their own disciplines, for example
ethics, epistemology and metaphysics. So, in this essay I
take a look at some of the meta-philosophical aspects of
the rapidly expanding sciences and philosophy of science.
In doing this, I point out some of the consequences and
obstacles of doing philosophy of science on the global and
the local approach, with the aim of contributing to a me-
ta-philosophical discussion in the philosophy of science. I
also argue that, because of the rapid expansion of discipli-
nes and elds in both science and philosophy of science,
a local approach, together with theoretical and metho-
dological pluralism, seem to oer the best framework for
understanding how scientic concepts work and should
work. However, a local approach and pluralism face chal-
lenges of their own. I consider these challenges, arguing
that a local approach, even with such challenges, is still
better o than a global approach.
The Global and Local Approach to the Philosophy
of Science
ere are at least two ways to theorize about scientic
concepts. First, one can start with a general theory, which
ideally covers all the concepts across all of science. is
is what I call the global approach. Another way is to start
theorizing within a science, scientic eld or discipline
and develop a theory of a concept that ts the practice
exhibited in that particular science. is is what I call the
local approach. ere are certainly other approaches for
theorizing about scientic concepts. However, looking at
dIffErENT APProAchES
To ThE PhILoSoPhy of
ScIENcE1
By Bendik Hellem Aaby
This is a meta-philosophical essay characterizing two possible approaches to theorizing about scientic
concepts: the global and the local approach. A global approach is characterized as taking philosophical
theories of scientic concepts to be about science as a whole, while a local approach focuses on a particular
scientic discipline. The consequences of doing philosophy of science on either approach are explored,
exemplied by two distinct philosophical theories of scientic explanation. I argue that the local approach
seems to be the better one when trying to keep in touch with today’s multitude of scientic disciplines.
13
bEnDIK hEllEM aaby
these two dierent approaches as two of the main approac-
hes can bring out some interesting features of how each
approach deals with the generalizability and application of
theories, and how one can (and should) construct counter-
examples to the theories of the dierent approaches.
Before moving on to these discussions, let us look at
an example of a global and a local approach to the same
scientic concept, namely scientic explanation. Let us
start with a rough sketch of what a scientic explanation
is, both on a local and a global approach. A common place
to start is that a scientic explanation is knowledge of why
or how something occurs or occurred (Salmon 1989). is
can be contrasted with merely descriptive knowledge. One
thing is to know that the seasons are cyclical; another is
to know why the seasons are cyclical. Both types of know-
ledge are attainable by observation, but in the latter case
we can say that we have an understanding of the seasons.3
So, understanding why or how is a good rst appro-
ximation to characterizing scientic explanation. For our
purposes, let us say that an explanation is the answer to
a why- or how-question. en, a philosophical theory of
scientic explanation should tell us what it takes to sa-
tisfactorily answer such questions in science. e global
approach I am using as an example, the deductive-no-
mological model of explanation, sets out to nd a general
structure for explanation, i.e. a logical pattern that gives us
the necessary and/or sucient conditions for satisfactorily
answering the why- or how-question. e local approach,
the new mechanistic theory of explanation, on the other
hand, starts with a characterization of the abstract pattern
of answering why- or how-questions by scientists in spe-
cic disciplines or elds.4 Let us start with the global ap-
proach in more detail.
The DN-model of Scientic Explanation
ere are two key elements in the DN-model of scientic
explanation, the explanandum and the explanans. e ex-
planandum is the phenomenon to be explained, while the
explanans are the class of sentences that together constitute
the explanation of the expla-
nandum. In other words, the
explanandum is that which
is being explained, while
the explanans is that which
does the explaining. Further, there are two sets of condi-
tions that have to be met in order for the explanans to be
an adequate explanation of the explanandum (Hempel &
Oppenheim 1948:136–137).
First, the deductive condition: e class of sentences
that constitute the explanans must gure as premises in a
deductive argument, where the explanandum gures as the
conclusion. e explanans-sentences must be true and the
argument must be sound.
Second, the nomological condition: e class of sen-
tences in the explanans must contain at least one law (of
nature) or a universal generalization. e law or genera-
lization has to be an essential premise in the argument.
at is, the “law-premise” needs to be essential in the sense
that the derivation of the explanandum from the explanans
would not be valid without it (1948:137).
Usually, when we want an explanation of something
specic, the why-question we want answered is concerned
with something particular. is means that something
more than universal generalizations such as laws of nature
or regularities are needed for a derivation of the explanan-
dum-phenomenon from the class of sentences included in
the explanans. Such sentences, which provide information
of particular or kinds of events, are called antecedent condi-
tions.5 ese sentences are descriptive. For example, an an-
swer to the question of why the seasons are cyclical needs
more than just Newton’s laws of motion; you need par-
ticular facts about the earth as well, such as earth orbital
tilt and its distance to the sun. Hempel and Oppenheim
(1948:138) give us the following schema for the formal
structure of a deductive-nomological explanation:
(P1) C1, C2… Cn (Antecedent condions)
(P2) L1, L2… Ln (General laws)
(C) E (Phenomenon to be explained)
(P1) and (P2) are the premises of the argument, the ex-
planans, and (C) is the conclusion, the explanandum.
(C) is derived, by logical deduction, from (P1) and (P2)
(1948:139).
Ideally, for the logical empiricists, all scientic explana-
tions should have this logical structure. However, it is im-
portant to note that the logical empiricists did not believe
that all explanations lacking
such structure or unable to
be restated this way were
non-scientic. ey were
well aware of other types
of explanation, such as functional, historical or statistical
explanations.6 For example, Hempel made a new model
to incorporate statistical explanations called the inductive
statistical model (Hempel 1965) for particular cases and
deductive-statistical model for universal cases. Together all
Artikkel Samtale og
kritikk Spalter Brev
A philosophical theory of scientic explanati-
on should tell us what it takes to satisfactorily
answer such questions in science.
14 DIFFErEnT aPPrOachES TO ThE PhIlOSOPhy OF ScIEncE
these models are called Hempel’s cowering law model for
explanation.7
In any case, the salient point here is that the logical
empiricists start with a general theory of explanation. e
theory has all of science as its scope, and the domain of
discussion is all the scientic elds and disciplines. e
DN-model has no restrictions, except that it is a model
for scientic explanation. e cases that do not t the
logical pattern can be seen as temporary challenges that
should ideally be overcome in time. Because of this, con-
structing descriptive counter-examples to the DN-model
can be done with fairly simple cases, dealing, e.g., with
cases where we nd an explanation that lacks the logi-
cal structure of the covering law model.8 A related point
about counter-examples is that you can use explanations
that are not riddled with technical scientic concepts
from a specic discipline. Because of the logical nature
of the model, examples of non-explanations that share
the logical structure are sucient as counter-examples to
the model. As an illustration, consider one of the most
famous counter-examples to the DN-model, the agpole
counter-example.9
e length of the shadow of a agpole can be explai-
ned deductively or derived by the relevant antecedent con-
Illustrasjon: Lisa Marie Mrakic
15
ditions; e.g. the length of the agpole and the angle of the
sun above the horizon, and the nomological conditions;
laws of rectilinear propagation of light. However, vice
versa, the length of the agpole can be derived from the
length of its shadow coupled with the antecedent and no-
mological conditions already mentioned. is “backward”
derivation does not explain why the agpole is n-meters
high; it gives us no explanatory relevant information as to
how the agpole obtained its height. is counter-example
illustrates the problem of explanatory asymmetry for the
DN-model. Such counter-examples can be constructed
from any science and still be relevant for the discussion of
the DN-model. at is, the domain of discussion on the
global approach is science as a whole.
The New Mechanistic Theory of Explanation
Let us turn our attention to an example of a local approach
to scientic explanation: New mechanistic explanation. It
is called the new mechanistic theory of explanation becau-
se it is not, strictly speaking, based on the old mechanistic
world picture of Descartes and Newton. A mechanism on
the new picture is explanatory in virtue of its constituents
and their activities and organization in a complex system.
In order to see how this approach is local and how it is dif-
ferent from the global approach exemplied by the DN-
model of the logical empiricists, we should look into the
details of what it takes to explain a phenomenon mecha-
nistically. Let us start with the denition of a mechanism
given by Machamer et al. (2000):
Mechanisms are entities and activities organized such that
they are productive of regular changes from start or set-up to
nish or termination conditions. (Machamer, et al. 2000:3)10
To explain a phenomenon mechanistically then, is to give
a description of the relevant entities and activities, and
how they are organized such that they regularly produce
a phenomenon or a part of a phenomenon. A mechanistic
explanation of a phenomenon, then, consists of (1) enti-
ties, i.e., the relevant constitutive parts; (2) activities, i.e.,
the relevant causal relationships obtaining between the en-
tities; and (3) the organization of the relevant entities and
activities (MDC:3). In the language of the logical empiri-
cists we can say that the explanandum is the phenomenon,
while the explanans are (1) – (3). e notion of production,
regularity and start or set-up and nish or termination con-
ditions are the conditions and constraints of the organized
activities and entities that operate within the mechanism
(MDC:3).
bEnDIK hEllEM aaby
e entities and activities are the ontologically funda-
mental part of a mechanism. e other conditions and
constraints (i.e. organization, regularity, production,
set-up and nish conditions) are relative to the relevant
entities and activities. e organization of the activities
and entities in a mechanism is what determines how the
mechanism operates, i.e., it determines the ways in which
activities and entities produce the phenomenon. e enti-
ties are organized in a spatial manner, while the activities
are organized in a temporal manner. e entities must be
spatially organized in such a way that they can engage in
activities with each other, while the activities must be tem-
porally organized in such a way that they ensure the beha-
vior of the entities and ultimately the mechanism (e.g. the
temporal order and duration must be right for the entities
to be capable of engaging in the relevant activities). e
notion of regularity is explained by the predictability of the
behavior of the mechanism under the same or similar cir-
cumstances. So, if the activities and entities are organized
in a particular way, a mechanism should go from set-up to
nish conditions without the lack of productive continuity.
Productive continuity is the seamless and complete beha-
vior of the entities and activities that produce the pheno-
menon (MDC:3).11 ere are several other mechanists
that characterize regularity dierently. Some hold that it is
not necessary for a mechanism to be regular, while others
say that regularity should be understood in a temporal
sense, as repetitive behavior of the mechanism.12
e phenomenon is the behavior of the mechanism
seen at the level of the system, i.e., of all the activities
and entities and their organization throughout the hierar-
chical structure of the mechanism and its components.
Components, relative to a mechanism, are just a shor-
thand term for describing (1) – (3). If we were to look
at the composition (i.e. the organization of the activities
and entities) of one of these components, we would refer
to them as sub-components relative to the mechanism.
But if one looks at a component in isolation, we would
refer to the component as the mechanism, and the sub-
component as the components. us, the boundaries of
a mechanism are context-relative; they are relative to the
mechanism that produces the phenomenon.
e boundaries of the mechanism pick out the rele-
vant activities and entities that are organized such that the
mechanism produces the phenomenon. Further, to under-
stand what entities and activities (and what organization)
that produce the phenomena one need to decompose the
mechanism into hierarchies. is is a process of picking
out the explanatory relevant parts (entities and activities).
Artikkel Samtale og
kritikk Spalter Brev
16 DIFFErEnT aPPrOachES TO ThE PhIlOSOPhy OF ScIEncE
In other words, in decomposition we individuate the entities
and activities of the mechanism. e next step is localizing
the kinds of entities and the relevant subset of properties
for the entities to engage in the activities and their spatial
location, and kinds of activities and their temporal loca-
tion. e specic organization of these relevant activities
and entities produces the phenomenon. Decomposition
and localization are most commonly referred to in the
discovery of mechanism (Bechtel & Richardson 2010
[1993]:23–27), but are essential for the explanation to cite
relevant facts.
e most common way for the mechanists to present
their theory is by showing how well its abstract structure
conforms to scientic practice (that is, how scientists ac-
tually explain), usually with examples from the relevant
science they are concerned with. For example:
My aim in this book is to construct a [mechanistic] model
of explanation that reects, rather than merely accommoda-
tes, the structure of explanations in neuroscience. (Craver
2007:2)
So, in contrast to the global approach of the logical em-
piricists, the mechanists begin locally, nding particular
disciplines or elds where the scientic practice conforms
to the mechanistic pattern of explanation. Whether or not
the mechanists want their theory of explanation to ap-
ply to all of science is unclear, the mechanists are most
often concerned with mechanisms in particular scientic
disciplines.
Since the approach is local, the extent to which the
mechanistic theory of explanation is, or should be, applied
to scientic explanation in general is not given in advance.
ere is no explicit and uniform “mechanistic program”,13
and no prior guidelines to the generality of the theory. e
preliminary restrictions of the theory has to be explicitly
stated, and is not implicitly given as with the global ap-
proach of the logical empiricists. We can probably nd,
on the one hand, a mechanist who nds deviation from
the explanatory relevant facts (1) – (3) unacceptable, and
on the other, someone who welcomes deviations in order
to apply the mechanistic theory of explanation to a vast
number of dierent sciences, scientic elds or disciplines.
e point being, in contrast to the global approach, when
doing philosophy of science on the local approach one has
to explicitly state the preliminary restrictions in order to
indicate the intended scope of the theory and domain of
discussion. e scope will determine the relevance and do-
main of possible counter-examples.
Consequences of the Global and Local Approach
Now that we have gone through the details of the examples,
let us turn our attention to the consequences that we have
only briey touched upon. On the local approach, you
have a need for explicitly stated restrictions for the theory
and domain of the discussion. On the global approach,
science as a whole is the domain, so no special restrictions
or specications of scope of the theory need to be stated.
us, on the global approach, one strives for theoretical
monism. A theory on the local approach, on the other
hand, can be one of many. For example, one can have sta-
tistical and mathematical explanations in science that are
not translatable into the mechanistic schema, without this
causing the mechanistic theory to be useless. It is just that
there are areas in science that do not warrant mechanistic
explanations. On the global approach, explanations that
do not conform to theory are seen as serious obstacles to
the theory. Another salient dierence between the global
and the local approach is how they deal with the genera-
lity of their theories. e global theory is general from the
outset, while the local is particular.
The Generality of Theories
We can characterize the dierence in generality on the
global and local approach by placing the approaches at
each end of a scale of generality, with the global approach
at the general end and the local at the particular end.14
However, through discussion and counter-examples the
theories both move towards the center of the scale. e
generality of the global approach is restricted by showing
the implausibility of its application to several sciences or
scientic explanations, while the generality of the local ap-
proach is expanded by applying the mechanistic theory of
explanation to more scientic disciplines than those origi-
nally intended to be within the scope of the theory. As we
saw earlier, the intended generality of a theory has conse-
quences for the discussion around it. Let me illustrate this
point with an example.
Let us say you are challenging the DN-model of expla-
nation. You argue that the DN-model has a serious aw
in its nomological requirement, for example the case of
the non-existence of laws of nature in biology or the social
sciences. If this argument is well-founded, the advocator
of the DN-model will have a serious problem with his
theory. Because the intended scope is all scientic expla-
nations he must nd a way to overcome such a challenge
or claim that biology or the social sciences do not explain
scientically. He could also claim that the laws gure at a
more fundamental level, so if the lawless theory in bio-
17
As we noted above, the body of scientic knowledge
has never been larger and the scientic disciplines and
elds have never been more numerous. Also, the number
of tools and methodologies utilized in the dierent areas
of sciences are larger than ever. Similarly, the philosophy
of science has expanded and never before have we seen
so many sub-categories, such as philosophy of psychology,
biology, chemistry, neuroscience, etc. Such sub-categories
reect how the philosophy
of science has moved to-
wards a local approach. I
think this development is a
healthy sign for philosophy
of science, and that this is a
reasonable reaction to the increasing complexity of scien-
ce. In order to take scientic practice into account when
philosophizing about science, a close look at the peculia-
rities of the dierent disciplines, elds, methodologies etc.
is needed. Of course, a global approach where a theory
ideally covers all the dierent sciences may still seem, at
least for many, the primary aim for the philosopher of
science. However, as the complexity of the current state
of science and philosophy of science suggests, this may be
harder than previously thought. So, in order to advance in
philosophy of science, it seems reasonable to model our
approach after science itself, where increased specialization
and divisions is an eective way to progress.
However, a local approach is not without its own chal-
lenges. As we have already mentioned the local approach
needs to be very clear in xing the relevant restrictions and
the intended scope of the theory or concept. Further, a
consequence of the increasing specialization that usually
follows from a local approach is that the philosopher of
science has to be well versed in the science or scientic
eld or discipline he is concerned with. is last challenge,
the challenge from specialization, may be a “necessary evil”
that cannot be circumvented, because the way science it-
self is developing should be of paramount importance to
the philosopher of science. e point here is merely to il-
lustrate that an increase in degree of specialization is just
as much a reality for the philosopher of science as for the
scientists themselves.
Conclusion
In this essay I have characterized what I take to be two chief
approaches to the philosophy of science, the global and
the local approach. I have illustrated, with examples from
scientic explanation, the consequences and challenges of
each approach. e main challenge for the global approach
bEnDIK hEllEM aaby
logy or the social sciences were reduced to physics, laws
of nature would be included in the explanans. is would
need a theory of reduction, and we will not go further into
that here. What is of importance to us is that no matter
what area you come from, no matter what scientic eld
or discipline you use as an example, the advocator of the
DN-model is always challenged by a scientic counter-
example and it will always gure in the relevant domain of
discussion.
Discussions on the lo-
cal approach, on the other
hand, are dierent. Let us
say that you are challenging
the applicability of mecha-
nistic explanation in the social sciences. For example, the
so-called mechanisms in explanations of social pheno-
mena at a group-level are not decomposable into orga-
nized and regularly working components (activities and
entities). How the advocator of the mechanistic approach
will deal with this is not given in advance as it is with the
global approach. He could say that this is of course true,
but he is only concerned with the proximal-physiological
parts of biology and sees no reasons why he should have to
deal with problems in the social sciences. Or he could take
the challenge, for example by loosening the constraints on
what it takes to be a mechanism in order to accommodate
the ndings of his challenger. e point being, on the local
approach one has to explicitly state the intended scope of
the theory in order to lay down the restrictions for the
domain of discussion one is willing to have or is interested
in having.15
Local or Global – How Should we Approach
Philosophy of Science?
ere is a normative question that has been looming
throughout this essay: should we opt for a local or global
approach in the philosophy of science? Firstly, we need
to take into account the plurality of scientic methodolo-
gies, disciplines and elds in order to get an accurate per-
spective on how we do and should theorize about science.
Secondly, and very much related to the rst point, the
perspective forced by the current plurality in science and
philosophy of science (and science in the foreseeable fu-
ture) point out the complexity in seeing science as a whole
and theorizing about scientic concepts in complete gene-
rality. ese considerations suggest that a local approach,
based on the current plurality of science, may be the most
reasonable approach to take, at least if one wants to take
current scientic practice seriously.
Artikkel Samtale og
kritikk Spalter Brev
Because the intended scope is all scientic ex-
planations he must nd a way to overcome
such a challenge or claim that biology or the
social sciences do not explain scientically.
18
is the diversity and plurality of science, which makes it dif-
cult to develop a theory for science as a whole. e local
approach is challenged by the increasing specialization of
science and the lack of given restrictions on the scope of
the relevant theory. It is argued that the local approach is
what reects current philosophy of science, and that this
is the way we should approach philosophy of science. e
challenge from specialization may be a “necessary evil” of
the ever-increasing complexity of science, and is perhaps
unavoidable. But the challenges the local approach faces
from restrictions, domain and scope should be faced head
on by the philosopher of science. is can be done by pay-
ing close attention to, and explicitly stating, the relevant
restrictions on the intended scope of the theory. Maybe
the only way to theorize globally about science is oblique-
ly: by engaging in meta-philosophical discussion.
lITEraTUrE
Andersen, H. 2012, “e Case for Regularities in Mechanistic Causal
Explanation”, Synthese, 189: 415–432.
Bechtel, W. 2006, Discovering Cell Mechanisms: e Creation of Modern
Cell Biology, Cambridge University Press, Cambridge,
NewYork, NY.
Bechtel, W. & Richardson R. C. 2010 [1993], Discovering Complexity:
Decomposition and Localization as Strategies in Scientic
Research, Second Edition, MIT Press/ Bradford Books,
Cambridge, MA.
Bromberger, S. 1966, “Why Questions”, in Mind and Cosmos: Essays
in Contemporary Science and Philosophy, Colodny, R. (ed.),
86 –11, University of Pittsburg Press, Pittsburg.
Craver, C. F. 2007, Explaining the Brain: Mechanisms and the Mosaic
Unity of Neuroscience, Oxford University Press, Oxford.
Craver, C. F. & Tabery, J. 2015, “Mechanisms in Science”, e
Stanford Encyclopedia of Philosophy, (Spring 2016 Edition),
Edward N. Zalta(ed.), URL = <http://plato.stanford.edu/
archives/spr2016/entries/science-mechanisms/>.
Hempel, C. G. 1942, “e Function of General Laws in History”,
Journal of Philosophy, 39 (2): 35–48.
––––1965, Aspects of Scientic Explanation; and Other Essays in the
Philosophy of Science, e Free Press, New York and Collier
and McMillian Ltd., London.
Hempel, C. G. & Oppenheim, P. 1948, “Studies in the Logic of
Explanation”, Philosophy of Science, Vol. 15: 135–175.
Machamer, P., Darden, L., & Craver, C. F. 2000, “inking About
Mechanisms”, Philosophy of Science, Vol. 67 (1): 1–25.
Nagel, E. 1961, e Structure of Science: Problems in the Logic of
Scientic Explanation, Harcourt, Brace and World, Inc.,
New York.
Salmon, W. C. 1989, “Four Decades of Scientic Explanation”, in
Scientic Explanation, Kitcher P. & Salmon W. C. (eds.),
Minnesota Studies in the Philosophy of Science, Vol. 13:
3–219, University of Minnesota Press, Minneapolis.
Woodward, J. 2014, “Scientic Explanation”, The Stanford
Encyclopedia of Philosophy (Winter 2014 Edition),
Edward N. Zalta(ed.), URL = <http://plato.stanford.edu
archives/win2014/entries/scientic-explanation/>.
nOTES
1 I would like to thank Hans Robin Solberg for extensive revisions and
helpful comments.
2 e DN-model can of course by construed locally. at is, there is
nothing inherently global about the DN-model. For the sake of this pa-
per, I will characterize the DN-model as a global approach, and I think
this is in line with the desiderata for the theory of many of the logical
empiricists. e same applies to the mechanistic theory of explanation.
ere is nothing inherently local about this theory either; it is just that
most, if not all, characterizations of the mechanistic theory are local. To
construe the mechanistic theory as a global approach would certainly
be an interesting thing to do. I imagine one would face some serious
metaphysical obstacles in doing this, but nonetheless it is not something
I will pursue in this paper.
3 At least we can say we are beginning to understand the seasons or why
and how the seasons are cyclical.
4 e most common application of the new mechanistic theory is in the
proximal-physiological disciplines of biology such as molecular biology,
biochemistry, histology, genetics, etc.
5 It should be added here that the antecedent conditions can both be
type and token conditions. at is, the antecedent conditions can be
specied more generally, for example “any container lled with gas”
(type), and “a container c with the volume v, lled with n liters of the
gas g” (token).
6 Functional explanation: Hempel (1965), Nagel (1961); historical ex-
planation: Hempel (1942); and statistical explanation: Hempel (1965).
7 See Salmon (1989) for a thorough discussion on the extensions of
Hempel’s deductive-nomological model.
8 e most devastating counter-example to the DN-model is the appar-
ent insensibility to explanatory asymmetry and explanatory relevance
because of the logical nature of the covering-law model, see Salmon
(1989) and Woodward (2014) for details.
9 Bromberger (1966) is the original creator of this counter-example.
10 From here on the abbreviation ‘(MDC)’ will refer to the Machamer
et al. (2000) article.
11 Many have interpreted the regularity condition in the denition of a
mechanism in (MDC) as a condition of recurrence or repetitive behav-
ior of mechanism; this is not how it is meant by (MDC). It is enough
with predictability (see Craver & Tabery 2015:§2.1.2).
12 See Andersen (2012) for a taxonomy of dierent regularity conditions
in mechanistic explanations.
13 Except, perhaps, the agreement of the abstract pattern of a mecha-
nism and that this pattern conforms well to scientic practice in (at
least) most of the proximal-physiological disciplines of biology.
14 It is important to note here that we nd at the particular or non-
general end of the scale cases of a local theory that covers only one
particular case of scientic explanation. I would argue that, in its origin,
the mechanistic theory of explanation is an example of this case. For
example, Craver (2007) starts with the explanation of the action poten-
tial in neurotransmission.
15 If the DN-model was construed locally and the mechanistic theory of
explanation globally, the roles would of course be reversed. See note 2.
DIFFErEnT aPPrOachES TO ThE PhIlOSOPhy OF ScIEncE
