The debate on scientific explanation in philosophy of science:

an ordering attempt

Abstract: The explanation topic, one of the major subjects in philosophy of science, is characterized by a plurality of
accounts that tried to articulate a general explication of the concept of scientific explanation. In this paper I propose
some general characteristics of the way the subject was approached and place the different particular accounts as
implementing some of these characteristics. This enables one to induce some order in the multitude of approaches
on the subject that were developed over the ime.

Keywords: scientific explanation, approaches on explanation

Richard David-Rus1

Anyone who dares to approach the explanation topic in a straight manner nowadays2 is
confronted with a spectrum of accounts comprising very different proposals that may lead to confusion or
even inhibition. This paper originates in an attempt to overcome this situation and to make the best out of
it. In order to achieve this aim my strategy was to achieve a bird’s-eye view of the multitude of proposals
with the intention to extract some general characteristics, which may account at least partially for the
divergence of approaches. At the same time I’ll be looking for some particular ways to determine certain
modalities of approaching the problem in a pertinent register. This procedure should allow one to evaluate
further the strengths and weaknesses of the different sorts of approaches for a future viable working
agenda in the actual context of philosophy of science.
Taking into account the actual situation in philosophy of science, one might say that there are also
certain advantages to work on the explanation problem in this period. Compared to previous decades the
debate has cooled down considerably, and leading ideas were crystallized for certain positions together
with different variations pertaining to them. By comparison to the mentioned previous decades, there is
now a more generous offer of different perspectives to approach the topic, and the offer extends even
more in regard to the means involved in working out the topic. There is a very rich environment for

1
*researcher, postdoctoral fellow of The Knowledge Based Society Project POSDRU/89/1.5/S/56815.
This paper was made within The Knowledge Based Society Project supported by the Sectorial Operational Program Human
Resources Development (SOP HRD), Financed by the European Social Fund and by the Romanian Government under the
contract POSDRU/89/1.5/S/56815.
2 After over six decades of debate if we are to paraphrase Wesley Salmon’s book, Four Decades of Scientific Explanation

(Pittsburgh: University of Pittsburgh Press, 2006).

research in a good sense, despite possible feelings of “embarrassment” as Newton-Smith describes the
today’s attitude towards the subject.3
Still why is there an embarrassment (or even a “scandal to philosophy of science” as the author
calls it)? It is because, according to Newton-Smith, we still lack a unified account on explanation despite
the existence of many pertinent in-depth studies. The need for such an account is pressing, given the
widely accepted claim that the main task of science is to provide explanations. Besides this, claims in the
realm of any other debates on different philosophical issues (as the one on scientific realism) need to be
substantiated through reference to a concept of explanation.
From an historical perspective it is also a suitable time to draw some morals as to the fate of the
debates in general, and how philosophical agendas evolve and influence other philosophical topics outside
their main scope. In trying to react to Newton-Smith challenge, we’ll acknowledge that promises were
high for a general account on explanation. This also means that we have to keep a close eye on the past.
From a more up-to-date perspective my conviction is that the explanation topic was not a dead-end; on
the contrary, it is full of potentialities for advancing the philosophical insights into the nature of science
and scientific activity. Moreover the aims of the philosophical approach to explanation have to be
adjusted in the light of the results of the debate.
In the following sections I’ll consider certain distinctions that will delineate the directions to
advance a plausible approach. Distinctions have already been made in the literature from which I tried to
cut across different accounts. These were often used to induce classifications over the accounts. A first
way to induce such classifications as we find in different overview articles (e.g., those in encyclopedias)
places the accounts according to a broader view that the authors have adopted as a working possibility. So
we intend to regard the received view adherents or the ordinary-language analysts as approaching the
explanation subject through their specific means. Another well-known classification used by Salmon
identifies three basic conceptions: epistemic, ontic and modal, with the first type subdivided into
inferential, information-theoretic and erotetic kinds4.
In the flow of my argumentation I will make use of some of such previous classifications, which
are widely accepted. But the distinctions I will be drawing do not aim so much to classify the existing
accounts, but to suggest also possible directions that might help us to advance further solutions.
One can talk in terms of the intuitions behind the approaches – although the adequacy of an
appeal to philosophical intuitions is nowadays heavily disputed. Viewing the task of the philosophical

3
In his article on Explanation in In W.H. Newton-Smith, (Ed.) A Companion to Philosophy of Science, (Oxford: Blackwell
Publishers, 2000)
4 In Four Decades of Scientific Explanation, p.118 ff.
analysis in Carnap’s terms5 as an explication of the explanation concept, these intuitions will be reflected
in the choices made for the explicatum and especially in the modalities of its similarity to the
explicandum. One could see a kind of core intuitions comprised in Hempel’s model6, which dominated
the development of the subject during the first decades of the debate. Major subsequent accounts have
developed on the split of the initial core: such as Friedman’s or Kitcher’s account7 exploiting the idea of
explanation as a deductive argument or Salmon’s focus on causality as explicatum. The fourth decade of
the debate, in Salmon’s own words one of the ‘radical innovations’, exhibits accounts totally and
mutually exclusive in their assumed intuitions. Of course different classification as mentioned above,
have to reflect the different intuitions.
Let us pursue with the distinctions:

Global versus local approaches

The first distinction I wish to point out is that between a local and a global kind of approach. This
distinction is based on the way in which two different kinds of considerations (global and local) are to be
seen as determinant for the scientific explanation and are therefore to be used in a conception of
explanation. A kind of global-type of approach will consequently be one in which the global
considerations are viewed as central. This does not mean that only explanations, which make direct
appeal to the most general principles are proper explanations, but that the right criteria that determine an
explanation are to be drawn properly from considerations at this level. Correspondingly, the same holds
for a local view.
Let us see in greater detail how this distinction could be made more explicit and by what other
means can we express it. We may look first at the uses encountered in scientific practice. Therefore we
will usually say that the scientists solved a problem by global considerations if they make appeal to some
general stipulations – e.g., such principles as those of conservation or invariance in physics – in
contradistinction to the situation in which the contextual information extracted from the particular case at
hand provides the solution. But it is not this sense that is intended by the philosophers who make explicit
use of the terms. In the explanation debate, the terms acquired a particular meaning, while Friedman
advocated a global approach to explanation. By drawing attention to the global aspects, he meant the

5
R. Carnap, Logical Foundations of Probability. (Chicago: University of Chicago Press. 1962)
6 Carl Hempel, “Aspects of Scientific Explanation,” in his Aspects of Scientific Explanation and other Essays in the Philosophy
of Science (New York: Free Press, 1965), 331-496.
7 As presented for example in Michael Friedman, “Explanation and Scientific Understanding,” Journal of Philosophy 71

(1974): 5-19, and in Philip Kitcher, “Explanatory Unification,” Philosophy of Science 48 (1981): 507-531
aspects regarding the relation of the phenomenon to be explained with “the total set of accepted
phenomena”.
To shortly recall Friedman’s account8. He begins by stating that the usual explanadum-
phenomena are expressed in the form of regularities constituting empirical generalizations or
phenomenological or more special laws. They are explained through some other more abstract and
general laws as is the case of Kepler’s and Galilei’s laws on motion that are explained by Newton’s laws.
The very essence of explanation in Friedman’s conception is basically the reduction of a plurality of
different law-like generalizations, which were previously accepted on independent grounds, to a more
general law. The explanatory relation is ultimately expressed by Friedman only through set-theoretical
relations; the two conditions providing the explanatory relation between two sentences being that the
explanadum should belong to the explanans’ consequences and the explanans should ‘reduce’ the
explanadum’s consequences. The notion of reduction from the last condition is expressed through an
inequality between the cardinalities of the following sets: the K-partitions of the set reduced and its
augmentation with the reducing sentence. This set-theoretical reconstruction of the explanatory relation
can be viewed as making explicit the global aspect that determines the set of genuine explanations in
science.
Kitcher’s account9 provides us with another example of a global type of approach. As Friedman’s
one, it also explicates scientific explanation by connecting it to the systematization of knowledge. But in
comparison to Friedman’s account, unification is seen here to be realized through the repetitive
application of a number of reasoning patterns to derive descriptions of different phenomena. The
systematization of the corpus of knowledge that is generated by the set of argumentative patterns that best
unifies the corpus will constitute the explanatory corpus, i.e., the set of explanatory arguments, which any
valid explanation has to instantiate. A set of patterns will be qualified as most unifying by comparison to
others, if it generates the greatest number of conclusions with a few stringent arguments. This way the
global considerations related to sets of patterns determine primarily the explanatory virtue of the
arguments.
In the view of the abovementioned authors, the global approach offers some major advantages for
solving some difficult issues raised by the explanation problem. For Friedman the global approach gives
an answer to the old argument against the existence of any genuine scientific explanation, argument that
invokes the unexplanatoriness of the premises assumed. Besides, both authors see such an approach
having the advantage to make clear the connection of explanation with understanding. A further benefit
resides in the fact that it recuperates this way the objectivity of scientific understanding.

8
As exposed in Friedman, “Explanation and Scientific Understanding,”
9 As exposed in Kitcher, “Explanatory Unification,”
 The more recent approaches, those developed by Bartelborth10 and Schurz11 would make for good
examples of global type of approaches. These authors pick on a sense directly from Kitcher’s basic
unification idea trying to articulate it in a more rigorous way. They do this either by adopting a well-
articulated conception of scientific knowledge - the structuralist conception on science in Bartelborth’s
case - or by carefully formalizing the process of “assimilation” of the new phenomenon to be explained in
the corpus of knowledge.
Examples of the local type of approach are those of Salmon12 and van Fraassen13. In van
Fraassen’s case the emphasis on the contextuality of explanation reflects directly the local character of the
approach. However this is not the way the abovementioned authors will use the term. For Kitcher and
Friedman local will most probably characterize an approach that does not take into consideration the
systematization aspects of scientific knowledge. As for Salmon, the attribute local is used for a type of
explanation (not of an approach) that shows “how particular occurrences come about.” He is viewing his
causal/mechanistic explanation as a typical exemplification for this case. If an approach focusing on this
type of explanation could be qualified as local, it is not clear from his discussion. Salmon is referring to
types of explanations not types of approaches as Friedman. Of course for Friedman and Kitcher this type
of explanation that focuses on singular occurrences was to be rendered only as a particular case, a limiting
one in their approaches. But the issue has not been really addressed since their analysis builds on the
assumption of explanandum being expressed through regularities.
In contradistinction, Salmon takes as paradigmatic for his analysis the explanation of singular
events; in his view the identification of a cause implies taking into account the particular situation, in
which the cause is acting. In his words: explanation is about opening the black boxes of nature and
revealing the hidden inner workings. The very particular configuration of the situation to be explained is
reflected in the network of particular causes at work that have to be captured by a valid explanation. A
further specification that we should be aware of is that it is not that global considerations could not play a
role in determining the explanatory relation. None of the two authors, Salmon and van Fraassen, will
probably deny this fact. It is rather the fact that the global factors are not taken as the only considerations
relevant in determining the explanatory quality; local factors are the ones that play the decisive role. For
both authors, considerations related to unification or reduction of laws could guide the building of
explanations; but if they are to be used and how they influence the building process, is ultimately
determined by local factors.

10
Thomas Bartelborth, “Explanatory Unification,” Synthese 130 (2002): 91-107.
11 Gerhard Schurz, “Explanation as Unification,” Synthese 120 (1999): 95-114.
12 Salmon, W. Scientific Explanation and the Causal Structure of the World, (Princeton: Princeton University Press:1984).
13 His pragmatic approach is presented in the fifth chapter of his book The Scientific Image (New York: Oxford University Press,

1980): 97-157.
 To summarize: Salmon casts the difference global-local as characterizing two types of
explanation (expl1 and expl2); this could be taken also to draw the difference between two types of
approaches. But as we have seen, the meanings are not entirely identical when qualifying approaches or
explanations. There is a difference in the way I’m using the terms local and global that takes Friedman’s
use as a starting point and becomes apparent through the fact that Salmon will not qualify van Fraassen’s
approach as dealing with local explanation. He regards it as rather neutral, since it could render either of
the two types of explanations when additional constrains are attached.14 By taking van Fraassen’s
approach as local, I plead implicitly for a broader meaning of the term local.
While keeping in mind these exemplifications I’ll try to render more explicit the sense in which I
read the distinction local-global by illuminating different facets through which it could be reflected.
A first plausible way that offers itself directly is to draw a separation in reference to the scientific
entities invoked in the considerations that determine the explanatory relations. In the global type of
approaches the authors are concerned with ‘bigger’ scientific structures. They concern primarily the
aggregation of scientific entities, sets of statements or laws, or patterns, or models, making out the
theories or corpuses of knowledge. Scientific knowledge is considered at the level of the overall structure
and the way this organization is inducing explanatory relations. For Friedman it’s about the relation of the
explanandum with the entire set of the accepted phenomena; for Kitcher it is the membership of the
pattern in the explanatory corpus E(K). Both accounts make reference to quantitative references invoking
sets of statements or patterns and a number of axioms to which the statements are reduced or simply the
number of patterns in a corpus. We can also claim that in global approaches there are usually proprieties
emerging at the level of the bigger entities considered (special proprieties as unification or coherence) that
bear directly on the explanatory power of these entities. I’ll return to this point later in my thesis.
The local approaches do not imply such references; the structure of scientific knowledge is not
particularly decisive in establishing the explanatory relation. For example, van Fraassen’s semantic
conception on theories is not brought to defend directly his account on explanation.
The classical Hempelian approach would have a somehow less clear positioning along this way of
viewing the distinction or one that makes reference to laws as the main determinants of explanation. A
natural move is to take theories as playing the central role in a global approach. As sets of statements or
models systematized in a certain way, they can deliver the necessary determinant for the explanatory
virtues. We could regard the universality of the laws as a sort of global feature in analogy with Friedman-
Kitcher approach. And following the loose analogy we could search for a quantitative measure to render
or to capture a feature, by pointing to the number of instantiations (or the degree of abstraction) of the

14
The point is made in the last part of the book Four Decades of Sceintific Explanation, when Salmon seeks a compromise
between the ontic approach (developed in his account) and the epistemic one (developed by Kitcher).
law. Obviously we lack here the central idea – the systematization of the corpus of knowledge and the
way it is achieved. The subsumption of facts under the laws or of more particular laws under more
general ones through deductive arguments alone does not tell us much enough about the way we
systematize knowledge.
There is a certain affinity between the DN-model and the global types of approaches. These
authors are considered to develop their account by continuing the main lines of the Hempel’s and the
received view agenda. The idea of explanation as an argument is central to both Friedman and Kitcher;
but the way they make use of it, by integrating it with the idea of knowledge systematization, underlines
the radical difference from Hempel’s model. There is also an affinity of the DN-model with Salmon’s
way of understanding explanation. Explanation1, the local type in Salmon’s sense, directly captures one
of Hempel’s intended explanations: the causal type. Therefore Hempel’s model can be viewed as
containing in nuce the main ideas of both approaches without being reducible to either of them.
I mentioned briefly that van Fraassen’s account can be regarded as a good example of a local type
approach. The main reason for this would be the role pragmatics plays in determining the explanatoriness
of a relation. The distinction pragmatics-non-pragmatics should, in my view, parallel in a good sense the
distinction global-local. Pragmatics should play a central role in a local approach allowing in this way a
better recuperation of the process of explanation as a part of the scientific process of inquiry. This way
Salmon’s meaning of the local is covered only partially. Though the causal-mechanism is not per se
subject to pragmatic factors, picking the right causal network from the ideal explanatory text in Railton’s
view, which Salmon considers to complete the causalist view, is pragmatics dependent. So identifying the
right causal mechanism and establishing in this way the adequacy of the explanation is influenced by the
contextual factors of the investigation.
Another distinction that may be seen as recasting the wanted intuition behind the global-local
distinction is the one used by Kitcher, i.e., between top-down and bottom up approaches. Salmon sees it
as cutting along his explanation1 and explanation2. The bottom-up explanation appeals to the microstate
of the facts targeted by the explanation, meanwhile the top-down appeals to more general principles and
law-like generalizations. We encounter here a partial overlapping with the sense of global mentioned at
the beginning of this section as being used by the scientists when drawing conclusions on global
considerations. In an effort to synthesize and render the approaches complementary, Salmon sees both
strategies: top-down and bottom–up, as being only different ways of reading the ideal explanatory text.
Pragmatics determines in his view what way is to be picked in a particular situation. In as far as I am
concerned, I cannot see everyone happy with this kind of compromise; there are passages of not
unimportant details that are suppressed in order to carry out the overall match. One of such problem that
becomes acute in this situation is the modality we should conceive how the ideal explanatory text is
actually read in a global approach of explanation. This is in fact similar to the problem encountered by
Kitcher when trying to account for causal explanation in his own conception.
In addition to the different ways of casting the meaning of the local-global distinction, I would
like to make reference to another sense of the localism. The local character of the approach as specified
above should also fall under this more general sense. The best statement of this sense can be found in
Huggett’s paper “Local philosophies of science”15. He calls local the approaches in philosophy of science
if “philosophical problems are to be found and treated using the resources of more-or-less delineated
scientific programs and not by trying to make science fit some prior vision”. This sense should qualify the
ones that I tried to pin down through the different distinctions. The modalities of casting the local
character through the proposed distinctions are in a natural way qualified through Huggett’s localism.
This last claim is justified if we take into account the fact that the search for local considerations, which
will determine the explanatory relations, is enhanced in the frame of a particular scientific program. On
the other side global considerations tend usually to overrun the particular programs.

Static versus dynamic approaches

Another distinction that I would like to take into consideration is the one between static and a
dynamic approach. It is not to be regarded as a distinction that can be drawn between different existing
accounts on explanation. The existing accounts are more of the first sort. The distinction refers to a more
general style of viewing the logico-philosophical reconstruction of scientific activity. The distinction is
primarily between approaches that give a place, i.e., take more into consideration aspects of temporal
evolution of the scientific structures versus the ones that ignore them. But how would temporal aspects be
thought in the case of the explanation problem? Further on I will try to delineate the ways we can do this.
The static view is naturally seen as a legacy of the received view. It is not unusual to qualify as
static not only the specific problem of explanation but the more general conception of science of the
logical-empiricist view. As known, the historical orientation in the philosophy of science embodied the
reaction against this aspect of the received view. Under the received view, the rejection of the pragmatics
is the background that justifies the neglect of the temporal aspects. This should fall into the areas of
psychological and sociological studies.
In the evolution of the problem of scientific explanation there are no obvious critical reactions to
point to this aspect or rather to pick on it as the central point of their critique. We find scattered remarks

15
Nick Huggett, “Local Philosophies of Science,” Philosophy of Science 67 (2000): 128-137.
and warnings only later in the debate (during the fifth decade) at such authors as Bunzl 16 or Sintonen17.
But someone could react by pointing to Kicher’s account as one that addresses and integrates the temporal
aspects of scientific knowledge into the explanation problem. This point seems right: Kitcher’s
conception can account for different corpuses of knowledge from different historical periods as
determining the explanations accepted as valid in those periods. The account captures the dynamics at the
macro level of scientific activity; a fact that could be seen as constrained directly by its global character.
In my opinion, the sense of dynamics recuperated by Kitcher’s account is only a specific one among
many, and not the relevant one for an adequate solution. A proper dynamic account should unfold at a
local level. To characterize it in a more general way, the main interest would be in describing how
elements of scientific knowledge are modified or new elements are constructed in the process of
providing an explanation for a phenomenon.
There are more ways available to unpack the above view. For example, Kitcher provides us
criteria for comparing and selecting between corpuses of knowledge but he does not provide any clue as
to how an explanation pattern is built. The solution in sight would be to provide rules as to how different
elements of the pattern evolve, i.e., are chosen, modified or dropped in the course of searching for an
explanation of a phenomenon. This will have of course to be integrated with the macro-constraints at the
level of the corpuses. The task does not seem to be obvious if it is not even unrealizable. A relativization
of the corpuses to more delimited fields of science will probably constitute a beneficial move.
In fact the dynamical aspects can be articulated in more diverse forms and Kitcher’s one is only
one among others. Variants that should be addressed are the ones that regard the way background
knowledge is being modified when building an explanation. Bunzl, pointing to this issue, expresses it by
requiring such accounts to be about “explanations as they arise in the context of building scientific
theories”18. I’ll tend to see this characterization extended to comprise also other types of scientific units,
as for example, models. The dynamical aspects of explanation will become obvious if approached in the
context of scientific modeling. Of course this is a particular angle of dealing with the problem, but a
promising one as I’ll argue further in my work.
Addressing the specific dynamics of the explanatory processes, one has to take into consideration
the relations between different sorts of scientific representation. Aspects of the interplay between the
representations should reflect the way scientists are looking for explanations, and the way explanations
arise and are modified. Such types of accounts were in fact already developed by some authors as

16 Bunzl, M. The Context of Explanation. (Dordrecht: Kluwer Academic Publisher: 1993)

17 Sintonen, M. “Why Questions, and why just Why-Questions?”, Synthese, 120(1), (1999): 125-135.
18 Bunzel, The Context of Explanation.
Hughes19 or Frigg and Hartmann20, with reference to more specific registers of explanation i.e.,
explanation through models. I’ll come to discuss these accounts and the virtues they exhibit in dealing
with the explanation problem.
Making room for pragmatics should go hand in hand with the dynamic view advocated here.
Contextual factors viewed at the local level shape in an essential way the process of scientific inquiry and
the search for explanations. In a global approach pragmatics could be discarded to a certain degree.
Kitcher’s account is a pragmatics-free one but the modalities of comparing different corpuses of
knowledge are not as much pragmatics-free as Kitcher wants to see them. Nevertheless it can be argued
that the dynamics captured in Kitcher’s account refers primarily to the units of scientific knowledge and
not to explanations. Of course this claim has to be pursued in a more detailed way.
Explanation plays a central role in scientific inquiry and in particular in scientific discovery; it
drives scientific inquiry and reflects in this way the process of knowledge expansion. This important
aspect can be properly captured only in a dynamical account on scientific explanation. I’ll touch also in a
more detailed way on this aspect further in my work.
One may also rightly raise the question of the means that we could use to capture the dynamics.
The dynamical aspects at the local level would be better exposed if we use specific frameworks. One such
framework that proves to be suitable for this is the interrogative view on scientific inquiry. This will
provide quite an efficient modality of modeling the dynamics of explanatory practices.21

Explanation as application versus explanation as construction

After a quick perusal of the previous discussion, I intend to advance a further distinction that
might shed some light from another angle on the separation between the two types of views. The basic
fact of one of the views can be put this way: it tends to conceive, to over-read explanation as application
of a law or a theory (or other scientific entity taken as a central unit of scientific knowledge). I will call
this sort of explanation, explanation as an application. The specificity of this approach is to be seen in the
assumed (more or less) understanding of the application. First we should notice that there is no
specialized, well-developed area of research in philosophy of science focused on the topic of application

19Hughes R.I.G., “Models and Representation”. Philosophy of Science, 64, Supplement. Proceedings of the 1996 Biennial
Meetings of the Philosophy of Science Association, (1997): 325-336.
20 As presented in their communication “Explanation through models” at the conference “Philosophical Perspectives on Scientific
Understanding”, Amsterdam, august 25-27, 2005.
21 I addressed in detail this point in my doctoral thesis Explanation and Understanding through Scientific Models. Perspectives

for a New Approach to Scientific Explanation (PhD diss., University of Munich, 2009).
only. The term application is used usually in an informal way. We could say in general that an application
of a law is an instantiation of it or that an application of a theory is a realization of it in some concrete
situation – spelled out it would mean that some of the laws expressed in its sentences are instantiated in
the situation under investigation. It could also appear that there are no further real issues related to this
topic. Nevertheless many of the classical authors have in mind a sort of situation analogical to the logical-
mathematical view. As in mathematical logic, an application will be thought along the lines of a sort of
plugging values into a formula or formal schema. In the empirical sciences the situation is quite different
from the mathematical one, and by using (even if only in a regulative way) the simplified version we
distort the situation and situate us on an unfruitful track.
We could regard Hempel’s and Kitcher’s accounts as implementing directly this simplified
version. In case of Kitcher’s account, we can interpret his detailed description of the patterns (through
filling instructions, classification rules etc.) as trying to deal with the limitation of the explanation as
application view. I think we can find this attitude of approaching explanation as an application as a more
constant and tacit assumption of the accounts in the classical positions of the debate. This last point is
further illuminated if we look to the connection with another kind of distinction (to be discussed in more
detail in the next section) – the one between theory-driven and non-theory driven accounts.
While we assume a broader view of application, we may regard explanatory processes as related
to the notion of application in several ways. The important thing to stress and to take into consideration is
the fact that applications in empirical sciences involve a series of scientific activities, which could be seen
as parts of the application. Certainly it is more proper to see application as a construction process. There
are several distinct activities involved in the application process such as decisions of what representation
shall we use or what approximation and idealization should we build and adopt. From the perspective of
the logical empiricist conception they will fall mostly into the realm of pragmatics and this could explain
their neglect in the classical accounts of the debate.
How to characterize application is in my opinion not a one shot business. Making an attempt to
delineate briefly some intended meaning I’ll probably say that by using the notion of ‘construction’ I want
to point to the different scientific activities involved in the process of explanation. Explanation as
‘construction’ takes seriously the idea that representation of phenomena must be constructed and it is
through them that we may get an explanation. Such processes as idealization and approximation are part
of this construction and of the explanatory practice.
 The explanation as application bears also an analogy to what Cartwright in her critique of the
received view on theories calls “the vending machine” view.22 Theories are thought as one will “fed them
an input in certain prescribed forms for the desired output” and after a while “it drops out the sought-for
representation […] fully formed.”
If there are chances to explicate the application concept in a more rigorous way, I think that the
precise meaning should be picked under the local approaches on explanation as described in the previous
paragraphs. But it is also highly plausible that we should regard it rather as an umbrella concept, which
spans over different scientific activities, the explanatory activity being one among many. Referring to it
here, my intention is to posit it as an opposite to the explanation as construction, in order to suggest the
general direction of advancing the inquiry.

Explanation in a theory-driven perspective versus in a non-theory driven one

The 60s and 70s have witnessed strong debates concerning the nature of a scientific theory. This
situation is to be seen also against the background of the reaction to the rather narrow logico-positivistic
proposal regarding this topic. The rise of the semantic view offered a new alternative solution to the
problem. But a more radical consequence of this reaction brought about distrust in the central role played
by the notion of theory in the analysis of scientific enterprise. The concentration on getting in the first line
a well articulated solution for the nature of a scientific theory, leaving the rest to follow from this or to be
constructed around the notion of scientific theory, was gradually abandoned. Some aspects, which were
thought as being secondary for the analysis of the scientific activity, gained much more importance; such
as the experimental or the modeling activity and their products. Consequently, different topics, among
them arguably explanation, gained (more or less) new valences in this new context.
The explanation topic rose at the status of a major subject in the philosophy of science in the
heydays of the theory-oriented philosophy of science. Therefore it bears some of the legacy of that
context. This could be seen also as one of the reasons why it is rejected in more recent philosophical
agendas that assume a radical departure from any received view influence. Nevertheless we could read
out influences of the theory-centric but also signs from the opposite attitude in today’s approaches on
explanation.

22Cartwright, N. “Models and the Limits of Theory: Quantum Hamiltonians and the BCS Model of Superconductivity”, in M.
Morgan & M. Morrison (Eds.), Models as Mediators. Perspectives on Natural and Social Science (Cambridge: Cambridge
University Press: 1999): 241-281.
 In the last and a half-decade one of the most active moves along this separation was undertaken
by focusing on the role of models in scientific knowledge. Authors such as M. Morrison, Mary Morgan,
Stephan Hartmann or Nancy Cartwright,23 are among the best known exponents of the modeling-oriented
approach. In an effort to center the philosophical investigation on scientific models, they criticized the
theory-driven view on models. What they mean by such a view on models is that models are seen as
entities derived from theories and that their building, status and functioning depend in an essential way on
theories. The authors engage in arguing for the autonomy of models, as for example in Morrison’s way of
making the point where models are autonomous agents in the production of scientific knowledge.
In an analogical way we can also argue for a more relaxed non-theory-centric kind of approach on
the explanation topic. Even more, we can take the position of considering models as a solution for the
status of explanation-bearers,24 i.e., to consider the articulation of the accounts on explanation by making
reference to models rather than theories or laws.
From an historical perspective one can distinguish between approaches on explanation that
manifest tendencies towards a theory-driven view in a more or less obvious way. So, one will regard
Hempel’s account as theory-centric, since the concept of a theory plays a central role in its articulation.
Kitcher’s account integrates a theory-centric view too by advancing a proposal as to the nature of
scientific theories as ‘families of patterns’. In general it would appear as a normal tendency that global
approaches should be more theory-centric biased. But different global accounts differ also in the role
theories play in the articulation of the account on explanation. Some appeal in a straight way to the
structure of theories as Kitcher or Bartelborth. But reference to more comprising units of knowledge such
as corpuses of knowledge can bypass the appeal to theories, as in Schurz’ case. Such a corpus of
knowledge can be interpreted as a theory, but it is meant to accommodate various other forms of
knowledge entities. In this sense, Schurz’ account evades a strict theory-centric approach.
In order to draw to an end my investigation I’ll restate an implicit observation. As I mentioned
already earlier, the proposed distinctions are not clear-cut. They are in a sense regulative, sketching some
broad directions along which one can comparatively refer when evaluating the different accounts. There
are obvious affinities between the different distinctions in the sense that by adopting some option of a
distinction, a specific option for the other distinction becomes more plausible.

Conclusion
In concluding, I want to restate the main intention behind the above proposals. I think the
analysis made clear that despite the variety of the approaches on the subject, we can detect behind them

23 A sample of works defining this orientation can be found in the volume edited by Mary S. Morgan and Margaret Morrison
Models as Mediators. Perspectives on Natural and Social Science, (Cambridge:Cambridge University Press, 1999).
24 As I have argued in my doctoral thesis.
some general suppositions that guided their articulation. By making them clear one can further place the
limitations of the known accounts in a larger context. This fact bears further on the way one would try to
approach the topic of explanation nowadays. As I argued in my doctoral thesis a plausible option will be a
local, dynamic and non-theory driven approach, that I saw as being possible implemented in a modelistic
frame. This option opens I think a fertile new perspective for approaching the respectable topic of
scientific explanation nowadays.