EXPLA NATORY UNIFICATION
A ND SCIENTIFIC UNDER STA NDING
?Zcc^[ZgL^ahdcBjac^m
1. >cigdYjXi^dc
;
]^h paper represents a response to the criticisms made by Eric Barnes in
“Explanatory Uniication and the Problem of Asymmetry” and “Explanatory Uniication and Scientiic Understanding” against the thesis of Explanatory Uniication.1 This paper responds to Barnes’ two main criticisms, that
of derivational skepticism and causal asymmetry, and successfully refutes his
objections. This paper also defends the plausibility of the uniicationist account of scientiic explanation because of its ability to render coherent the
notion of scientiic understanding. In doing so, this paper focuses on Michael
Friedman’s account of Explanatory Uniication. This is chosen because of his
extensive treatment of scientiic understanding. Other (perhaps technically
more sound) accounts of uniication, especially Philip Kitcher’s detailed formulation, will also be utilized in responding to Barnes’ objections.2
I begin in Section One with a positive defense of Explanatory Uniication,
elucidating Friedman’s account of uniication, paying particular attention to
the notion of scientiic understanding. In Sections Two and Three, I defend
Explanatory Uniication against the charges of Barnes on the issues of derivational skepticism and causal asymmetry. In doing so, I aim to clarify the basic
motivations of the Explanatory Uniicationist. The inal section, Section Four,
presents some additional issues relevant to scientiic explanation. I conclude
that until these other questions are resolved, an ultimate evaluation of the
success of Explanatory Uniication as a comprehensive theory of explanation
cannot be made. However, although Explanatory Uniication is susceptible to
* University of Massachusetts Dartmouth, 285 Old Westport Road, North Dartmouth,
bV 02747. E-mail: jmulnix@umassd.edu
1 E. 7VgcZh, Explanatory Uniication and the Problem of Asymmetry, (ZjeV hereafter) «Philosophy of Science», 59 (1992), pp. 558-571 and E. 7VgcZh, Explanatory Uniication and Scientiic Understanding, (Zjhj hereafter), Proceedings of the Biennial Meeting of the Philosophy of
Science Association, 1 (1992), pp. 3-12.
2 For some recent alternative defenses of Explanatory Uniication see: G. HX]jgo, Explanation as Uniication, «Synthese», 120 (1999), pp. 95-114; and R. HX]lZYZg, A Defense of a
Uniicationist Theory of Explanation, «Foundations of Science», 10 (2005), pp. 421-435.
®VXiVe]^adhde]^XV¯Æ^^!'%!'%&&Çee#&,*"&.)
&,+
_Zcc^[Zgl^ahdcbjac^m
certain diiculties, it does answer the objections of derivational skepticism and
causal asymmetry, and thus we should continue to search for better formulations of the uniicationist approach to scientiic explanation.
'#HX^Zci^[^XJcYZghiVcY^c\
VcY;g^ZYbVc » h:meaVcVidgnJc^[^XVi^dc
Wesley Salmon maintains that the purpose of science is not just to describe
the world, but also to provide understanding, comprehension, and enlightenment. Furthermore, science provides these things through scientiic explanations. Thus, scientiic explanation is fundamentally tied to understanding of
some sort. Given this, Salmon believes the questions to which we should be
seeking answers are: «If explanation does involve something over and above
mere description, just what sort of thing is it?» and «What, over and above descriptive knowledge of the world, is required in order to achieve understanding?».3
It is appropriate, then, that central to Michael Friedman’s Explanatory Uniication doctrine is the notion of ‘scientiic understanding’. For Friedman also
believes the principal problem for any theory of scientiic explanation is the
following: «What is the relation between phenomena in virtue of which one
phenomenon can constitute an explanation of another, and what is it about
this relation that gives understanding of the explained phenomenon?”4 Resolving this dilemma, however, also requires discovering what it is for a phenomenon to be ‘scientiically understandable’. And, according to Friedman,
“We can ind out what scientiic understanding consists in only by inding
out what scientiic explanation is and vice versa».5 He further asserts that the
concept of scientiic understanding is vague, but that what is explained is a
general regularity or pattern of behavior (a law) and a regularity is explained
by relating (reducing) it to another regularity.
Friedman’s project, then, is to provide accounts of traditional theories of
explanation and show how they are insuicient to account for understanding.
He also outlines the general properties that the concept of scientiic understanding should have, and he proposes an account of scientiic explanation
that possesses these desirable properties. In addition, Friedman discriminates
among two primary groups of scientiic explanation. The irst group represents the traditional received view and includes Hempel and Nagel. The
second group consists in the causal/mechanistic/pragmatic approaches and
includes Scriven, Toulmin, and Dray. Friedman claims that neither of these
3 W. HVabdc, Scientiic Explanation and the Causal Structure of the World, Princeton UP,
Princeton 1984, p. 9.
4 M. ;g^ZYbVc, Explanation and Scientiic Understanding, «Journal of Philosophy», 71
(1974) 5-19, p. 6.
5 Ibidem.
ZmeaVcVidgnjc^[^XVi^dcVcYhX^Zci^[^XjcYZghiVcY^c\
&,,
groups suiciently explain both scientiic explanation and scientiic understanding:
«Some philosophers, like Hempel and Nagel, have relatively precise proposals as to
the nature of the explanation relation, but relatively little to say about the connection
between their proposals and scientiic understanding, i.e., about what it is about the
relation they propose that gives us understanding of the world. Other philosophers,
like Toulmin, Scriven, and Dray, have a lot to say about understanding, but relatively
vague ideas about just what relation it is that produces this understanding».6
Philip Kitcher agrees with Friedman that the nature of scientiic explanation
should include understanding: «A theory of explanation should show us how
scientiic explanation advances our understanding».7 Moreover, Kitcher claims
that scientiic understanding is closely tied to uniication:
«My view of explanation as uniication suggests how scientiic explanation yields understanding. By using a few patterns of argument in the derivation of many beliefs
we minimize the number of types of premises we must take as underived. That is,
we reduce, in so far as possible, the number of types of facts we must accept as brute.
Hence we can endorse something close to Friedman’s view of the merits of explanatory uniication».8
Many, however, disagree with Friedman’s thesis that scientiic understanding
is important to scientiic explanation.9 For Carl Hempel, scientiic explanation is not intimately tied with scientiic understanding. Hempel declares that,
«such expressions as ‘realm of understanding’ and ‘comprehensible’ do not
belong to the vocabulary of logic, for they refer to the psychological or pragmatic aspects of explanation».10 Furthermore, «explanation in this pragmatic
6 Ibidem.
7 P. @^iX]Zg, Explanatory Uniication, «Philosophy of Science», 48 (1981), pp. 507-531, p.
508. See H. YZGZ\i and D. 9^Z`h, A Contextual Approach to Scientiic Understanding, «Synthese», 144 (2005), pp. 137-170 for a nice discussion of the notion of scientiic understanding
as it inds expression in diferent theories of scientiic explanation.
8 P. @^iX]Zg, Explanatory Uniication, cit., pp. 529-530.
9 Bas van Fraassen, for instance, is opposed to the claim there is a necessary connection
between scientiic explanation and understanding. «The real focus of disagreement may
be the relation between explanation and understanding. …All I have to say here is that our
understanding can sometimes be increased otherwise than by receiving explanation and
that we may sometimes be in a position (in view of our interests, background information,
and beliefs) to receive an explanation of one case and not of another even though the total
scientiic information, by itself, harbors no such asymmetry» (B. kVc;gVVhhZc, Salmon on
Explanation, «Journal of Philosophy», 82 (1985), pp. 639-651, p. 642).
10 C. =ZbeZa, Aspects of Scientiic Explanation and Other Essays in the Philosophy of Science,
Free Press, New York 1965, p. 413. Hempel is very clear in distinguishing scientiic understanding from psychological or empathic understanding: «It is important to distinguish
here understanding in the psychological sense of feeling of empathic familiarity from un-
&,-
_Zcc^[Zgl^ahdcbjac^m
sense is thus a relative notion: something can be signiicantly said to constitute an explanation in this sense only for this or that individual».11
Hempel, instead, provides an account of scientiic explanation that regards
understanding as predictive power. He says, «the [D-N] argument shows that,
given the particular circumstances and the laws in question, the occurrence of
the phenomenon was to be expected; and it is in this sense that the explanation
enables us to understand why the phenomenon occurred».12 However, according to Friedman, Hempel’s claim amounts to the idea that knowing “the particular circumstances and laws in question” is just to have “rational grounds”
for expecting the explained phenomenon to occur. But, to have grounds for
rationally expecting some phenomenon is not the same as to understand it.13
For example, a barometer may provide rational grounds for expecting a storm
to occur. But, most agree a barometer does not help us to understand why the
storm is about to occur. Rather, more information is necessary to provide understanding of the occurrence of the storm.
Friedman is willing to concede that the D-N model has the many advantages. For example, the D-N model provides entailment, which is a necessary
condition for explanation. Furthermore, it retains the objective character of
explanation, not depending on the arbitrary attitudes of the scientist or the
historical period. However, Friedman reminds us, the D-N model must ultimately be replaced because it has not succeeded in saying what it is about the
explanation relation that provides understanding of the world. Any adequate
theory of explanation must somehow connect explanation and understanding, and the D-N model is deicient in this respect.
Another issue central to scientiic understanding and explanation is the notion of ‘familiarity’. Hempel rejects the idea that scientiic explanation resides
in a reduction of phenomena to things familiar: «the frequent insistence that
explanation means the reduction of something unfamiliar to ideas or experiences already familiar to us is indeed misleading. For while some scientiic
explanations do have this psychological efect, it is by no means universal».14
derstanding in the theoretical, or cognitive, sense of exhibiting the phenomenon to be explained as a special case of some general regularity» (ibidem, pp. 256-67). See also J. Igdji,
Scientiic Explanation and the Sense of Understanding, «Philosophy of Science», 69 (2002), pp.
212-233, and The Psychology of Scientiic Explanation, «Philosophy Compass» 2 (2007), pp. 564591.
11 C. =ZbeZa, Aspects of Scientiic Explanation and Other Essays in the Philosophy of Science,
cit., p. 426.
12 Ibidem, p. 337.
13 M. ;g^ZYbVc, o.c., p. 8.
14 C. =ZbeZa, Aspects of Scientiic Explanation and Other Essays in the Philosophy of Science,
cit., p. 257. See also p. 329. On the other side, however, there are those, such as Michael Scriven and Gerard Schurz, who argue that an explanation of something unfamiliar in terms of
something else unfamiliar does not constitute a suicient explanation. For example, Bohr’s
stability postulates for electrons in an atom were able to successfully predict the hydrogen
ZmeaVcVidgnjc^[^XVi^dcVcYhX^Zci^[^XjcYZghiVcY^c\ &,.
In spite of their disagreement over the importance of understanding for explanation, Friedman agrees with Hempel that understanding does not reside
in familiarity, and thus, familiarity is not essential to explanation. «For, being
familiar, just like being expected, is not at all the same thing as being understood. We are all familiar with the behavior of household appliances… but
how many of us understand why they behave the way they do?»15 In rejecting
the claim that scientiic explanation must reduce phenomena to the familiar,
Friedman argues that there are many cases of genuine scientiic explanation
whose explanans are more unfamiliar than the explanandum. For example,
science explains the refraction of light, something familiar, by appealing to
electromagnetic waves, something relatively unfamiliar. Friedman further
supports his case by pointing out that if explanations had to reduce phenomena to the familiar, most of the explanations ofered by contemporary physics
could not possibly explain.
However, Friedman concedes that in many of these scientiic explanations,
no gain in understanding is produced. For instance, in the paradigm case of
using the kinetic theory to explain the Boyle-Charles law of gases, this scientiic explanation does not amount to a gain in understanding since it “merely
replaces one brute fact with another.” But, this concession does not rule out
the necessity of understanding for explanation, according to Friedman. Rather, this allows Friedman to formulate a distinction between ‘local’ and ‘global’
understanding. In other words, Friedman believes that scientiic explanation
involves an increase in global understanding even if, locally, one unfamiliar
fact is replaced with another.16 Thus, Friedman’s position relies upon a subtle
distinction between types of understanding. The idea here is that accounts of
scientiic explanation that require reducing unfamiliar phenomena to familiar
spectrum, but because they were inconsistent with classical mechanics, they were themselves in need of explanation. Thus, most scientists did not believe Bohr’s postulates could
be an explanation for atomic behavior. Hence, even if the unfamiliar phenomena are able
to successfully predict, and even causally explain, the explanandum, they cannot provide
explanation. According to Scriven, explanation is fundamentally tied to both familiarity
and understanding, because understanding resides in familiarity (M. HXg^kZc, Truisms as the
Grounds for Historical Explanation, in P. <VgY^cZg (ed.), Theories of History, Free Press, Glencoe 1959). Moreover, Schurz believes there is a semantic connection between scientiic explanation and scientiic understanding directly related to familiarity. His position asserts
that something cannot be an adequate explanation even if what does the explaining is not
itself completely understood or familiar. «[There is a] widespread view that explanation
and understanding are semantically related in the sense that an explanation (of P) is satisfying if it provides understanding (of P). For it is reasonable to argue that one cannot understand something (P) by means of some other thing (Prem) which one has not understood»
(G. HX]jgo, o.c., pp. 97-98).
15 M. ;g^ZYbVc, o.c., p. 10.
16 See also L. ;V]gWVX], Understanding Brute Facts, «Synthese», 145 (2005), pp. 449-466.
&-%
_Zcc^[Zgl^ahdcbjac^m
ones are inadequate because they require a special epistemological status on
the part of the phenomenon doing the explaining. This is unacceptable, according to Friedman, because often the phenomena doing the explaining are
themselves brute facts. This type of explanation equates ‘understanding’ with
‘familiarity’. Friedman, however, wishes to retain the notion of ‘understanding’ as a necessary condition for explanation, but does not believe understanding resides in familiarity.
Moreover, ‘mere gain of information’ is also an inadequate account of understanding because a gain of information does not mean a gain of understanding. Thus, while the explanation of the Boyle-Charles law of gases by
kinetic theory is a genuine explanation, it is not because it produces a gain
of understanding through a gain of information, but instead, because understanding is produced on the global level.17 Understanding, rather, resides in
reducing the total number of independent phenomena that we have to accept
as ultimate or given. Friedman states:
«On the view of explanation that I am proposing, this kind of understanding provided by science is global rather than local. Scientiic explanations do not confer intelligibility on individual phenomena by showing them to be somehow natural, necessary,
familiar, or inevitable. However, our over-all understanding of the world is increased;
our total picture of nature is simpliied via a reduction in the number of independent
phenomena that we have to accept as ultimate. …[P]revious attempts…have failed
through ignoring the global nature of scientiic understanding. If one concentrates
only on the local aspects of explanation – the phenomenon being explained, the phenomenon doing the explaining, and the relation (deductive or otherwise) between
them – one ends up trying to ind some special epistemological status – familiarity,
naturalness, or being an ‘ideal of natural order’ – for the phenomenon doing the explaining. … [A]ttention to the global aspects of explanation…allows one to dispense
with any special epistemological status for the phenomenon doing the explaining. As
long as a reduction in the total number of independent phenomena is achieved, the
17 Immediately one might interject here and object to the idea that global understanding
is related to scientiic explanation. Bas van Fraassen, for example, objects that Friedman’s
view of uniication «totally discount[s] a speciic relation of explanation altogether» (B. kVc
;gVVhhZc, The Scientiic Image, Clarendon, Oxford 1980, p. 109). In other words, explanation
is a local afair and thus requires local understanding. Schurz sidesteps this objection by
combining Friedman’s global account with a local component. His theory of Explanatory
Uniication takes the following form: (E) A is an explanatory satisfying answer to the question Why-P? in the cognitive state C if (i) A claims (for some Prem) Prem & Prem ĺ P,
where Prem ĺ P is a premise-relevant correct inference in a broad sense, and (ii) C + A >
C,” where ‘C’ is the cognitive state of the questioner; ‘Why-P?’ is the explanation-seeking
u
question and ‘P’ is in need of explanation; ‘A’ is the answer; ‘Prem’ are premises; and ‘> u’
is the well-deined partial order relation between the cognitive states (of cognitive agents
or systems) measuring their degree of uniication. Condition (i) is local and condition (ii) is
global, though dependent upon the local condition (i) (G. HX]jgo, o.c., pp. 98-99).
ZmeaVcVidgnjc^[^XVi^dcVcYhX^Zci^[^XjcYZghiVcY^c\
&-&
basic phenomena to which all others are reduced can be as strange, unfamiliar, and
unnatural as you wish – even as strange as the basic facts of quantum mechanics».18
Thus, Friedman’s motivation in rejecting this account and the previous ones is
that they all require the phenomenon doing the explaining to have some sort
of special epistemological status.19 But many of our explanans are brute facts.
The elementary particles of physics and their behavior, for example, are brute
facts that are not themselves understood, but are still able to explain other
phenomena.
The question this raises, however, is how something not understood can
provide an “intelligibility transfer” to another phenomena to produce understanding. In other words, many standard accounts of scientiic explanation,
including the ones most recently mentioned, believe that «the intelligible
quality of the explanans lows down the argument, as it were, across the line
dividing explanans and explanandum, and into the explanandum, displacing
the latter’s mysterious quality».20 Friedman’s answer to the above question is
simply that there is no such “intelligibility transfer.” Furthermore, “to derive a
fact from any premises whatsoever…is never by itself suicient to render that
fact as less mysterious than it was before, because the premises are inevitably
ultimately mysterious».21 Barnes refers to this as Friedman’s «Thesis of Derivational Skepticism.” This anticipates the problem of deductive explanations
of laws, the discussion of which will be postponed until the next section.
Given all of the above considerations, how are we now to evaluate the relation of explanation to understanding, if it is not a matter of familiarity or
intelligibility transfer, according to Friedman? What does understanding consist of? Friedman’s answer is that «science increases our understanding of the
world by reducing the total number of independent phenomena that we have
to accept as ultimate or given. A world with fewer independent phenomena
18 M. ;g^ZYbVc, o.c., p. 18.
19 In line with this, Friedman also criticizes what he calls the “intellectual fashion view.”
According to this position, «the meaning of ‘scientiic understanding’ varies with historical
tradition, since what counts as an ideal of intelligibility does. Consequently, the very same
theory may count as explanatory for one tradition but may fail to explain for another. …
most believe that there can be good reasons, usually having to do with predictive power,
for choosing one ideal over another. Indeed, one writer, N.R. Hanson, practically identiies
predictive power with intelligibility». But, Friedman argues, «it would be desirable to ind a
concept of explanation according to which what counts as an explanation does not depend
on what phenomena one inds particularly natural or self-explanatory. In fact, although there may be good reasons for picking one ‘ideal of natural order’ over another, I cannot see
any reason but prejudice for regarding some phenomena as somehow more natural, intelligible, or self-explanatory than others. All phenomena…are equally in need of explanation
– although it is impossible, of course, that they all be explained at once» (M. ;g^ZYbVc,
o.c., pp. 12-13).
20 E. 7VgcZh (Zjhj), o.c., p. 4.
21 Ibidem.
&-'
_Zcc^[Zgl^ahdcbjac^m
is, other things equal, more comprehensible than one with more».22 Furthermore, it is not simply replacing one puzzling phenomenon with another, but
replacing one phenomenon with a more comprehensive phenomenon, and
thereby efecting a reduction in the total number of accepted phenomena. A
phenomenon’s being itself unexplained does not prevent it from explaining
other phenomena.
To demonstrate this, Friedman introduces the notion of independently acceptable law-like sentences. A law-like sentence is independently acceptable if
«there are suicient grounds for accepting one which are not also suicient
grounds for accepting the other». It also satisies the following two conditions:
1) If SŌ Q then S is not acceptable independently of Q.
2) If S is acceptable independently of P and QŌ P, then S is acceptable independently of Q.
Friedman supposes that there is at any given time a set K of law-like sentences that are accepted by the scientiic community. A sentence is then Katomic if it has no partition: “if there is no pair {S1, S2} such that S1 and S2 are
acceptable independently of S and S1 & S2 is logically equivalent to S.” Friedman then deines scientiic explanation using this terminology. He explains
that if S is to explain some Sƍ in K, then the set that S must reduce is the set
of independently acceptable consequences of S (conK(S)). Thus, Friedman’s
deinition of scientiic explanation is:
(D1ƍ) S1 explains S2 if there exists a partition ī of S1 and an Si ī such that
S2 conK(Si) and Si reduces conK(Si).23
Friedman’s archetypal example of a case of scientiic uniication is the subsumption of several laws under the laws of mechanics. He states:
«Consider a typical scientiic theory – e.g., the kinetic theory of gases. This theory
explains phenomena involving the behavior of gases, such as the fact that gases approximately obey the Boyle-Charles law, by reference to the behavior of the molecules of which gases are composed. For example, we can deduce that any collection
of molecules of the sort that gases are, which obeys the laws of mechanics will also
approximately obey the Boyle-Charles law. How does this make us understand the
behavior of gases? I submit that if this were all the kinetic theory did we would have
added nothing to our understanding. We would have simply replaced one brute fact
with another. But this is not all the kinetic theory does – it also permits us to derive
other phenomena involving the behavior of gases, such as the fact that they obey
Graham’s law of difusion and (within certain limits) that they have the speciic-heat
capacities that they do have, from the same laws of mechanics. The kinetic theory
efects a signiicant uniication in what we have to accept. Where we once had three
independent brute facts – that gases approximately obey the Boyle-Charles law, that
22 M. ;g^ZYbVc, o.c., p. 15.
23 Ibidem, pp. 16-17.
ZmeaVcVidgnjc^[^XVi^dcVcYhX^Zci^[^XjcYZghiVcY^c\
&-(
they obey Graham’s law, and that they have the speciic-heat capacities they do have
– we now have only one – that molecules obey the laws of mechanics. Furthermore,
the kinetic theory also allows us to integrate the behavior of gases with other phenomena, such as the motions of the planets and of falling bodies near the earth. This
is because the laws of mechanics also permit us to derive both the fact that planets
obey Kepler’s laws and the fact that falling bodies obey Galileo’s laws. From the fact
that all bodies obey the laws of mechanics it follows that the planets behave as they
do, falling bodies behave as they do, and gases behave as they do. Once again, we have
reduced a multiplicity of unexplained, independent phenomena to one. I claim that
this is the crucial property of scientiic explanation – science increases our understanding of the world by reducing the total number of independent phenomena that
we have to accept as ultimate or given. A world with fewer independent phenomena
is, other things equal, more comprehensible than one with more».24
However, there have been two chief objections raised against the technical
formulations of Friedman’s account. The irst is that Friedman’s account only allows K-atomic sentences to explain. The second objection is that under
Friedman’s account there are no K-atomic sentences. If these two objections
hold weight, then Friedman’s account of explanation does no explaining, or
as Salmon says, it «cannot get of the ground».25
In regard to the irst objection, Friedman’s deinition of explanation implies
that only K-atomic sentences can explain, which is clearly not the case. Kitcher provides two types of counterexamples to demonstrate that Friedman’s
deinition is too restrictive. One is when we explain the behavior of a complex
system by using laws from diferent theories and the other is when we have independently acceptable laws which belong to the same theory and can be put
together in genuine explanations. His example for the second type of counterexample is the usual derivation of the law of adiabatic expansion of an ideal
gas. The explanans is the conjunction of the Boyle-Charles law and the irst
law of thermodynamics. But, because these laws are acceptable on the basis of
quite independent tests, their conjunction is not K-atomic. However, Kitcher
argues, the derivation of the law of adiabatic expansion from the conjunction
is a genuine explanation. Kitcher also argues that Friedman’s example of the
explanation of the Boyle-Charles law by the kinetic theory is not K-atomic.
Briely, the second criticism comes out of Friedman’s attempt to accommodate this irst criticism by revising his deinition of explanation. Kitcher argues that it now entails that no K-atomic sentence can explain anything. The
main weakness Kitcher sees in Friedman’s account of Explanatory Uniication
is Friedman’s failure to provide a technically sound account of uniication:
24 Ibidem, pp. 14-15.
25 W. HVabdc, Four Decades of Scientiic Explanation, University of Minnesota Press, Minneapolis 1989, p. 99.
&-)
_Zcc^[Zgl^ahdcbjac^m
«Consider the intuitive method of counting laws which we normally employ and according to which the conjunction of Newton’s laws would be regarded as three laws.
If we look at the class of explanations that use the principles of Newtonian mechanics we ind that, by our intuitive count, the number of laws invoked in explanantia is
not signiicantly less than the number of laws derived as explananda. What is much
more striking than the relation between these numbers is the fact that Newton’s laws
of motion are used again and again and that they are always supplemented by laws
of the same types, to wit, laws specifying force distributions, mass distributions, initial velocity distributions, etc. Hence the uniication achieved by Newtonian theory
seems to consist not in the replacement of a large number of independent laws by a
smaller number, but in the repeated use of a small number of types of law which relate a large class of apparently diverse phenomena to a few fundamental magnitudes
and properties. Each explanation embodies a similar pattern: from the laws governing the fundamental magnitudes and properties together with laws that specify those
magnitudes and properties for a class of systems, we derive the laws that apply to
systems of that class».26
Consequently, Kitcher wishes to retain the connection between uniication
and explanation, but believes he can provide a more technically sound approach. His account, instead, relies upon derivational patterns for the uniication of scientiic explanation. Kitcher’s formulation does much to ameliorate Friedman’s diiculties, and the diiculties in Friedman’s formulation of
K-atomicity are not carried over into Kitcher’s uniicationist view. However,
there are other objections Friedman and Kitcher must face if their uniication
accounts are to stand up as plausible theories of scientiic explanation. I will
focus on two criticisms leveled by Eric Barnes: (1) Explanatory Uniication
is internally inconsistent and fails to provide an adequate account of understanding, and (2) Explanatory Uniication cannot account for causal asymmetries. The next section will address the irst criticism.
(#:meaVcVidgnJc^[^XVi^dcVcY9Zg^kVi^dcVaH`Zei^X^hb
Barnes argues that Explanatory Uniication is inconsistent and thus, cannot
adequately account for scientiic understanding. Moreover, Barnes believes
he can successfully replace Explanatory Uniication (which he thinks has become futile) with a causal account of understanding. Given Friedman’s commitment to the importance of scientiic understanding, this is a substantial
worry for the overall cogency of Explanatory Uniication, if these objections
are successful. In formulating his objection, Barnes charges Friedman with
having the very same problems that Friedman accuses competing theories of
26 P. @^iX]Zg, Explanation, Conjunction, and Uniication, «Journal of Philosophy», 73 (1976),
pp. 207-212, pp. 209-10.
ZmeaVcVidgnjc^[^XVi^dcVcYhX^Zci^[^XjcYZghiVcY^c\ &-*
having. Essentially, Barnes argues that Friedman’s view is no less derivational
than the traditional deductive models.
To demonstrate this, Barnes elucidates what he believes is Friedman’s main
criticism of competing theories. He refers to this as the “Thesis of Derivational Skepticism.” The idea here is that Friedman rejects competing theories on
the basis of the idea that derivation does not equal understanding. Scientiic
understanding cannot be produced merely by the derivation of an explanandum from some explanans. Moreover, Barnes claims that Friedman attacks
the traditional views on the basis that these views are mistaken in thinking
that the “mysterious quality” of explananda is understood in deriving them
from some suitable explanans. Next, Barnes argues that the premises that led
Friedman to his Derivational Skepticism are the same ones he relies upon in
formulating his thesis of Explanatory Uniication – Friedman’s view utilizes
the very same deductive principles. This is because he characterizes Friedman’s thesis of Explanatory Uniication as follows: «We understand the world
better when we countenance fewer independent phenomena because the
presence of fewer independent phenomena in our picture of nature amounts
to the presence of fewer fundamental mysteries. The fewer fundamental mysteries we are forced to accept, the less mystiied we are: the better we understand the world. QED».27
Barnes then illuminates the supposed inconsistency by claiming that the
plausibility of the Uniicationist Thesis of Understanding is dependent entirely upon the premise that an underived, independent phenomenon represents
a mystery not represented by a derived, dependent phenomenon. Barnes
states:
«To derive a fact from any premises whatsoever, Friedman is suggesting, is never
by itself suicient to render the fact as less mysterious than it was before, because
the premises are inevitably ultimately mysterious. If the premises are themselves
derivable from some deeper theory T, this theory will either be mysterious (in the
sense that its truth is unexplained) or derivable from some still deeper theory….
Hence, from Friedman’s viewpoint, it is entirely unclear how understanding of any
explanandum is (or will ever be) possible at all. …It seems to me, in other words,
that the Uniicationist Thesis of Understanding ought to be considered as plausible
only if we antecedently assume some account of understanding according to which
individual explananda are rendered understood, in some sense, by means of their
standing under a suitable explanans. But the falsehood of any such account was the
punchline of the Thesis of Derivational Skepticism, which is in turn supposed to be
one of the most crucial motivations for adopting the Uniicationist Thesis of Understanding. If we are to reject, with Friedman and Kitcher, the local nature of explanation and understanding on the grounds of the Thesis of Derivational Skepticism, the
27 E. 7VgcZh (Zjhj), o.c., p. 6.
&-+
_Zcc^[Zgl^ahdcbjac^m
question why we should take understanding to be proportional to the uniied status
of our global theoretical picture emerges as a very pressing one».28
However, as formulated, this objection appears to miss the point. Friedman
is quite willing to countenance genuine explanations that involve unfamiliar
or mysterious explanans. Perhaps, Barnes’ point is that Friedman’s notion of
global uniication presupposes a notion of local explanation. That is, global
uniication is a matter of a few brute premises being able to locally explain
lots of independently acceptable laws. Even so, I think that Friedman can fend
of this objection.29 Moreover, Barnes claims that Friedman’s account fails to
provide an adequate account of event explanation. Instead, Barnes argues
that Salmon’s causal mechanistic account can successfully resolve Friedman’s
problems. Yet, this objection appears to be misguided as well. First, it is the
case that Friedman attacks certain derivational accounts of explanation, but
these are accounts that deal with deriving laws from other laws. But, Barnes’
attack is based on the derivation of events, which does not appear to be the
motivation for Friedman’s adoption of the thesis of Explanatory Uniication.
Rather, it is more plausible to assert that Friedman’s actual motivation is to resolve Hempel and Oppenheim’s diiculty with explaining general laws by the
D-N model. This is further supported by Salmon’s statement that Friedman
restricted the applicability of uniication to regularities.30
Thus, the real problem Explanatory Uniication must face is the ability to
explain laws by other laws without trivial deductions of general regularities.31
In other words, Friedman’s account must be able to rule out, for example, the
derivation of a law from a conjunction of itself and another unrelated law as
a case of genuine explanation. This ofers no explanation or understanding
because the conjunction of these two laws does not reduce its “independently
acceptable consequences.” Hempel and Oppenheim present the problem of
explaining general regularities in a famous footnote:
28 Ibidem, pp. 4, 6.
29 See also L. ;V]gWVX], o.c., pp.460-464, in which he defends Friedman against Barnes
by emphasizing that many brute facts increase our global scientiic understanding.
30 W. HVabdc, Four Decades of Scientiic Explanation, cit., p. 184.
31 William Kneale, who ofers his own account of uniication, provides an example of
the problem of explaining laws: “When we explain a given proposition we show that it
follows logically from some other proposition or propositions. But this can scarcely be a
complete account of the matter. For if I hear that there is a lion in my garden and demand
an explanation of this curious fact, I am certainly not satisied by a statement that there are
two lions in my garden, although the irst proposition follows logically from the second.
An explanation must in some sense simplify what we have to accept. Now the explanation
of laws by showing that they follow from other laws is a simpliication of what we have to
accept because it reduces the number of untransparent necessitations we need to assume,
but this may not be obvious at irst sight” (W. @cZVaZ, Probability and Induction, Clarendon,
Oxford 1952, p. 91).
ZmeaVcVidgnjc^[^XVi^dcVcYhX^Zci^[^XjcYZghiVcY^c\ &-,
«The precise rational reconstruction of explanation as applied to general regularities presents peculiar problems for which we can ofer no solution at present. The
core of the diiculty can be indicated by reference to an example: Kepler’s laws, K,
may be conjoined with Boyle’s law, B, to [form] a stronger law K.B; but derivation of
K from the latter would not be considered an explanation of the regularities stated
in Kepler’s laws; rather, it would be viewed as representing, in efect, a pointless
“explanation” of Kepler’s laws by themselves. The derivation of Kepler’s laws from
Newton’s laws of motion and gravitation, on the other hand, would be recognized
as a genuine explanation in terms of more comprehensive regularities, or so-called
higher-level laws. The problem therefore arises of setting up clear-cut criteria for the
distinction of levels of explanation or for a comparison of generalized sentences as
to their comprehensiveness. The establishment of adequate criteria for this purpose
is as yet an open problem».32
Salmon argues that if Friedman can resolve the problems of K-atomic sentences, then his account of explanation will solve Hempel and Oppenheim’s
problem.33
Barnes’ objection to Friedman, then, must be that only a causal account of
explanation can solve this problem. This, however, has yet to be demonstrated
and now the burden of proof is on Barnes’ side. Barnes believes that he can
replace Explanatory Uniication and global understanding by explaining particular facts through their causal basis. Even if this is the case, however, this
provides no compelling reason to reject the thesis of Explanatory Uniication
altogether. This can be seen by examining Salmon’s views on the compatibility of uniication theories with causal mechanistic theories.
Salmon claims that the uniication view (top-down) is compatible with a
causal or mechanistic view (bottom-up). Neither of these views need be rejected, but can rather be seen as two diferent ways of utilizing the ideal explanatory text (the causal view being primarily used for events and the uniication view being used primarily for laws). As an illustration, he ofers a scenario
where a physicist is asked to explain the motion of a balloon in an airplane.
Salmon explains that this explanation can be given on either a global or local
level. That is, the physicist can either provide a causal mechanistic explanation
involving the behavior of molecules or he can provide a uniication explanation by citing Einstein’s principle of equivalence and showing how the motion
of the balloon its under this general law. Neither of these explanations is incorrect, and which one to utilize is only a matter of context.34
32 C. =ZbeZa - P. DeeZc]Z^b, Studies in the Logic of Explanation, «Philosophy of Science»,
15 (1948), pp. 135-175, p. 159, n. 28.
33 W. HVabdc, Four Decades of Scientiic Explanation, cit., p. 100.
34 Ibidem, pp. 183-85. And, how distinct these explanations really are depends on the relation between causation and the equivalence principle. See also W. HVabdc, Scientiic Expla-
&--
_Zcc^[Zgl^ahdcbjac^m
This insight by Salmon raises the question over whether Explanatory Uniication is a genuine rival to the causal approach. Moreover, what exactly is the
relation between uniication and causality? Most importantly, can the demand
for causal reasons be solely explained in terms of uniication increase? Kitcher,
however, particularly in his 1989 work, does seem to advocate the exclusive
acceptance of Explanatory Uniication over the causal approach, because the
explanatory powers of causal reasons can be solely explained by uniication.
Kitcher argues that causal claims should be grounded in explanatory dependency. Causal dependence is derivative of explanatory dependence. For Kitcher, then, Explanatory Uniication is not set up as a complement to the causal
approach, uniication being used for laws and causality being used for events.
Rather, Explanatory Uniication can explain both laws and events by an appeal
to the explanatory store.35 In order to do this however, Kitcher needs to be able
to account for our intuitive understanding of causal asymmetry. If Kitcher is
able to demonstrate this, then Explanatory Uniication should be evaluated on
its merits alone without conceding any lack of ability to explain events. Gerhard Schurz, to the contrary, believes that uniication cannot explain causal
reasons of singular events, but that it can provide causal explanations of laws.
His reasoning for this is also based on the consideration of causal asymmetry
and his contention that uniication cannot adequately account for it. In view
of this, I will now turn to an examination of the problem of causal asymmetry, with an eye to whether Explanatory Uniication is a view that can stand on
its own, particularly with respect to the explanation of particular events.
)#:meaVcVidgnJc^[^XVi^dcVcY6hnbbZign
It is commonly recognized in the famous tower-shadow example that the
height of tower is a genuine explanation for the length of the shadow, whereas the length of the shadow is not a genuine explanation for the height of the
tower.36 However, this example presents several problems for the traditional
nation: Causation and Uniication, in Causality and Explanation, Oxford UP, Oxford 1998. For
an opposing view, see E. LZWZg and J. kVc7djlZa, Causation, Uniication, and the Adequacy of Explanations of Facts, «Theoria», 24 (2009), pp. 301-320.
35 P. @^iX]Zg, Explanatory Uniication and Causal Structure, in «Minnesota Studies in the
Philosophy of Science», Volume m^: Scientiic Explanation, U of Minnesota Press, Minneapolis 1989, p. 436.
36 This example was provided by B. kVc;gVVhhZc, The Scientiic Image, cit., pp. 132-134,
and seems to be derived from Bromberger’s lagpole-shadow example in A Theory about the
Theory of Theory and about the Theory of Theories in B. 7Vjbg^c (ed.), Philosophy of Science:
The Delaware Seminar, Vol. 2, Wiley & Sons, New York 1963; and Why Questions, in R. G.
8dadYcn (ed.), Mind and Cosmos: Essays in Contemporary Science and Philosophy, University
of Pittsburgh P, Pittsburgh 1966. See W. HVabdc, Four Decades of Scientiic Explanation, cit.,
p. 47 for treatment of the lagpole-shadow example.
ZmeaVcVidgnjc^[^XVi^dcVcYhX^Zci^[^XjcYZghiVcY^c\ &-.
D-N model. In addition to issues of temporal asymmetry there are also issues of causal asymmetries. The reason we do not want to explain the tower
height in terms of the shadow’s length is because of our belief that efects do
not explain their causes. However, according to the D-N model, both directions are equally explanatory because of inferential symmetry. Built into the
D-N model is the symmetry of explanation and prediction. The same logical
schema is applied both to explanation and prediction so that nomically connected temporally later events including efects can serve in genuine explanations. Because this is counterintuitive, modiications of the D-N model must
be made to accommodate such asymmetries. Explanation cannot merely consist in deductive derivation.37
Barnes charges Explanatory Uniication with being unable to explain why
one is a genuine explanation and not the other. In other words, under Explanatory Uniication, derivations moving in both directions would count as explanations, much like the D-N model. Thus, Explanatory Uniication cannot
accommodate the causally asymmetric nature of explanation.38 Because the
issue of asymmetry is explicitly addressed in Kitcher and not by Friedman,
I will utilize Kitcher’s formulation of Explanatory Uniication when defending it against the charge of being unable to account for causal asymmetry.
Barnes charges Kitcher’s account with having two equally unifying argument
patterns in the tower-shadow case. Barnes argues that both the origin and
development (O&D) pattern of length explanation and the shadow-based
explanation (hWZ) pattern have an equally unifying efect, and thus, equally
unifying explanatory stores. Schurz agrees with Barnes and explains that for
individual events, utilizing the uniication approach fails to capture the causal
direction:
«Concerning the explanation of singular events, considering a biconditional law
AļB where only the direction from A to B is the causal one. Why should it be
more unifying to explain B-events by A-events than to explain A-events by B-events?
A detailed analysis shows that if the law is deterministic, the situation is completely
equivalent vis-à-vis uniication, since every third event C must be correlated with A
and with B to exactly the same probabilistic degree. Thus, in the area of explana-
37 W. HVabdc, Four Decades of Scientiic Explanation, cit., pp. 46-50.
38 Barnes cites Kitcher’s antirealist stance toward causation due to Humean worries and
then wonders how explanatory stories can account for the causal structure of the world.
He says, “Kitcher thus takes his theory of explanation to do what causal theories cannot
do: establish the very possibility of understanding the causal structure of the world (despite
the Humean predicament), and thus establish the very possibility of possessing knowledge
with explanatory force.” Barnes is thus quite skeptical that Kitcher’s project can be carried
out (E. Barnes (ZjeV), o.c., p. 560). See also V. <^_hWZgh, Why Uniication is Neither Necessary
nor Suicient for Explanation, «Philosophy of Science», 74 (2007), pp. 481-500.
&.%
_Zcc^[Zgl^ahdcbjac^m
tions of singular events, the causal requirement does not follow from the uniication
requirement…».39
Schurz, however, does not see this as damaging to the uniicationist account
because he believes causal theories are only themselves justiied by their unifying power. In other words, to explain events Explanatory Uniication requires
an appeal to causal theories, but on the theoretical level, causal theories are
themselves only justiied by their unifying power. Nonetheless, Explanatory
Uniicationists need not take this route. In fact, I think that both Barnes and
Schurz are wrong in their assessment of Explanatory Uniication’s ability to
account for asymmetry. Kitcher already has available, via appeal to explanatory stores, a strategy for basing causality of particular events on uniication.
To explain, Kitcher’s immediate response to the tower-shadow asymmetry
problem is to widen the scope of phenomena to all objects of length and not
just tower-lengths. From this, he can formulate two patterns: The O&D pattern and the SBE pattern. The O&D pattern can be repeated over and over to
a wide range of phenomena (people, lagpoles, mountains, particles, etc.). Thus,
even though SBE can explain the height of the tower by the length of the
shadow, the SBE pattern is not a genuine explanation because there are other
ways, namely the O&D pattern, to derive the same conclusion but with more
unifying power. The key to Kitcher’s analysis, then, is in widening the scope of
phenomena under consideration. Doing so easily allows us to see that hWZ is
less unifying, for not all objects of length cast shadows. Furthermore, because
hWZ fails to explain the length of all objects, whereas Kitcher argues O&D can,
an explanatory store E(K) will be more unifying with only the O&D pattern
rather than with both or just the SBE.
But, even if this answer is able to account for the tower-shadow example,
one has only to point out that there may be other asymmetrical examples
that are not so easy to resolve. In other words, the fact that not all objects
cast shadows was able to save Kitcher’s account on that score, but what about
situations in which there are two equally unifying explanations? Or, let us just
imagine we are in a world where all objects do cast shadows. What would
Kitcher say then? Putting aside the possible world in which all objects do cast
shadows, one could argue equally as efectively that the tower has a complex
dispositional property. This property is «the disposition to cast a shadow of
such-and-such a length on such-and-such a surface if illuminated by a lightsource at such-and-such an elevation above the surface».40 From this, we can
derive a dispositional shadow pattern to replace SBE. Let us call this pattern
DSE. This, one could argue, has an equally unifying pattern as O&D. And if
39 G. HX]jgo, o.c., p. 100.
40 P. @^iX]Zg, Explanatory Uniication and Causal Structure, cit., pp. 485-486.
ZmeaVcVidgnjc^[^XVi^dcVcYhX^Zci^[^XjcYZghiVcY^c\
&.&
one questions whether this pattern applies to all objects with length, one need
only add to the pattern a disjunctive predicate of the form:
«X has the disposition to cast a shadow if illuminated by a light source or x has the
disposition to produce an absorption pattern if x is suitably coated and irradiated or x
has the disposition to cast a shadow if x is covered with opaque material or x has the
disposition to cast a shadow if x is sectioned and unrolled or x has the disposition to
cast a shadow after x has been treated to block its own light sources or…[etc]».41
The critic of the uniicationist view can in this way supply two equally unifying patterns. What is Kitcher now to do? Kitcher’s irst line of response is
to claim that the above dispositional property is an ‘unprojectable predicate’.
That is, it is a property that has been gerrymandered to include an «artiicial congeries of properties as a single characteristic».42 Furthermore, because
there is no common dispositional property that can be employed for all objects, they do not it into a single genuine pattern.
However, Barnes argues that there are numerous other examples of argument patterns that do not rely upon unprojectable predicates and that are
“maximally unifying” but non-explanatory. The three types of counterexamples Barnes uses are temporally symmetric closed systems, evidentiary arguments, and unexplained facts. All of these examples show, Barnes argues, that
under Kitcher’s account there are backward explanations that go from efects
to causes, which are more unifying than forward explanations that proceed
from causes to efects. Consequently, backward explanations are genuine explanations according Explanatory Uniication. Todd Jones, however, believes
Barnes is wrong on this score. Explanatory Uniication is able to account for
asymmetry in these backward or retrodictive pseudo-explanations. This is so
because the theory of Explanatory Uniication rests on two important theses:
«1) In cases where the only way to derive a precise description of some phenomenon
is to retrodict it from later information, there are other, more unifying, forwardlooking patterns which can partially derive these conclusions and are preferred on
uniication grounds.
2) The forward-looking derivations of conclusions, even if only partial, help unify
our knowledge because the patterns involved belong to related families of patterns.
Membership in these larger families make forward-looking patterns preferred on
uniication grounds».43
41 Ibidem.
42 By a “projectable predicate” Kitcher means something which refers to a natural kind.
So, an “unprojectable predicate” does not refer to a natural kind (E. 7VgcZh (ZjeV), o.c.,
p. 564).
43 T. ?dcZh, How the Uniication Theory of Explanation Escapes Asymmetry Problems, «Erkenntnis», 43 (1995), pp. 229-240, p. 231.
&.'
_Zcc^[Zgl^ahdcbjac^m
Given these two important features, none of Barnes’ examples count as unifying argument patterns or explanations. Jones is able to refute each of these
supposed counterexamples by appealing to “epistemologically intermediate”
accounts, whereby although the explanation provided is partial or speculative,
it is more unifying. Jones explains:
«There are several reasons that the partial account should be considered the more
unifying derivation, and would thus be counted as the truly explanatory account by
uniicationists. The guiding intuition of the uniication perspective is that we should
try to derive as many conclusions as we can, using the same argument patterns again
and again. A partial account tells us a great deal about numerous features of the conclusion, without having to add any new patterns to our special maximally unifying
set. Additionally, the less precise, more schematic premises used in partial explanations employ what might be called ‘placeholder’ terms. …The use of placeholder
terms not only gives us partial conclusions, but also guides us in our gathering of
new information, telling us what to look for to enable us to generate more precise
conclusione».44
Additionally, Barnes’ claim that the Newtonian Predictive Pattern is equally as
unifying as a Newtonian Retrodictive pattern (where later states of the system
are used to derive earlier states) can also be refuted by appeal to “families” of
patterns. The idea here is that one way to unify our knowledge is by using the
fewest number of patterns. But this is not the only way our knowledge can be
uniied: it can also be uniied through using similar patterns belonging to the
same “family.” Hence, some derivation patterns are more unifying by their
membership in a family of patterns that derives far more conclusions than any
rival patterns or rival families of patterns. Jones then goes on to explain that
the Newtonian Predictive Pattern belongs to the family of O&D patterns. So,
even if one can use the Newtonian Retrodictive Pattern to derive the same conclusion, it does not belong to this family of patterns and is thus less unifying.
The obvious objection here, however, is that the Newtonian Retrodictive
Pattern may belong to an equally unifying family of “backward” derivations.
If this is the case, then Kitcher’s problem reappears. Jones, however, argues
that even if there were backwards family patterns, these families would be
less unifying than forward-looking patterns because the backward patterns
cannot derive conclusions about the present and future. Moreover, even if
one had access to future events, these backward patterns cannot also derive
conclusions about the past.45 Thus, Jones concludes, forward-looking families
44 Ibidem, pp. 234-35.
45 This is because «the second law of thermodynamics guarantees that for things looked
at from an aggregate ‘macro’ perspective, increases in entropy will continually wipe out the
traces of past events» (ibidem, p. 239).
ZmeaVcVidgnjc^[^XVi^dcVcYhX^Zci^[^XjcYZghiVcY^c\
&.(
of patterns will always be more unifying than retrodictive families of patterns.
Hence, Kitcher’s uniication account is saved from Barnes’ charges that it cannot account for causal asymmetry.
Accordingly, Explanatory Uniication is a viable theory for the explanation
of both laws and events. Granted, there are other problems that Kitcher’s account of Explanatory Uniication faces that need to be resolved such as how
one is able to determine what its in each explanatory store in addition to the
pragmatic consideration of how one would cognitively be able to access the
unifying power of a possibly ininite explanatory store; but, these are issues
that can perhaps be resolved with a greater reinement of Kitcher’s unifying
method. What is important is the fact that Explanatory Uniication can account for causal asymmetry, spurious uniications, and explanations of laws.
*#;jgi]Zg>hhjZh[dg:meaVcVidgnJc^[^XVi^dc
Obviously, there remain additional questions one must consider before one is
able to evaluate the overall success of Explanatory Uniication. Some of these
questions include:
1) Is Salmon is right in claiming the causal mechanistic and the uniication
views are compatible, or do the commitments that lead one to accept the
causal view force one to reject the uniication view?
2) If we grant that laws are better explained by the uniication view and
events or statistical facts are better explained by a causal view, what does this
tell us about the nature of explanation?
3) Finally, what does the utilization of two difering views say about the ability of science to be understood or uniied?
I am not prepared here to provide answers to these questions, as treatment
of these issues is beyond the scope of this present paper. Rather, I ofer these
questions to make clear that the issue of Explanatory Uniication’s success as
an adequate theory of explanation does not merely reside in its ability to account for trivial deductions or asymmetries. To the contrary, until the nature
of scientiic explanation is itself better understood, this judgment will have to
be reserved. Explanatory Uniication, however, has numerous virtues, one of
which is its ability to render coherent the nature of scientiic understanding,
which is no small task indeed.
Uniication also has a tremendous amount of support in the scientiic community. Salmon, for example, asserts that uniication is a fundamental goal
of explanation. Hempel also agrees that uniication is important to scientiic
explanation, stating that «a worthwhile scientiic theory explains an empirical
law by exhibiting it as one aspect of more comprehensive underlying regularities, which have a variety of other testable aspects as well, i.e., which also imply various other empirical laws. Such a theory thus provides a systematically
&.)
_Zcc^[Zgl^ahdcbjac^m
uniied account of many diferent empirical laws».46 Hempel further claims
that, «what scientiic explanation, especially theoretical explanation, aims at
is not [an] intuitive and highly subjective kind of understanding, but an objective kind of insight that is achieved by a systematic uniication, by exhibiting
the phenomena as manifestations of common, underlying structures and processes that conform to speciic, testable, basic principles».47
Finally, there are other noteworthy advantages in accepting a uniication
account of scientiic explanation. First, uniication is intimately tied to understanding and as we have seen, understanding is an important part of scientiic
explanation. Second, uniication bypasses the diiculties presented by Hume’s
analysis of causation. Third, the thesis of Explanatory Uniication resolves
Hempel’s diiculty in explaining laws with the D-N model, or at least comes
quite close. Fourth, where the nomic expectability approach and the causal
approach are mutually inconsistent, Explanatory Uniication is compatible
with either view as a supplement to their local analyses. Thus, an acceptance
of Explanatory Uniication does not entail a rejection of the causal approach;
they may be two diferent ways of interpreting the ideal explanatory text.
Finally, though it can be compatible with other theories, Explanatory Uniication is also able to explain laws and events on its own. There appear to be
no insuperable challenges to the coherence or consistency of the thesis of
Explanatory Uniication, and it is adequately able to respond to both the issues of asymmetry and spurious uniications. Although I believe I have successfully refuted the criticisms of Barnes in demonstrating that the thesis of
Explanatory Uniication is able to explain laws and account for causal asymmetry, there still remain many issues to be addressed before a inal assessment
of Explanatory Uniication’s success as a theory of scientiic explanation can
be made. It is my hope that with future reinements of the technical aspects
of the theory that these problems with be resolved as well.
6WhigVXi: This paper represents a response to the criticisms made by Eric Barnes in
“Explanatory Uniication and the Problem of Asymmetry” and “Explanatory Uniication and Scientiic Understanding” against the thesis of Explanatory Uniication.
This paper responds to Barnes’ two main criticisms, that of derivational skepticism
and casual asymmetry, and successfully refutes his objections. This paper also defends the plausibility of the uniicationist account of scientiic explanation because
of its ability to render coherent the notion of scientiic understanding, focusing in
particular on the work by Michael Friedman and Philip Kitcher.
@ZnldgYh: uniication, scientiic explanation, derivational skepticism, causal asymmetry,
philosophy of science, Kitcher, Friedman, Barnes.
46 C. =ZbeZa, Aspects of Scientiic Explanation and Other Essays in the Philosophy of Science,
cit., p. 444.
47 >YZb, Philosophy of Natural Science, Prentice Hall, Englewood Clifs 1966, p. 83, taken
from P. @^iX]Zg, Explanatory Uniication, cit., p. 508.