Briggs on Antirealist Accounts of Scientific Law

Briggs on Antirealist Accounts of Scientific Law

Rachel Briggs’ (2009a) critique of “antirealist” accounts of scientific law – including

(Halpin 1994) – is part of a project meant to show that Humean conceptions of scientific law are

more problematic than has been commonly realized.1 Her introduction to the full project, (Briggs

2009b), nicely sets out the Humean view and her general objection:

One fundamental schism in the philosophy of science involves the status of nomological

properties: does the universe at bottom contain any laws, dispositions, or objective chances?

… The debate can be cast as an argument about Humean Supervenience: roughly, the thesis

that nomological facts supervene on non-nomological facts…. Closer attention to David

Lewis’s ‘Big Bad Bug’, (1986, xiv-xvi; 1994), may help advance the discussion. The Big

Bad Bug is an argument to the effect that objective chances fail to supervene on non-

nomological matters of fact. If sound, the Bug is disastrous for Humean Supervenience. I

argue that the Big Bad Bug is a stronger argument than many defenders of Humean

Supervenience have realized. (428)

Indeed, Briggs’ (2009a) argument provides a new challenge to the Humean, a thoroughly

epistemic version of David Lewis’ Bug. Still, I will argue, the antirealist (perspectivalist and

expressivist) accounts she criticizes have the resources to withstand the challenge and come out

stronger for it.

A Humean attempts to show that all facts supervene on a basis of non-nomological,

“occurrent” facts. As I would put it, a scientific account involving lawful connection, chance,

cause or other dispositional concepts provides an interpretative overlay for the totality of events.

For a typical proponent of “Humean Supervenience”2, nomic claims may count as true but only

insofar as nomic concepts are required for a best scientific systematization of all occurrences. At

1

“bottom”, then, the world postulated by the Humean involves only non-nomic fact. But, Briggs

argues, Humean chance (seen as supervenient on non-nomic facts) proves problematic because

of chance’s close connection to epistemic probability or credence. She contends that though

antirealist accounts of law may provide Humean solutions to metaphysical issues regarding

chance, e.g., how to attribute chance to non-actual possible worlds, they leave untouched the

epistemological “Bug” for Humeanism: attribution of chance and credence for epistemic

possibilities.

After section one’s discussion of the Bug, section two of this article contends that a

perspectivalist (or, alternatively, an expressivist) account has a natural solution to Briggs’

epistemological argument. Still this antirealism has a cost, viz., relativity to point of view.

Section three argues that the price is right: there are significant advantages to a perspective

dependent account of laws.

1. Brigg’s Better Bug

Many contemporary Humean accounts of scientific law, including my own, are indebted to

Lewis’s definition in terms of ideal scientific theory. Put roughly, Lewis defines laws for a

possible world w to be the consequences of the best scientific systematization for that world’s

non-nomological facts. So to think about w’s laws, one is to imagine what ideal science would be

for w’s events and take w’s laws to be defined as the logical consequences of this “best-system”.

The details of just what constitutes “ideal” science or “best-system” (some sort of concise

description) are controversial but not at issue in Briggs’ critique. Rather, she is concerned with

the Humean’s treatment of the credence-chance relationship (i.e., the relationship between

objective single case probability and reasonable degree of belief). First consider the objective

side of the relationship. Lewis’s Humean Supervenience position defines truth about chance in

2

terms of ideal science’s probability measure: a measure providing the best statistical fit to the

pattern of occurrences. Then insofar as the best system makes probability claims, these claims

are true. But as Lewis saw, this Humean reduction for chance may be problematic in its

relationship to credence; this is the “Bug” and requires a few paragraphs to develop.

Quantum mechanics is contemporary science’s attempt to give a best system for

fundamental physical reality. But a toy example of chance systematization, for normal coins and

their tosses, is sufficient here. For example, a Canadian loonie coming up tails (the loon

showing) has chance ½. No matter that loonies have not actually come up tails exactly half the

time: The conceptualization in terms of chance is far simpler than a flip-by-flip description yet

gives significant predictive value. According to Lewis’s Principal Principle (PP), in a broad

range of cases rational credence follows chance. Knowing only the chance 1/2 for a loonie toss

result, one should have a partial belief or credence of 1/2 in tails for that toss. PP spells out the

details and implies mixtures when chance is uncertain. However, we will need only a simple and

intuitive consequence. In normal contexts3 and for ‘X’ regarding future events, the principle

implies the following about a reasonable credence function.

(*) If theory T logically implies that Chance(X)=p, then Credence(X|T)=p.

Lewis (1986) stresses a concern with so-called “undermining” worlds, worlds he feared

might be the ruin of Humean Supervenience. The idea is easily spelled out using the loonie

example. The probabilities assumed, 50% each for heads or tails with each toss result

independent of the others, leave open a wide range of future sequences. For example, if loonies

are tossed a settled n times in the future, then there is a chance, 1 in 2n, that every future toss

lands with the loon up. Given a long run of all tails, future observers might well conclude that

the loonie coin was unfair – even were there no discernable structural asymmetry – so assign

3

chance in a non-standard way. However, by our hypothesis, a conclusion assigning non-standard

chance would be wrong: the coins just happen to all come up tails. The Humean, though, has

difficulty with this judgment. The future coin tosses – even if all loon-up – are part of a history

determining truth about chance. If the loonie results form part of a pattern best fit by an

alternative probability theory, then chances are not 50% after all. This is undermining for the

Humean: A theory of chance may allow a possible future which together with past and present

facts implies a contrary theory of chance.

Worse, argues Lewis, such undermining leads to contradiction given the Principal Principle.

To state Lewis’s argument, I will use ‘F’ as a parameter assigned an undermining world and let it

do double duty naming the world and the proposition that this world occurs. For this section, take

F to indicate the loonie world just discussed in which all future loonie tosses come up tails. On

the best-system account, F undermines T and so implies that T is false. Thus, the credence of F

given T should be zero. On the other hand, our theory of chance T gives F a probability 1/2n for

the assumed n future tosses. So, by PP (and specifically * with ‘F’ in place of ‘X’), our credence

for F given this theory should also be 1/2n, contradicting the first result of zero. In Lewis’s terms,

this reductio is the “big, bad bug” for Humean Supervenience.

Lewis’ argument explicitly concerns metaphysically possible but non-actual worlds

(those extending the actual past and present but with a non-actual future that currently has some

chance). Thus, the undermining relationship, that F implies ~T, is a matter of logical entailment.

I will restate the first half of the reductio, this way:

i) F logically entails the denial of theory of chance T, thus Credence(F|T) is zero.

The second half of the reductio can be compactly formulated as

ii) PP is true, so Credence(F|T)=Chance(F) which is non-zero.

4

By his (1994), Lewis is prepared to reject PP (replacing it with a new principle relating chance to

credence, “NP”) while continuing to premise F as a non-actual world.

My (1994) perspectivalist best-system account of laws is meant to preserve PP by rejecting

the premise of i), that F logically entails the denial of T. A practical understanding, I argue, takes

actual laws and projects them onto a counterfactual world like F. This is the normal and

appropriate scientific perspective to take on a nomically possible world. We understand it as

having our laws because that interpretation illustrates these laws and their chances and so

provides a scientifically useful model. (Again, the Humean assumes only a world of occurrent

facts while law attribution is an interpretive overlay on these facts. Thus, there need not be a

single right way to attribute laws to a world. After all, a world may be a model of many different

and conflicting scientific theories.) From this normal scientific point of view, we assign F actual

laws including T, so F does not contradict T, the first half of the reductio is false, and Lewis’s

version of the Bug fails given the perspectival best-system account.

Briggs, however, argues that these modal considerations – regarding non-actual but

nomologically possible worlds like F – are not relevant to the underlying epistemic problem: the

contradictory credence claims.

Anti-realist accounts of chance make significant headway against four metaphysical

problems related to the Bug. They do not, however, address the Bug itself. Their failure is

no accident: the Bug is driven chiefly by epistemic modality rather than nomological

modality, and by Humeanism rather than HS. Humeanism entails that if there are chances

at all, they can be constructed from categorical facts (perhaps together with facts about

the observer’s interests). Thus, anyone with epistemic access to the array of categorical

facts (perhaps together with the facts about her own interests) can rule out some theories

5

of chance which grant nonzero probability to the array of categorical facts. But according

to PP, knowledge of the categorical facts is sufficient reason to dismiss a theory of

chance only if theory assigns probability 0 to the categorical facts. At bottom, the Bug

involves a conflict between two principles of rational credence. Tinkering with

nomological modality misses the point and leaves the Bug intact. (Briggs 2009a, p. 91-

92)

The reductio, then, need not suppose F to be a nonactual possible world. Instead, because the

Bug depends on credence, taking F to be epistemically possible is sufficient for the reductio. To

see this, follow Briggs and replace logical entailment with a priori entailment in i):

i′) F a priori entails ~T, so Credence (F|T) is zero.

The premise of i′) may seem definitive of best-system theory: If undermining world F happens to

be actual, then the actual best system will be contrary to T. And, I want to admit, the conclusion

of i′) is a straightforward consequence of its premise. On the other hand, given PP, ii)’s

1 I am a bit uneasy with the “antirealist” characterization of the perspectivalist account of

science’s laws. For my concerns, see (Halpin 2003) on the objectivity of perspectivalist laws.

Still, as the pages below will indicate, I am much indebted to Briggs for her critique and

subsequent correspondence.

2 (Ward 2005), also criticized in (Briggs 2009a), gives an expressivist account of laws, so rejects

truth-claims about laws and any supervenience for facts about laws: there are no such facts. So,

in this sense he rejects Humean Supervenience. Still, he is a Humean for postulating no facts that

fail to supervene on the non-nomic occurrences.

3 “Normal” here means there are no crystal balls or other means to see directly into the future.

The full principle applies at all times when credence depends only on “admissible” evidence and

consequently is much more powerful than the simple (*).

6

conclusion that Credence (F|T) is non-zero would seem inescapable. So, the contradiction

remains with i) replaced by Briggs’ i′).

Briggs argues that my (1994) account is subject to argument i′), summarizing my account of

laws (for world w from the perspective from which w is actual) as follows:

(**) “The laws of a world from the perspective of its inhabitants are completely fixed by

the world’s array of categorical properties.” (2009b, p. 87)

That is, there is a single perspective and so a determinate assignment of laws for a theorist’s

actual world; given a best-system account, these laws are just the consequences of the best-

system for the given world. Thus, if the actual world happens to have a future undermining T,

then the actual best system fixes actual laws contradicting T. This conditional would appear a

priori by the argument just given, consequently the premise of i′) is true and its conclusion, Cr(F|

T)=0, may seem inescapable.

In fact, I think, my (1994) and (1999) are both confused on this point about laws for the

actual world and specifically on the proper perspective of scientists working within a world

rather than taking a God’s eye view of the full world throughout all time. Nonetheless, the

perspectivalist can and should reject (**), i′), and Cr(F|T)=0 for reasons independent of the

undermining argument. Roughly, the laws properly assigned to the actual world depend on the

context of the assigner. To show this, I will argue that law and chance assignments for actual

world epistemic possibilities are perspectival just as for metaphysical possibilities. (Also, I will

briefly argue that the rebuttal works for Ward’s expressivism just as for the perspectival best-

system account.) For these purposes, F will be updated to a more fully-fleshed out undermining

world.

7

2. Contextual Chance Assignments

The following story, an American loony story, is probably (indeed hopefully) false. Yet it is

an epistemic possibility regarding the former U.S. Republican party nominee for Vice President.

Sara Palin’s move into television entertainment and “tea-party” fame, the story goes, revitalizes

her public standing and launches her successful quest to become President of the United States in

2017. Mid-century her still-deluded admirers impose a tribute: The Sara Palin $100 coin (needed

after a long inflationary bout caused by the inept economic policies of her administration). The

story is a bit far-fetched but epistemically possible circa 2010. So too are many possibilities for

the large number, n, of coin-tosses of these American loonies. Suppose for our story that Palin’s

face comes up each and every time the coin is tossed. And suppose that this happens even though

the coin is minted in a normal way and tossed just like any other coin. One last detail: Barack

Obama lives to see the coin minted but refuses to touch the thing.

That is the story. How should this American loonie world (hereafter F replacing the F-world

of section one) be assigned laws and chances? I do not insist that there is a unique answer.

Nonetheless, the story’s provenance suggests a preferred account. The possible Palin-coin-

tossing results of F are generated by thinking about the standard chances for a standard coin,

again 50% chance of heads, allowing a 1 in 2n chance for the all-heads sequence. Thus, we are

inclined to see hypothetical F as a bizarre statistical anomaly. Or so I claim. Here is another way

to make the same intuitive point about our perspective on hypothetical F. Again, F may or may

not represent the actual future. But on the hypothesis that F is actual, we may ask about what

would happen if in his old age Barak Obama were (counter to the facts of hypothetically actual

F) to lower himself to flip a few Palin-coins. Because the coin in question has normal symmetry,

and because there is no reason to see the hypothetical coin toss results of F as anything but an

8

anomaly, one plausibly thinks Obama would likely get about half the coins Palin-up. That is,

from our early 21st century perspective, one naturally assigns the Palin coin normal chance.

I take it to be part of rationality to model the future in accordance with the known or

justifiably believed chances. We apply our best theory to model both clearly counterfactual

possibilities and epistemic ones like the American loony story at hand. Of course, given that the

loony story is undermining, if it happens to be actual, scientists of the future will disagree with

our chance theory when they see the data come in. We can imagine these scientists first

skeptical, then very puzzled as their own repeated tests find nothing awry with the coin’s

construction or the claim it always comes up heads. They will be exasperated, no doubt, when

their own tosses all result in Palin glibly smiling up at them. (I imagine these theorists concerned

that some sort of cheating or even magic is afoot and then, perhaps, minting and testing other

coins with similar representations – the Tina Fey coin?, the George W. Bush coin? Perhaps some

may even be drawn to Plantinga’s Augustinian science. Still, the circa 2010 point of view sees all

this as mistaken. If the future holds in store what we now see as a statistical glitch, then the

resulting science will be wrong because based on a huge accident.) In any case, I take there to be

a good puzzle of hypothetical disagreement:

Present theorists interpret F as a world governed by T (so as a statistical anomaly) while

those possible theorists of the year 2060 who are inhabiting F (and may be in our future)

disagree and do so on the basis of more information, albeit information we take as mere

accident.

This hypothetical disagreement will seem puzzling on any account of law: Why should we

rational thinkers disagree with potential future experts, better informed observers, about their

possible future, a future we may in fact share? It is odd to say that they are just wrong about their

9

world but that we will come to agree with them if only (a) per improbable, it turns out that they

are in the actual future and (b) we wait fifty years and ourselves became better informed. For

present purposes, I want to explore only how the perspectival best-system account should handle

the American loony world and its puzzle of epistemic disagreement.

On the perspectival account, I think, this puzzle is straightforwardly resolved. We of the

early 21st century tentatively accept T. It is, then, practical to project T even onto undermining

possibilities. This projection is justified for the reasons noted above in the Canadian loony case:

attributing laws to a model of those laws is useful, indeed almost irresistible. Thus we see the

American loony possibility (F) as a statistical glitch governed (or, better) generated by T. That

said, if indeed the facts turn out in the shocking way described by the American loony story,

future theorists will rightly attribute contrary laws and chances to this same possibility F.

(Assuming that F is clearly undermining, future theorists will reasonably reject T for the good,

practical reason that a better overall system is available for the actual data.4) Accordingly, one

ends up with two reasonable perspectives on F rather than real disagreement.

Taking stock, Briggs’ reductio (as I have restated it) goes as follows:

i′) F a priori entails ~T, so Credence(F|T) is zero, while,

ii) PP is true, so Credence(F|T)=Chance(F) which is non-zero.

But I have just argued against i′). Even for the actual world, we bring perspective and, from the

point of view of the second decade of the 21st century (well before any Palin coins might be

made), see F as a statistical anomaly assigning it the 50% theory of chance, T. Thus from our

perspective, F does not a priori entail ~T and the reasonable credence of F given T is small but

non-zero as PP requires.

10

I believe that Ward’s view is consistent with this perspectival analysis though his account of

laws replaces claims about chance facts with scientific “norms”, e.g., “model following the 50%

chance rule”, to be projected onto possible worlds.

If chances are projections of our confidences, and our standards for projection are not so

stringent as to associate a unique chance with each total history, there need be no paradox

[of undermining]. (Ward 2005, 242)

Now, Briggs responds that the standards are strict enough to generate the Bug for Ward even if

not strict enough to assign a unique norm. About a world with an unusual number of coins

landing heads, she writes that if a rational believer “knew what the ideal observer knows—that

such a future is the actual future—then she would reject the ½ rule outright.” (2009a, 91) But this

last is wrong, or so I conclude from the analysis of the hypothetical disagreement puzzle: A fully

rational believer at the present time, somehow informed that F is true (i.e., given a God’s eye

4 In personal correspondence, Briggs describes a concern with “induction” to discover a best-

system rather than apply an accepted theory that is tentatively assumed best. As she points out, I

stress that Credence(F|T) is non-zero when projecting the settled view of 50-50 coin chances

onto F. However, she notes, when T itself is in question, it is important, e.g., when using Bayes’

theorem, that Credence(F|T) be zero. Quite right: Only from a perspective from which T is

tentatively accepted should one project T onto F and take Credence(F|T) to be non-zero. When

T’s status as best-system is in question, one is in something like the context of the exasperated

future scientists of the American loonie story, struggling with how the interpret/systematize the

data of world F. From this perspective, one of determining best-system and truth of T,

Credence(F|T) will be zero on the assumption that F undermines T. Halpin (1994) attempts to

deal with some of the complications of other perspectives and the PP for these.

11

view of total history), would still understand that the ½ rule is appropriate for normally informed

believers of her time and thus not to be rejected outright.

3. Advantages

To many it will seem odd (if not absurd) to take nomic claims as perspective dependent.

But the Humean need not find this perspectivalism problematic. Again, the Humean postulates

that “at bottom” there is only the occurrent. The rest, all non-occurrent fact, is what I call

“interpretative” and perspective relative (that is, theorists interpret the patchwork of occurrences

by assigning nomic connections legitimized by their place in theory: laws, chances, dispositions,

etcetera). Moreover, there are significant advantages to this “antirealism”. These can be seen in

new light given Briggs’ epistemological concern regarding law-attribution for the actual world’s

possibilities.

First, reiterating, the epistemological problem for undermining futures – the hypothetical

disagreement puzzle – extends to all views about chance. However, a proponent of the

perspectivalist view has a straightforward solution: What appears to be disagreement is, for the

perspectivalist, a difference of point of view. Thus, even if F is indeed actual, we early 21st

century thinkers properly take the Palin-coin to have only a 50% chance of heads while future

denizens of our undermining world are right to see chance of heads as very high.

Next, there are well known apparent failures of Humean Supervenience used to argue

against best-system accounts. It is argued that there are intuitively clear examples of world pairs

with the same occurrent facts but different nomic truths. In other worlds, there is a supposed

metaphysical disagreement between possible worlds with the same occurrent facts. But the

perspectivalist can treat this difference just like the first: two perspectives rather than real

12

disagreement. For example, Tooley and Carroll ask us to consider worlds with hypothetical

forces and particles. (Tooley 1977, Carroll 1994). In one case considered by Carroll, worlds w1

and w2 include X particles and Y fields: In w1 all X particles that enter a Y field follow law of

nature L1 and acquire spin UP. On the other hand, w2 is just like w1 excepting X-Y interactions:

In w2 when X particles enter Y fields they acquire spin DOWN as required by the laws L2 of w2.

Next consider what happens if a mirror were to deflect all X particles of both worlds away from

Y fields. The two resulting possibilities, w1′ and w2′ have exactly the same occurrences (we may

suppose). But, because inserting a few mirrors should not change laws of nature, it is intuitive to

see the first mirror world as having L1 and the second as having L2. So, the argument goes, the

two mirror worlds w1′ and w2′ have the same occurrences but different laws – a violation of

Humean supervenience. In contrast, the perspectivalist takes the differences between w1′ and w2′

as interpretive differences regarding one world rather than differences between two possible

worlds. From the point of view from which w1 is salient (e.g., considered actual) L1 is to be

projected onto the one mirror world. Here too there is no metaphysical disagreement but only

two perspectives; Humean Supervenience is salvaged – relative to a perspective. And, again, a

world like the mirror world is an epistemic possibility but wide open with regard to its nomic

interpretation.5

Another Briggs criticism, I believe, points instead to an advantage for perspectivalism.

She writes that the Bug bites my view “slightly harder” than Lewis’s view because “Lewis, who

adopted the philosopher’s perspective [i.e., a perspective assigning each world laws according to

its own best system], could abandon PP for an approximation called the New Principle” (2009a,

89). According to this new principle, NP, the credence for a claim given theory T is identical to

5 To be fair, Carroll is well aware of the contextualist response, indeed he highlights it in a

review article (forthcoming, p. 6).

13

the conditional chance of X given T: Credence(X|T)=Chance(X|T). For an undermining world F,

then, Credence(F|T)=Chance(F|T)=0 because F and T are seen as contradictory. This is “as

desired” writes Hall so that the Bug’s contradiction cannot be derived: ii) is not sound. (See Hall,

1994 p. 512.) Instead, the new principle allows only something like this pair of arguments:

i) F logically entails the denial of theory of chance T thus Credence(F|T) is zero,

iii) NP is true so Credence(F|T)=Chance(F|T) which is also zero.

Thus, PP’s problem of contradiction does not arise for NP.

Although the NP is Bug-resistant, it is not desirable when the concern is scientific use of

chance. Worlds like F are just extreme cases of statistical anomalies with respect to a theory like

T. Typical scientific analysis expects such glitches to occur from time to time in the universe.

Perhaps the first replicating molecule’s formation is such a glitch. Or again, multiverse theorists

usually see our pocket universe as anomalous because life-allowing physical parameters are so

unlikely. A straightforward analogy: we should have a minute credence that a coin will come up

loon-up in million straight trials given that the coin is fair. Accordingly, Credence(F|T) should be

minute but non-zero as PP advises but both NP (without perspectivalism) and iii) deny.

Yet another advantage for a perspectival account is its conceptual openness. Just as

chance claims are true (as Lewis supposes) in virtue of their place in ideal system, other

sciences’ concepts (I propose) can be similarly legitimized. The best-system is supposed to be

ideal and so a concise and useful description for the occurrent facts. How conciseness and

usefulness is best realized depends on perspective: compression of the data for a given purpose.

Biology may make use of functional and even teleological concepts (Millikan, 1990). Arguably,

psychology too systematizes the occurrent facts but with hugely different concepts and very

different standards from biology or physics. Still, there is no disagreement in this difference but

14

only change in perspective.6 The actual world, then, is properly interpreted in very different

ways, e.g., as merely mechanical or as involving purpose or mind, depending on perspective-

dependent theoretical need.

One nice application of this conceptual pragmatism has to do with concerns about the

“elegant” or “rational” or “intelligible” universe. We human animals, the story goes, use a

mathematical science that “deciphers the Mind of God contained in His work of creation.”

(Heller and Coyne, 2008). Robin Collins (2008) argues that “[t]he beauty and elegance of the

laws of nature also point to Divine design” and that

the laws of Nature themselves seem to be carefully arranged so that they are

intelligible, and in addition discoverable, by beings with our level of intelligence--like

solving a clever puzzle. This has been stressed by many prominent physicists. Albert

Einstein, for example, famously remarked that "the eternal mystery of the world is that

it is comprehensible.... The fact that it is comprehensible is a miracle".

Now, if laws govern creation in a metaphysically significant way, i.e., are prior to and

responsible for the universe’s events, then laws might, I suppose, be both inelegant and either

completely hidden from human endeavor or incomprehensible to those with human conceptual

abilities. But given the perspectival best-system account, laws are not governing but descriptive.

Scientific concepts are not mystically deciphered, on this view, but are legitimized in scientific

6 The eliminative materialist about beliefs and desires refuses to systematize from an intentional

stance. Dennett, on the other hand, takes beliefs and desires to be real in virtue of the success of

intentional systematizing. For the proponent of the perspectival best system account, a zombie

world is just the actual world from a perspective refusing to systematize using mentalistic

concepts. Again there are not two worlds with metaphysical differences but one world viewed in

different ways. See Halpin (manuscript).

15

theorizing given the concepts and standards of particular human perspectives. The resulting

concise descriptions, attempts at best-system, cover a broad range of phenomena, elegantly

unifying a subject matter. The consequence is discoverability, intelligibility, and elegance for the

perspectivalist’s laws.

Summarizing, Brigg’s epistemological argument provides a new and important challenge

to Humeanism. And she is correct to critique my conflation of epistemic and metaphysical

possibilities. Still, I have tried to show that the Humean “antirealist” has good reasons to resist.

Indeed, bringing perspectivalism to bear on epistemic possibilities (as well as metaphysical ones)

proves advantageous to the Humean in a number of ways: each of the advantages described in

this last section involve or may involve interpretation of epistemic possibilities. The upshot, I

have argued, is that the perspectivalist not only has the resources to meet her objection and

maintain the PP, but also solves certain problems of disagreement and failure of supervenience

while respecting conceptual openness in science.

16

REFERENCES

Briggs, R. (2009a) “The big bad bug bites anti-realists about chance”, Synthese 167: 81-92.

Briggs, R. (2009b) “The Anatomy of the Big Bad Bug”, Noûs 43 (3):428-449.

Carroll, J. (1994) Laws of Nature, Cambridge University Press, Cambridge.

Carrol, J. (forthcoming) Review of “John T. Roberts’ The Law-Governed Universe”, British

Journal for the Philosophy of Science, first published online September 10, 2011

doi:10.1093/bjps/axr031.

Collins, R. (2008) “The Case for Cosmic Design”

http://www.infidels.org/library/modern/robin_collins/design.html (retrieved August 7, 2011)

Hall, N. (1994) “Correcting the Guide to Objective Chance. Mind 103 (412):505-518.

Halpin, J. F. (manuscript) “Hard Problems, Interpretive Concepts, and Humean Laws”.

Halpin, J. F. (2003) “Scientific Law: A Perspectival Account”, Erkenntnis 58 (2):137 - 168.

Halpin, J. F. (1999) "Nomic Necessity and Empiricism", NOUS Vol. 33: 630-643.

17

Halpin, J. F. (1994) “Legitimizing Chance: The Best-System Approach to Probabilistic Laws in

Physical Theory” Australasian Journal of Philosophy, 72(3): 317-338.

Heller, M. and Coyne, G. (2008) A Comprehensible Universe: The Interplay of Science and

Theology, Springer-Verlag New York.

Lewis, D. (1994) “Humean Supervenience Debugged”, Mind 103 (412):473-490.

Millikan, R. (1990) “Truth Rules, Hoverflies, and the Kripke-Wittgenstein Paradox”, The

Philosophical Review, Vol. XCIX, No. 3: 323-353.

Tooley, M. (1977) "The Nature of Laws", Canadian Journal of Philosophy VII: 667-698.

Ward, B (2005) “Projecting Chances: A Humean Vindication and Justification of the Principal

Principle”, Philosophy of Science 72 (1):241-261.

18

19