Pre-proof - Einar Duenger Bohn

WHYTHISUNIVERSE?
EinarDuengerBohn
UniversityofAgder
Either the universe is intentionally designed or it is not. Which one is more
likely?Inwhatfollows,Idiscussanargumentincomparativefavoroftheformer,
aparticularversionoftheso-calledfine-tuningargument.Iclarifytheargument
(section1),replytosomeweakerobjections(section2),andpresentandreply
(albeit less conclusively) to what I take to be two stronger objections to it
(section3).Boththeargumentandalltheobjectionstoitaremoreorlesswell
known from the literature; my contribution here is simply to clarify the
argument and objections, reply to the objections, and thus argue that the
argument in fact stands surprisingly strong. Being agnostic about the existence
of an intention behind the universe, I for one find this conclusion deeply
interestingbecauseIfindallthepremisesplausible,sototheextentIconsider
theconclusionareductio,IhavetogiveupsomethingelseIbelieve.SoI’mjust
notsurewhattobelieveatthispoint.
1.THEARGUMENT
Firstbackgroundassumption: there exists a universe U and it contains rational
life.1Though I will here neither define ‘universe’ nor ‘rational life’, I take this
background assumption to be uncontroversial. Pace radical skepticism, that U
exists is evident from perception and that rational life exists is evident from
introspection.Itishereassumedthatwhat’sthusperceived,namelyU,includes
what’s thus introspected, namely rational life, as well as that what’s thus
introspectedisdependentonthephysicalnatureofU.Inotherwords,itishere
assumed that rational life as we know it could not have been realized in
somethingalientoourphysicaluniverse.
1Rationallifequarationalisnotessentialtotheargument,butitarguablycries
outlouderforanexplanationthanotherkindsoflife,and,aswe’llseelateron,
thuscryingoutforanexplanationisimportantfortheargument’sstrength.
1
Second background assumption: let’s call the hypothesis that U is intentionally
designed, I. I is a propositional hypothesis that we can grasp such that we can
sufficientlyunderstanditandworkwithit.Iamountstonothingmorethanthis:
U exists and contains rational life because there was an intention to design U
withrationallifeinit,andwhateverhadthatintentionhadsufficientknowledge
and powers to actually realize what it intended, and as a matter of fact, it did
realize it (as per the first background assumption above). There is no
presumptionthatsuchadesignerisGod,inanyparticularfamiliarsenseofthe
term,northatsuchadesignerissomethingparticularlygood.
Thirdbackgroundassumption:Uisfine-tunedforrationallife.Ourbestmodelof
Ucontainssomequantifiedlawsl,constantscandinitialconditionsci.Eachoneof
thesehasasamatterofempiricalfactbeenobservedtohaveauniquenumerical
value taken from some very wide ranges of theoretically possible such values.
Thenumericalvaluesthatl,candciinfacthaveareaccidentalinthesensethat
each one of the values could have been any other value from the ranges those
values are picked from. The ratios between the numerical values are also thus
accidental.Ifwesetanyoneofl,candciasunitvalue,theothervaluesstillare
arbitrary.Thatis,thereisnonecessaryconnectionbetweenanyofthenumerical
values. In short, the values of l, c and ci are each one “selected” seemingly at
randomfromsomewiderangesofpossiblesuchvalues.
What’s more, our best models of U tell us that only some few numerical
values selected from those wide ranges of possible such values are compatible
withrationallife.Thatis,ifl,candcihadhad(almost)anyothervaluesfromthe
ones they actually have, rational life would have been impossible (cf. our first
background assumption above). U is thus fine-tuned for rational life: l, c and ci
needtotakeonparticularvaluesthatstandinparticularratiostoeachotherin
orderforrationallifetobepossible.(Theratio-partisworthkeepinginmindfor
what’stocomebecause,forexample,obtainingeightbythrowingtwodiceisone
thing,butobtainingeightbythrowingafourandafourisanother.Thelatteris
2
lessprobablethantheformer.)Let’scallthehypothesisthatUisthusfine-tuned
forrationallife,F.2
ThehypothesisFissupportedbyaposteriorievidenceinphysics,notby
a priori evidence in philosophy (or theology), though, as with most evidence,
someapriorireasoningisneededtomakesenseofit.Inanycase,hereisatoy
modelofwhatwe’resupposedlygivenfromphysics,whichwe’llgetbacktoand
work with some more later on. It goes without saying that this is merely a toy
model; 3 it is the structure of the model that matters for the philosophical
argument in question, not the exact numerical values. Now, consider a 20
dimensionalspaceU20witheachoneofits(20orthogonal)axesbeingdividedup
into 1055 “points”. Any such point located in U20 is uniquely identified by a 20tuple of natural numbers: <n1,n2,…,n20>, where each 1≤ni≤1055. We let one and
onlyonepointinU20be“litup”,andletalltherestbenot“litup”.Callthepoint
thatislitupU,andcalltherestofthenon-lituppointsUi.IfweletthefactthatU
islitupcorrespondtothefactthatUcontainsrationallife,andthefactthatUiis
not lit up correspond to the fact that Ui contains no rational life, then our
perceptionandintrospectiontellusthatUistheactuallyexistinguniverse,and
thehypothesisFtellsusthatUiisnotlitup(i.e.thatUistheoneandonlypoint
inU20thatislitup).What’ssurprising,orshocking,cryingoutforanexplanation,
isthefactthatofallthepointsinU20,U–theoneandonlypointthatislitup!–is
thepointthatisactuallythecase.Whataretheoddsofthat?Answer:1:1055x20=
1:101100!
Fourth background assumption: let’s call the hypothesis that U exists and
contains rational life because of pure chance, C. C is a propositional hypothesis
thatwecangraspsuchthatwecansufficientlyunderstanditandworkwithit.C
amountstonothingmorethanthis:anyvalueofl,candciareequallylikely,and
the fact that they have the values they have is due to pure chance; it’s just like
throwingsomefairdice.Inourtoymodelabove,itamountstothis:anypointin
U20was“initially”equallylikelytoactuallyhappen(thisisbasedontheso-called
2Foradiscussionoftheactualfine-tuningevidencefromphysics,seee.g.Collins
(2003).
3Theexactnumbersarejustforfun,butnottoofarfromthesupposedtruth.
3
PrincipleofIndifference),andthefactthatUactuallyhappenedratherthanUiis
justpurechance;justlikeafair101100-sideddiemightlandontheoneandonly
onesidethatisred,say.
Fifthbackgroundassumption:thereisacoherentnotionofepistemicconditional
probabilitythatwecananddolegitimatelyworkwith.Iwillnottrytorigorously
defineithere(infact,Itakeittobeindefinable),buttheroughideaisthatwe
cananddo(allthetime!)legitimatelymakeassessmentsofrelationsofwarrant
between propositions and evidence (or simply between propositions, if we
understandevidenceaspropositional).Forexample,wecananddolegitimately
makeassessmentsofhowprobableitisthattherearepeopleoutsideofmyoffice
doorgiventhatIhearvoicesoutsideofmyofficedoor,orhowprobableitisthat
IwillbeabletoraisemyhandwithinthenextminutegiventhatIintendtoraise
myhandwithinthenextminute,orhowprobableitisthatadieisfairgiventhat
it lands six twenty times in a row. For a more scientific example, discussed in
Collins(2012),considerthecontinentaldrifttheory.Thediscoverythatmillions
ofyearsoldfossilsoflandanimalsinAfricaandSouthAmericawereverysimilar
to each other was taken to lend considerable empirical support to the
continentaldrifttheorybecauseitwasjudgedveryimprobablethattheywould
bethussimilarifthespecieshadnotatsomepointinteractedwitheachother,
somethingtheycouldnotatthetimehavedoneifnotconnectedbyland.Also,as
Sober(2003)pointsout,objectiveprobabilityisassignedtoresultsofachance
process, e.g. the throw of a fair die, but it is natural to think that, say, a law of
nature is not the result of such a chance process, so we must use a notion of
epistemicprobabilityinstead.Also,epistemicprobability,asopposedtoobjective
probability, let’s us assign a probability to a one-off event. What’s common in
suchcasesisthatwebracketanduseselectedbackgroundinformationtoassess
howprobablepisgivensomeq,i.e.howprobablepisgivenqplusssomeselected
backgroundinformation.We follow custom and write the epistemic conditional
probabilityofpgivenqandsomebackgroundinformationkas:Pr(p|q&k).4
4‘Pr’isoftenusedforotherkindsofinterpretationsofprobabilityaswell,e.g.
subjectiveandobjective,buthenceforth’Pr’,andthewords’probability’and
’probable’moregenerally,meansexclusivelyepistemicprobability(asdescribed
4
First foreground premise: the first premise we need is one direction of the socalledLikelihoodPrinciple.TheLikelihoodPrinciplestatesthatsomeevidencee
favorsapropositionpoveranotherpropositionqifandonlyiftheprobabilityof
e given p is greater than the probability of e given q. We also assume that if e
favors p over q, then, given e, p is morelikely than q, and if, given e, p is more
likely than q, then, all else being equal, we should believe p over q. So, for
example, my arguing with my wife is evidence that favors, and makes more
likely,thehypothesisthatsheisahumanbeingratherthanarobotifandonlyif
theprobabilityofmyarguingwithhergiventhatsheisahumanbeingisgreater
thantheprobabilityofmyarguingwithhergiventhatsheisarobot.Butweonly
needtheright-to-leftdirectionofit,andrecall,wetaketheprobabilityinvolved
tobeepistemic,asperthefifthbackgroundassumptionabove.Now,supposemy
arguingwithmywifeismoreprobable,giventhatsheisahumanbeingrather
thanarobot.Then,bytheLikelihoodPrinciple,myarguingwithherevidentially
favors that she is a human being over her being a robot, i.e. I should, all else
being equal, find it more likely, and hence believe, that she is a human being
rather than a robot, given my arguing with her. And that seems right. So, the
likelihood principle, at least the direction of it that we need, seems a good
principleindeed(thoughwe’llgetbacktoitinsection3below).Itiswidelyused,
arguablyindispensablyso,bothinscienceandordinarylife.5
Second foreground premise: if U is intentionally designed, i.e. if hypothesis I is
true,thehypothesisthattheuniverseUisfine-tunedforlife,i.e.hypothesisF,is
more probable than if U is due to pure chance, i.e. hypothesis C is true. This is
simply due to the fact that according to C the odds of U being the case is, as
pointedoutinthethirdbackgroundassumptionabove,somethinglike1:101100,
but according to I it is much, much higher, in fact close to 1:1. If someone
intendedtocreateU,andhadsufficientknowledgeandpowerstoinfactrealize
above),unlessnotedotherwise.Formoreonanepistemicnotionofprobability,
seee.g.Collins(2012)andWilliamson(2000:ch.10).
5Forafurtherdiscussionofthelikelihoodprinciple,seee.g.Collins(2012).
5
it,thenitisnaturaltothinkthatitisverylikelythatitwillinfactberealized;in
anycasemoresothan1:101100!
Giventhesetwoforegroundpremises,itlogicallyfollowsthatFfavorsIoverC,
or that given F, I is more likely than C, hence, all else being equal, we should
believeIoverCconditionalonF.Weputtheoverallargumentinthefollowing
logicallyvalidform:
1. Pr(F|I)>Pr(F|C)→L(I|F)>L(C|F) (InstanceoftheLikelihoodPrinciple)
2. Pr(F|I)>Pr(F|C)
(Premise)
3. ∴L(I|F)>L(C|F)
(1,2,MP)
whereLisevidentialfavoring,or(conditional)likelihood:FfavorsIoverC,or,in
otherwords,givenF,thelikelihoodofIisgreaterthanthelikelihoodofC;where
backgroundknowledgekisheresuppressedforthesakeofsimplicity.Sincethe
argumentislogicallyvalid,onemustdenyeitherpremise1orpremise2,ifone
wantstodenytheconclusion.
Before we consider some immediate objections that spring to mind, some few
thingsareworthnotingrightaway,ifnotonlytoavoidconfusion.
First note: conclusion 3 is a claim of evidential favoring, or comparative
likelihood,aclaimastowhichoneoftwohypothesesarefavoredormademore
likelybytheevidence;it’snotaclaimastowhichoneofthemweshouldbelieve
onitsown.Thatis,theargumentisnotanargumentfortheconclusionthatthe
universe is intentionally designed, but an argument for the conclusion that the
evidencefavors,ormakesmorelikelythattheuniverseisintentionallydesigned
over that it is due to pure chance. To the extent we should believe what our
evidencefavors,weshouldthen,allotherevidentialfavoringbeingequal,believe
thattheuniverseisintentionallydesignedoverthatitisduetopurechance.But
if all other evidential favoring is not equal, it might still be, for all the above
argumentsays,thatweshouldnotbelieveanyoneofthem.
6
Secondnote:theargumentisstillaprettystrongargumentbecauseitmeansthat
unless we can come up with a far, far better explanation than pure chance, we
should, following our evidence, believe the universe is intentionally designed.
And I’m not sure we can come up with a far, far better explanation than pure
chance; at least not one that beats the likelihood of the hypothesis that the
universeisintentionallydesigned, given F.Thoughweshouldofcoursealways
lookforsuchabetterexplanation.
Thirdnote:beingaconclusionofcomparativelikelihood,thereisnoneedtotake
astandonhowprobableorimprobableIorCisindividually,i.e.thereisnoneed
tofixthepriorprobabilityofIorC.Allweneeddoiscomparethemconditional
onF,whichisamucheasierthingtodo–andhasmuchmoreplausibility–than
assigningeachoneofthemadefiniteprobabilitytobeginwith.
2.SOMEWEAKOBJECTIONS
Inthissection,Ipresent,butrejectsomemoreorlesswell-knownobjectionsto
theargumentabove.Alltheobjectionsareultimatelyobjectionstopremise2of
the argument unless noted otherwise. I consider them in what I take to be,
roughly,anincreasingorderofstrength.
Firstobjection:wecannotassumethatthedesignerhasanintentiontocreateU
withoutassumingwhatistobeargued,namelythatIismorelikelythanC,given
F. That is, premise 2 already assumes conclusion 3. Reply: we’re not assuming
thatsomedefinitedesignerhasthedefiniteintentionofdesigningU,andhence
that U is more likely on that assumption than it is on chance. Rather, we’re
proposing the hypothesis that there is some designer that had the intention of
designingUasweactuallyfindit,asanexplanationofwhatweactuallyfind.And
then,onthebasisofthat,wecompareittocompetinghypotheses,andcometo
theconclusionthatitisthesuperiorhypothesis,thebetterexplanationofwhat
weactuallyfind.Thereisnoviciouscircularityhere.
7
Secondobjection:wecannotknowthemindofthedesigner,so,inparticular,we
cannot know what the designer intended; so, more generally, premise 2 fails.
Reply:thisobjectionseemstoconfusethatUisintentionallydesignedwithwhyU
is intentionally designed. Of course, we cannot know why U is intentionally
designed,butthatisnotthepoint.ThepointisthatweactuallydiscoverF,and
on the basis of that discovery conclude that we should believe that U is
intentionallydesignedoverthatUistheresultofpurechance.Thereisnowhere
a claim as to knowing exactly why the designer intended U, only that it did. So
the argument is not operating at an illegitimate level of specificity. If, on the
other hand, the objection is not confusing that with why in this context, then I
simply deny the objection: the whole point of the argument is that we should
believe that some designer designed U over U being the result of pure chance
(givenF).Itisnoobjectiontosimplystatethatwecannotknowthis.
Third objection: we’re unable to assess the proper probability of F given C
because our sample is biased in favor of a universe with rational life in it. In
other words, there might be many other unknown ways for the universe to be
other than those represented by the points in U20 discussed in the fourth
background assumption above, and these other ways matter to the proper
assessmentofhowprobableUis.Reply:ofcourse,thereisalwaysthedangerof
there being relevant data that we don’t know of yet, but that neither does nor
should stop us from making assessments of how probable a hypothesis is
conditionalonthedataweinfactdohaveavailable.Thatisalsopartofthewhole
point of appealing to epistemic probability to begin with. If not, all inquiry is
crippledbyskepticism.So,untilnewdatacomesin,1-3stands.
Fourth objection: the argument 1-3 as a whole has no philosophical interest
because there are too many alternative hypotheses we can easily cook up
conditionaluponwhichFismoreprobablethanFisconditionaluponC.Reply:of
coursewecancookupmanydifferentsuchhypotheses,buttheywouldallhave
littleindependentinterest,andhenceseemobjectionablyadhoc,i.e.postulated
forthesolepurposeofexplainingF.Whyweshouldtakesomehypothesesand
theories more seriously than others is a difficult issue in the philosophy of
8
science,butthatweinfactshould(anddo!)takesomehypothesesandtheories
moreseriouslythanothersisuncontroversial.Plausibly,bothhypothesesIandC
(andthemultiversehypothesisM,aswe’llseelater)areamongthosehypotheses
thatshouldbetakenseriouslyasexplanationsofFbecauseeachoneofthemhas
beenpostulatedindependentlyofF,andsoforgoodreasons,andassucharenot
objectionably ad hoc. They are all the viable candidates on the market, so to
speak.Thispointwillcomeupagaininsection3below.
Fifthobjection:theprobabilityofFgivenImightonitsfaceseemhigherthanthe
probability of F given C, but in fact the probability of I is so incredibly low to
begin with as to counterweigh the advantage of I over C even given F. Reply:
recall that hypothesis I only says that there is an intention behind U, and that
whoever (or whatever) so intended, had sufficient power and knowledge to
actuallyrealizethatintention.Assuch,itisaveryminimalthesis,sayingnothing
ofbeinganyparticularGodwithlogicallyproblematicproperties(suchasbeing
all-good, all-knowing, all-powerful, being both three persons and one person,
being both human and divine, etc.). Why is the probability of such a minimal
thesis so incredibly low that discovering F cannot outweigh it? Recall, our
probabilityisepistemic,notobjective,andourconclusionisevidentialfavoring,
notobjectiveprobability.Inanycase,moreneedstobesaidbeforethisobjection
hasanyforce.6
3.TWOSTRONGEROBJECTIONS
Inthissection,IreplytowhatItaketobetwostronger(andmoreinteresting!)
objectionstotheargument1-3fromsection1above.
6NotealsothatthesisIiscompatiblewith,forexample,thedesignerbeinga
nerdysciencestudent,existinginamuch(much)larger(betterorworse)
universethanours,with,say,someincredibleluckdesigningouruniverseina
lab.Itisperhapsunfortunatethattheargumentfromdesignisalwaysconnected
toasupernatural,divineandgooddesigner.Itneednotbeso.Though,ofcourse,
themoreremoveditisfromanyhistoricalnotionofaGod,themoreadhocthe
designhypothesismightbe(cf.thefourthobjectionabove).
9
Firstobjection:UisinnomoreneedofanexplanationthananyoneofUi,sowe
shouldnotinvokeIoverCtoexplainF,justlikeweshouldnotinvokeIoverCto
explain any of the “tunings” of Ui. This thus ultimately amounts to a denial of
premise 1 of the main argument. Reply: this is a harder objection because it is
hard to say anything definite about the line between what is in need of an
explanation and what is not. We have paradigm examples, but no definite
criterion. For example, a monkey typing Shakespeare’s complete works in one
attemptdefinitelycriesoutforanexplanation,butamonkeytyping‘jshdfoihne’
doesnot.Whatifthemonkeytypes‘bannanna’?Shouldwejudgeittobeacool
coincidence,orshouldweseekafurtherexplanation?Whatifhetypes‘banana’?
Idon’tknowhowtosettlethequestioningeneral.
The only suggestions I know of for how to begin drawing a definite line
between what does what does not need an explanation are two necessary
conditionsforwhensomethingneedsanexplanation,namely,ontheonehand,
when it challenges our presumptions of how it came about (White, 2000), and,
on the other hand, when we can glimpse a possible, but not too implausible
explanation of it (Horwich, 1982; Leslie, 1989).7In order for rational life, and
hence U, to be in need of an explanation, it needs to be more like the monkey
typing Shakespeare’s complete works than the monkey typing ‘jshdfoihne’, and
on the one hand, rational life needs to challenge our presumptions of how it
cameabout,andontheotherhand,weneedaglimpseofapossible,butnottoo
implausible explanation of it. In the monkey case, the monkey typing
Shakespeare’scompleteworkschallengesourpresumptionsofhowthattyping
came about (namely, roughly by chance), and the possible, but not too
implausible explanation might be that the typewriter is rigged, or perhaps that
themonkeyisverywelltrainedintypingShakespeare’sworks,insteadofitall
7ThereisalsoanaccountinGrimm(2008),accordingtowhichasituationstands
inneedofanexplanationforsomesubjectinvirtueofthatsubject’ssenseof
variousalternativewaysforthatsituationtohavebeen.But,asfarasIcantell,
thiscriterioniswaytooweaktobeofinterest.Mycoffeemugisinacertainplace
onmydesk,andIhaveasenseofalternativeplacesonthedeskforittopossibly
havebeen,butthesituationisinnoneedofanexplanation(tome).Many
mathematical,metaphysical,physicalandpsychologicalfactsareinneedofan
explanation(tome),butoftenIhavenosenseofalternativewaysforthemto
havebeen.
10
being due to chance. Both conditions thus seem satisfied in the case of the
monkey. In the case of rational life, hypothesis F, i.e. how improbable it is that
the universe contains rational life, challenges our scientifically informed
presumptionsofhowrationallifecameabout(namely,roughlybychance),and
the possible, but not too implausible explanation might be that it is all
intentionally created, instead of being due to chance. So both of the above
necessary conditions seem satisfied in the case of rational life as well. And the
existence of rational lifedoes seem utterly mysterious to me, so it does cry out
foranexplanation,atleasttome.
An obvious objection at this point would be that, yes, at some stage of
inquiry,rationallifewasinneedofanexplanation,butnolonger:wehavefound
itsbestexplanationintheDarwiniantheoryofnaturalselection.Giventhethree
concepts of arbitrary mutations (genetic changes)8, genetic inheritance (from
one generation to the next), and natural selection (roughly, environmental
pressure towards survival of the fittest), we can explain how rational life has
developedovertime:roughly,thechainofarbitrarygeneticchangestransmitted
throughgenerationsending(sofar)inourrationallifewasasufficientlygoodfit
for survival along the way. So, the initial objection stands unrefuted: at the
present stage of inquiry, U is in no more need of an explanation than Ui. The
theoryofnaturalselectionshowsthatthereisnothingspecialaboutrationallife
thatcriesoutforanexplanation.
But, as already pointed out by Paley (1802:11-15), this kind of reply
providesthewrongkindofexplanation.Itprovidesapurelycausalexplanation
ofhowrationallifecametobe,notajustificatoryexplanationofwhyrationallife
cametobeinthefirstplace.Obviously,itisthelatterthatputsanexplanatory
demandonusatthepresentstageofinquiry.Inotherwords,whatcriesoutfor
anexplanationiswhythecausalhistoryoftheuniverseshould(sofar)endup
withrationallife,nothowitactuallyhasmadethatrationallifecomeabout.In
yetotherwords,whatcriesoutforanexplanationiswhythewatchmakermade
awatch,nothowhedidit.ADarwiniantheoryofnaturalselectionprovidesthe
8Notethataccordingtoevolutionarytheory,thecurrentexistenceofrationallife
isnotateverysteparesultofpureunconditionalchance,butmorelike
accumulativeconditionalchance;butnoneofthismattersmuchtoour
argument.
11
latter, not the former, so the former still cries out for an explanation, not the
latter.9
Being more precise about why rational life needs a justificatory
explanationisnotoriouslyhard;infact,itissohardthatincoldermomentsit’s
temptingtoblameitonwishfulthinking,onakindofdelusion,andthusaccept
theinitialobjection.Butmaybewecanseejustabitbetterwhythelattermight
beamistakebyconsideringourtoymodelfromsection1,namelyU20.Wesaw
thatthespaceU20had101100manypoints,onlyoneofwhichwasspecialbybeing
“lit up”, namely U. Now, assume (i) that a machine operating at random – a
randomizer–issettoactualizeoneandonlyonepointinU20;(ii)thateachpoint
Ui in U20 is equally likely to be actualized (this is the Principle of Indifference
again); and (iii) that one and only one point in U20, namely U, is the point that
shouldbeactualized(maybejustbecauseit’stheoneandonlypointinU20that
hassomeintrinsicvalue;ormaybe,asIaminclinedtothink,becauseitistheone
andonlypointthathassomemeaning).Theseassumptionsamountstotheodds
fortheoneandonlypointUthatshouldbeactualizedactuallybeingactualized
being 1:101100. We are and should be surprised upon discovering that U in fact
wasthepointbeingactualized.Ourpresumptionsforhowitcameabout,namely
(i)-(ii), is challenged by the incredible odds against the point that should be
actualized in fact actualized together with the fact that we have a possible, but
not too implausible explanation for it. It’s sort of as if you wonder which of
Shakespeare’s plays is the most famous one, and when you put the monkey in
front of the typewriter he types out Hamlet (or perhaps Romeo and Juliet?) in
oneshot.Itseemshard,ifnotimpossibletoshrugthatoffasamerecoincidence
inneedofnomoreofanexplanationthanifthemonkeyhadtyped‘jshdfoihne’.
Onemightobjectthattheabovethoughtexperimentisincoherentdueto
assumption(i)beingincoherent:thepointoftheimaginedrandomizermakesno
9Itisalsoworthnoting,asindirectlypointedoutbyPaleyalready(butoften
seemstohavegonemissingeversinceDarwin),thatevolutionarytheoryjust
cannotprovidesuchajustificatoryexplanationbecausethequestioniswhy
rationallifeshoulddeveloptobeginwith(understoodnon-causally!),which
evolutionarytheorysaysnothingabout,andwiselyso.Somemight,ase.g.
Dawkins(2006)does,simplyrejecttheneedforsuchajustificatoryexplanation,
andoptforchanceinstead,butthattakesusbacktowhetherrationallifeisin
needofanexplanationbeyondpurechance,asdiscussedabove.
12
sensebecausepriortoouruniversetherewasnothing,andhencenoalternatives
amongwhichUwas“selected”.Butsuchanobjectionhasalreadybeenanswered
above: the probability in question is epistemic (non-objective) probability, and
thealternativesarethevariousvalues(numbers)theconstants,lawsandinitial
conditions could(theoretically) have taken on. Given epistemic probability, the
thought experiment makes as much sense as any epistemic probability
assignment.
One might also object to the fact that one and only one point should be
actualized,i.e.denythatrationallifeshouldbeactualizedovernon-rationallife,
or over inanimate matter for that matter. If the normativity in question is
subjective,dependentonussomehow,dependentontheveryonesthatmakethe
epistemicassessmentofprobabilityinquestion,thenitmightbefurtherclaimed
that the specialness that raises the demand of explanation, i.e. the value of
rationallife,isalsosubjective,dependentonussomehow,andhencethatinitself
UisnomoreinneedofanexplanationthanUi.Inshort,justbecausearational
life finds itself qua rational life very special, it doesn’t follow that a universe
containingsuchrationallifeisveryspecial;10thatUistheoneandonlyuniverse
thatshouldberealizedmightallbeaself-conceiteddelusion.Butthentheinitial
objectionstillstandsstrong:UisinnospecialneedofexplanationoverUi.
Theargument1-3doesnotdependonrationallifebeingwhatobjectively
shouldberealizedovernon-rationallife.Theargumentonlydependsonrational
lifeasopposedtonon-rationallifebeingsomethingthatweasinquirersassess
asbeinginneedofanexplanation.Onewayforittobeinneedofanexplanation
isforittobetheoneandonlypointinU20thatshouldberealized,butitisnotthe
onlywayforittobeinneedofanexplanation.Itmightsufficetojustseethatitis
special,notatalllikeitscomplement,i.e.thatrationallifeissomethingspecialas
opposedtonon-rationallife,somethingthatwefindcannotbeleftunexplained,or
lefttomerechance.Wefindsuchcasesinscienceandoureverydaylivesallthe
time, and we seek explanations of such facts, we don’t leave them to mere
chance. There is no presumption in those cases that they need to be what
objectively should be the case, so there need not be such a presumption in the
case of rational life either. It is enough that we epistemically assess them as
10JustlikemefindingmyselfveryspecialdoesnotentailthatIamveryspecial.
13
beinginneedofanexplanation,notlefttomerechance.Butofcoursewewould
like an account of why and in what sense they are thus special, in need of an
explanation;thattheyaretheonlycandidateforwhatobjectivelyshouldbethe
caseisonepossiblesuchaccount,thoughnottheonlyone.
Now,ofcourse,evengrantingthatrationallifedoescryoutforanexplanation,
there might be other non-ad hoc alternative explanations for it that does not
invoke intentional design, which brings us over to the second more serious
objectiondiscussedinthissection.
Secondobjection:arguably,everyUiactuallyexist,justasU,soneitheroneofUi
andUisanymoreorlessprobablethantheother;infact,sincetheyallexist,ina
sense,theyallhaveprobability1.Let’sfollowcustomandcallthisthemultiverse
hypothesis, or simply M. We should also not find U any more surprising or in
needofanexplanationthananyUibecausewehavenochoicebuttoexperience
onlyU.WecalltheclaimthatweshouldnotfindUanymoresurprisingthanUi
the observation selection effect, or simply O. Now, given M and O, there is no
reasontoinvokehypothesisI,because,plausibly,Fismoreprobableconditional
onMthanonI,andwithOonboardthereisnothinglefttoexplain.Furthermore,
Misalivetheoryincontemporaryphysics,postulatedongroundsindependent
of F, so it cannot be excluded as an alternative explanation on the grounds of
being objectionably ad hoc. M is simply proposed as a better explanation of F
thanI,bothsuperiortoC.So,inotherwords,wecanrunaparallelversionofthe
argument1-3whereMandIarethecompetitorsratherthanIandC,andcome
toseethatMwillwinoverI.So,thoughIisbetterthanCbyvirtueof1-3,Mis
likewisebetterthanI;hence,weshouldbelieveMoverbothIandC.
Reply:Mismoretheoreticallyspeculativethanempiricallyverified,butso
is I;11M is, in one sense, extravagant, but so is I; and M is in many ways
incredible,butsoisI.Therefore,itwouldbeunwisetotrytoruleoutM(orI)on
anyofthesegrounds.Moregenerally,itwouldbeunwisetotrytoarguedirectly
against M on a priori grounds because for all we currently know M might be
11Admittedly,Idon’tthinkthevariouslyexistingtestimoniesastodirect
encounterswithGodcountsformuchhere.
14
established on more or less a posteriori grounds, 12 and if so, the a priori
argumentshouldbesacrificedinfavoroftheaposterioriargument.
Ofcourse,Fdoesnotallbyitselfgiveanysignificantevidenceinfavorof
M, but neither does F all by itself give any significant evidence in favor of I.
Rather, F raises the likelihood of either M or I, given independent evidence for
eitherMorI.So,itisnotsomuchthatwehavereasontobelieveinM(orI),given
F, but rather that given M (or I), we can explain F, which in turn raises the
likelihood of M (or I). That is, in compliance with what we saw in the fourth
objectioninsection2above:wemusthaveindependentevidenceinfavorofM
orIinorderfortheargumenttobesufficientlyinteresting.So,inonesense,the
questionbecomes:whichoneofMandIhasthemostevidenceindependentlyof
F?Tothenon-fanaticsamongus,theanswertothatquestionshouldnotbeatall
obvious.Personally,Iforsurehavenoempiricalevidenceinfavorofhypothesis
I,butIknowotherswhoclaimtohavehadempiricalencounterswithGod,which
in turn could be taken to be empirical evidence for hypothesis I. On the other
hand,nooneevenclaimstohaveanyempiricalevidenceinfavorofM.Allagree
that,asoftoday,Mispuretheoreticalspeculation.13Ofcourse,thatisnottosay
that those theoretical speculations are not good theoretical speculations,
speculationsthatmighteventuallybeempiricallysupported;it’sjusttosaythat
theyhavenoempiricalsupportyet.So,Ibelievetheanswertothequestionasto
whichoneofMandIhavethemostindependentevidenceisfarfromobvious.I
personallydon’tbelieveinempiricalencounterswithGod,thoughothersdo,but
thereisalsoasoftodaynoempiricalevidenceinfavorofM,so,asfarasIcansee,
asoftoday,itisperhapsadrawwhenitcomestoindependentempiricalsupport.
Now,judgingbythehistoryofphysics,whereeachepochseemstohave
increased the size and extent of the universe, it should not surprise us if M
turnedouttobethenextstepinthesamedirection.Weshouldnothavethought
thatoursolarsystemistheonlysolarsystem,andweshouldnothavethought
12If,forexample,wefoundGUT(TheGrandUnifiedTheory),andGUTlogically
impliedamultiverse.Cf.Rees(2003).
13Cf.thevariousessaysinCarr(2009).
15
thatourgalaxyistheonlygalaxy,sowhyshouldwethinkthatouruniverseisthe
onlyuniverse?14
On the other hand, M might not matter to the issue at hand, namely
whether we should believe I. Because assume M is true, and sufficiently
supportedindependentlyofF;perhapsbytherebeingsomegrandunifiedtheory
(GUT)ofeverythingthatentailsthatthereareverymany,ifnotinfinitelymany
Big Bangs, all with different laws l, constants c, and/or initialconditionsci. But
howexactlywouldthatexplainU,asauniversewithrationallife,tobeginwith?
First, unless we can provide some appropriate connection between the
existenceofthedifferentuniverses,wewouldnotexplainwhythereisrational
life in our universe; just like having rolled five double six in a row does not
explain the outcome of this next roll (this is the so-called Gambler’s Fallacy15).
But then again maybe we eventually can provide the appropriate connection.
Still,thereisanotherproblem.
Second,ifallthetheoreticallypossibleuniversesUiexist,thenofcourseU
existstoo,butUdemandsanexplanationoverUitotheextentthattherational
lifeitcontainsdemandsanexplanation,butMjustdoesnotexplainrationallife.
Atbest,likenaturalselection,Mhelpsexplainhowrationallifecameabout,but
not why it came about to begin with, which is what we want. If rational life as
such is in some sense special in comparison to its complement, if rational life
morethanitscomplementissomehowcryingoutforanexplanation,thenitjust
doesnothelptoexpanditscomplement.Thecomplementofrationallifecan,as
far as I can see, be of no use in explaining rational life itself, and a mere
expansion of it can certainly not be of any such use. So, short of rejecting that
rationallifeisspecialcomparedtoitscomplement,orinneedofanexplanation,I
conclude that M just does not help explaining rational life as such. But the
problem now is that rational life as such is pulling all the strings; F is neither
herenorthere.That,itseemstome,weakensthecaseforI.Ifallwehaveisthat
rational life somehow seems special, then maybe what we should look for is
therapy,notexplanation.Still,thereisafinalpotentialproblem.
14Assuming,ofcourse,that’universe’doesnotjustmeanallthatexist,inwhich
caseitistrivialthatthereisonlyonesuchthing.
15Cf.White(2000).
16
Namely,third,whatever“mechanism”causesalltheuniversesUitoexist
mustalso befine-tuned;otherwise,rationallifewouldbeimpossible.Ofallthe
logically or metaphysically possible such mechanisms there could have been,
whythisonethatcausedrationallifeasweknowit?16
I conclude that the argument 1-3 from section 1 stands surprisingly strong.
WithoutMonthetable,FfavorsIoverC.WithMonthetable,Ffavorswhichever
one of M and I have the most independent evidence. But also, with M on the
table, it might be that rational life as such still cries out for an explanation
independent of our explanation of F. What’s more, M might just push the
questionoffine-tuningonestepbackratherthanexplainittobeginwith.After
all,wearelookingforanexplanationofwhyrationallifeexiststobeginwith,so
noexplanationdependingonarbitrarinessisreallysatisfying.17
BIBLIOGRAPHY
-
Carr,B.(ed.)(2009).UniverseorMultiverse?CambridgeUniversityPress.
-
Collins,R.(2012).Theteleologicalargument:aninvestigationofthefinetuningoftheuniverse.AsinCraig&Moreland(2012).
-
-------------(2003).Evidenceforfine-tuning.AsinManson(2003).
-
Craig, W. L. & Moreland, J. P. (eds.) (2012). Natural Theology. WileyBlackwell.
-
Dawkins,R.(2013).TheBlindWatchmaker.Penguin.
-
Grimm, S. R. (2008). Explanatory enquiry and the need for explanation.
Britishjournalforthephilosophyofscience,Vol.59,No.3.
-
Horwich,P.(1982).Probabilityandevidence.CambridgeUniversityPress.
-
Leslie,J.(1989).Universes.Routledge.
-
Manson,N.A.(ed.)(2003).GodandDesign.Routledge.
-
Paley,W.(1802).NaturalTheology.OxfordWorld’sClassics,2006.
16SeeCollins(2012).
17ThankstoJeffDunnandtheaudienceattheNorwegianSchoolofTheologyfor
comments.
17
-
Rees, M. (2003). Other universes: a scientific perspective. As in Manson
(2003).
-
Sober,E.(2003).Thedesignargument.AsinManson(2003).
-
White,R.(2003)Fine-tuningandmultipleuniverses.InManson(2003).
-
Williamson,T.(2000).Knowledgeanditslimits.OxfordUniversityPress.
18