Journal of Philosophy, Inc.

Quantum Mysteries for Anyone

Author(s): N. David Mermin
Source: The Journal of Philosophy, Vol. 78, No. 7 (Jul., 1981), pp. 397-408
Published by: Journal of Philosophy, Inc.
Stable URL: [Link] .
Accessed: 11/01/2014 21:10

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
[Link]

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@[Link].

Journal of Philosophy, Inc. is collaborating with JSTOR to digitize, preserve and extend access to The Journal
of Philosophy.

[Link]

This content downloaded from [Link] on Sat, 11 Jan 2014 [Link] PM

All use subject to JSTOR Terms and Conditions
 SCENES AND OTHER SITUATIONS 397

logic); others might see it as injunction against speaking in the

way we do in natural language, of arbitraryscenes and situations.
JON BARWISE
StanfordUniversity
Universityof Wisconsin/Madison

QUANTUM MYSTERIES FOR ANYONE

Weoftendiscussedhisnotionson objectivereality.I recallthatdur-
ing one walk Einsteinsuddenlystopped,turnedto me and asked
whether I reallybelievedthatthemoonexistsonlywhenI lookat it.
A. Pais'
As 0. Sternsaid recently, one should no morerackone's brain
about theproblemof whether somethingone cannotknowanything
about existsall the same, thanabout theancientquestionof how
manyangelsareable tositon thepointofa [Link] seemstome
thatEinstein'squestionsare ultimately
alwaysof thiskind.
W. Pauli2
AULI and Einsteinwereboth [Link] questionswith
which Einstein attacked the quantum theory do have
answers; but theyare not the answers Einstein expectedthem
to have. We now know that the moon is demonstrablynot there
when nobody looks.
The impact of this discovery on philosophy may have been
blunted by the way in which it is conventionally stated, which
leaves it fullyaccessible only to thosewith a workingknowledgeof
quantum mechanics. I hope to remove that barrierby describing
this remarkable aspect of nature in a way that presupposes no
backgroundwhateverin the quantum theoryor, forthatmatter,in
classical physics either. I shall describea piece of machinerythat
presentswithoutany distortionone of the most strikinglypeculiar
featuresof the atomic world. No formal training in physics or
mathematicsis needed to grasp and ponder the extraordinarybe-
havior of the device; it is only necessaryto follow a simple count-
ing argumenton the level of a newspaper braintwister.
Being a physicist,and not a philosopher, I aim only to bring
home some strangeand simple factswhich mightraise issues phil-
osophers would be interestedin addressing. I shall try,perhaps
' Reviews of Modern Physics,Li, 863 (1979): 907.
2From a 1954 letterto M. Born, in The Born-EinsteinLetters(New York: Walker,
1971), p. 223.
0022-362X/81/7807/0397$01.20 ? 1981 The Journal of Philosophy, Inc.

This content downloaded from [Link] on Sat, 11 Jan 2014 [Link] PM

All use subject to JSTOR Terms and Conditions
398 JOURNAL OF PHILOSOPHY

without notable success, to avoid raising and addressing such

issues [Link] I describe should be regardedas something
betweena parable and a [Link] is less than a lec-
turedemonstrationfortechnicalreasons:even if thiswerea lecture,
I lack the time, money, and particular expertise to build the
machineryI shall describe. It is more than a parable because the
device could in fact be built with an effortalmost certainlyless
than, say, the Manhattan project, and because the conundrum
posed by the behavior of the device is no mere analogy, but the
atomic world itself,acting at its most perverse.
There are some black boxes within thedevice whose contentscan
be describedonly in highly technical [Link] is of no impor-
tance. The wonder of the device lies in what it does, not in how it
is put togetherto do it. One need not understandsilicon chips to
learn fromplaying with a pocket calculator thata machine can do
arithmeticwith superhuman speed and precision; one need not
understandelectronicsor electrodynamicsto grasp thata small box
can imitatehuman speech or an [Link] theend of theessay I
shall give a brief technical description of what is in the black
boxes. That description can be skipped. It is thereto serveas an
existenceproofonly because you cannot buy thedeviceat thedrug-
[Link] is no more essential to appreciating the conundrumof the
device than a circuitdiagram is to using a calculator or a radio.
The device has threeunconnected parts. The question of con-
nectednesslies near the heart of the conundrum,but I shall set it
aside in favorof a fewsimple [Link] are neither
mechanical connections (pipes, rods, strings,wires) nor electro-
magneticconnections(radio, radar,telephoneor light signals) nor
any [Link] be hard
to avoid. All threepartsmight,forexample, sit atop a single table.
There is nothing in the design of the parts,however,thattakesad-
vantage of such connectionsto signal fromone to another,forex-
ample, by inducing and detectingvibrationsin the table top.
By insisting so on the absence of connections I am inevitably
suggestingthat the wondersto be revealedcan be fullyappreciated
only by expertson connectionsor theirlack. This is not the right
attitudeto take. Werewe togetherand had I thedeviceat hand, you
could pick up the parts,open themup, and poke around as much
as you liked. You would find no connections. Neitherwould an
experton hidden bugs, the Amazing Randi, or any physicistsyou
called in as [Link] real worryis unknown connections.
Who is to say that the parts are not connectedby the transmission
of unknown Q-rays and their detection by unrecognizable Q-
detectors?One can only offeraffidavitsfromthemanufacturertesti-

This content downloaded from [Link] on Sat, 11 Jan 2014 [Link] PM

All use subject to JSTOR Terms and Conditions
 QUANTUM MYSTERIES FOR ANYONE 399
fyingto an ignorance of Q-technologyand, in any event,no such
intent.
Evidentlyit is impossible to rule out conclusivelythepossibility
of [Link] proper point of view to take,however,is that
it is preciselythe wonderand gloryof the device thatit impels one
to doubt these assurances fromone's own eyes and hands, profes-
sional magicians, and technical expertsof all kinds. Sufficeit to
say that thereare no connections thatsuspicious lay people or ex-
pertsof broad eruditionand unimpeachable integritycan discern.
If you find yourselfquestioning this, then you have grasped the
mystery of the atomic world.
Two of the threepartsof the device (A and B) functionas detec-
[Link] detectorhas a switchthatcan be set in one of threeposi-
tions (1, 2, and 3) and a red and a greenlightbulb (Fig. 1). When a
detectoris set offit flashes eitherits red light or its green. It does
this no matterhow its switchis set, though whetherit flashesred
or green may well depend on the [Link] only purpose of the
lights is to communicate informationto us; markson a ribbon of
tape would serveas well. I mention thisonly to emphasize thatthe
unconnectednessof the parts prohibitsa mechanism in eitherde-
tectorthat might modifyits behavior according to the color that
may have flashedat the other.

2
R
~~~3

Fig. 1. A detector.
Particlesenteron [Link] red(R) andgreen
(G) lightsareon [Link] switchis setto 1.

The thirdand last part of the device is a box (C) placed between
the [Link] button on the box is pushed, shortly
thereaftertwo particles emerge,moving offin opposite directions
towardthe two detectors(Fig. 2). Each detectorflasheseitherredor
green whenevera particle reaches it. Thus within a second or two

This content downloaded from [Link] on Sat, 11 Jan 2014 [Link] PM

All use subject to JSTOR Terms and Conditions
400 JOURNAL OF PHILOSOPHY

of everypush of the button,each detectorflashesone or the other

of its two colored lights.

(A) (C) (B)

Fig. 2. The completedevice.A and B are thetwodetectors.
C is the
box fromwhichthetwoparticlesemerge.

Because thereare no connectionsbetweenpartsof thedevice, the

link between pressing the button on the box and the subsequent
flashingof the detectorscan be providedonly by the passage of the
particlesfromthe box to the [Link] passage could be con-
firmedby subsidiarydetectorsbetweenthe box and the main detec-
torsA and B, which can be designed so as not to alter thefunction-
ing of the device. Additional instrumentsor shields could also be
used to confirmthe lack of othercommunication betweenthe box
and the two detectorsor between the detectorsthemselves(Fig. 3).

CR VEIl
2

of [Link] box is embeddedin

Fig. 3. Possiblerefinement
a wall thatcutsoffone detector
[Link] detectors
confirm thepassageof theparticlesto themaindetectors.

The device is operated repeatedlyin the following way. The

switchon each detectoris set at random to one of its threepossible
positions, giving nine equally likely settingsforthe pair of detec-
tors: 11, 12, 13, 21, 22, 23, 31, 32, and 33. The buttonon the box is

This content downloaded from [Link] on Sat, 11 Jan 2014 [Link] PM

All use subject to JSTOR Terms and Conditions
 QUANTUM MYSTERIES FOR ANYONE 401

then pushed, and somewhat later each detectorflashes one of its

lights. The flashingof the detectorsneed not be simultaneous. By
changing thedistance betweenthe box and thedetectorswe can ar-
range [Link] can also let the switchesbe given
theirrandom settingseitherbeforeor afterthe particles leave the
box. One could even arrangeforthe switchon B not to be set until
afterA had flashed(but, of course, beforeB flashed).
Afterboth detectorshave flashed theirlights,the settingsof the
switchesand the colors thatflashedare recorded,using the follow-
ing notation: 31 GR means thatdetectorA was set to 3 and flashed
green,while B was set to 1 and flashedred; 12 RR describesa run in
which A was at 1, B at 2, and both flashed red; 22 RG describesa
run in which both detectorswere set to 2, A flashed red and B
flashedgreen;and so on. A typicalfragmentfroma recordof many
runs is shown in Fig. 4.

_uG 2ZGGi
,2RR 31RG 13RG 22GG 221
i7R 21GR 32RG 11GG 32GR 33GG 21
22GG IIRR 1CGG 23GG 12RR 32GR 11CG
G 12RG 13RG 33GG 21RG 13GR 31RR 32GR
,GR 13GR 21RG 33RR 13GR IIRR 1CGG 13RG 31
.2GG 32GR 33GG 21GR 21GG 33RR 23RG 21GG 21h
13GR 11GG 32GG 31GR 32RG 33RR 13RR 13RG 12R'
11GG 31RG 33RR 12RG 21GR 11GG 22GG 33GG 23GC
11RR 22RR 12RG 22GG 23GR 12GR 33GG 31GG 13GC
13GR 21RR 33RR 33RR 13RG 23RG 33GG 32RR 12R
3RR 32RG IIRR 11RR 11RR 32RG 12RG 21RG lIG
RG 23RR 21RG 33RR 13GR 12GR 23RG 21RR 32
R 21GR 12RR 31GR 12RG 13GR 13RG 22RR I
23GR 11RR 12RR 33RR 21RG 13GR 21RP
"R 12RR 23GG 13RG 21RG 1CGG 12
'2RG 32RG 32GR 11GG 22W
^n31RG 7
Fig. 4. Fragment
of a page of a volumefromtheset of notebooks
recordinga long seriesofruns.

This content downloaded from [Link] on Sat, 11 Jan 2014 [Link] PM

All use subject to JSTOR Terms and Conditions
402 JOURNAL OF PHILOSOPHY

The accumulated data have a random character,but, like data

collected in many tossingsof a coin, theyreveal certainunmistak-
able featureswhen enormouslymany runs are examined. The sta-
tisticalcharacterof the data should not be a source of concern or
suspicion. Blaming the behavior of the device on repeated,syste-
matic, and reproducibleaccidents,is to offeran explanation even
more astonishingthan the conundrumit is invoked to dispel.
The data accumulated over millions (or, if you prefer,billions or
trillions)of runs can be summarizedby distinguishingtwo cases:
Case a. In those runs in which each switch ends up with the
same setting(11, 22, or 33) both detectorsalways flash the same
color. RR and GG occur in a random patternwith equal frequency;
RG and GR neveroccur.
Case b. In the remaining runs, those in which the switchesend
up with differentsettings(12, 13, 21, 23, 31, or 32), both detectors
flash the same color only a quarter of the time (RR and GG occur-
ring randomly with equal frequency);the other threequarters of
the time the detectorsflash differentcolors (RG and GR occurring
randomlywith equal frequency).
These resultsare subject to the fluctuationsaccompanying any
statisticalpredictions,but, as in the case of a coin-tossingexperi-
ment, the observed ratios will differless and less fromthose pre-
dicted,as the numberof runs becomes largerand larger.
This is all it is necessaryto know about how thedevice operates.
The particularfractionsI and 3 arising in case b are of criticalim-
portance. If the smaller of the two were I or more (and the larger2
or less) therewould be nothing wonderfulabout the device. To
produce theconundrumit is necessaryto run theexperimentsuffi-
cientlymany timesto establishwithoverwhelmingprobabilitythat
the observedfrequencies(which will be close to 25% and 75%) are
not chance fluctuationsaway fromexpected frequenciesof 333%
and 662%. (A million runs is more than enough forthis purpose.)
These statisticsmay seem harmless enough, but some scrutiny
reveals themto be as surprisingas anythingseen in a magic show,
and leads to similar suspicions of hidden wires,mirrors,or confed-
eratesunder the [Link] begin by seeking to explain why the de-
tectorsinvariablyflashthesame colors when theswitchesare in the
same positions (case a). There would be any numberof ways to ar-
range this were the detectorsconnected,but theyare not. Nothing
in the constructionof eitherdetectoris designed to allow its func-
tioning to be affectedin any way by thesettingof the switchon the
other,or by the color of the light flashedby the other.
Given the unconnectednessof the detectors,thereis one (and, I

This content downloaded from [Link] on Sat, 11 Jan 2014 [Link] PM

All use subject to JSTOR Terms and Conditions
 QUANTUM MYSTERIES FOR ANYONE 403
would think,only one) extremelysimple way to explain the behav-
ior in case a. We need only suppose that some propertyof each
particle (such as its speed, size, or shape) determinesthe color its
detectorwill flash foreach of the threeswitchpositions. What that
propertyhappens to be is of no consequence; we require only that
the various statesor conditions of each particlecan be divided into
eight types;RRR, RRG, RGR, RGG, GRR, GRG, GGR, and GGG. A parti-
cle whose stateis of typeRGG, forexample, will always cause its de-
tectorto flash red for setting 1 of the switch,green for setting2,
and green forsetting3; a particle in a stateof typeGGG will cause
its detectorto flash green forany settingof the switch;and so on.
The eight typesof statesencompass all possible cases. The detector
is sensitive to the state of the particle and responds accordingly;
puttingit anotherway, a particlecan be regardedas carryinga spe-
cificset of flashinginstructionsto its detector,dependingon which
of the eight statesthe particleis in.
The absence of RG or GR when the two switcheshave the same
settingscan then be simply explained by assuming that the two
particles produced in a given run are both produced in the same
state; i.e., theycarryidentical instructionsets. Thus if both parti-
cles in a run are produced in statesof typeRRG, thenboth detectors
will flash red if both switchesare set to 1 or 2, and both will flash
green if both switchesare set to 3. The detectorsflash the same
colors when the switcheshave the same settingsbecause the parti-
cles carrythe same instructions.
This hypothesisis the obvious way to account forwhat happens
in case a. I cannot prove thatit is theonly way, but I challenge the
reader,given the lack of connectionsbetweenthe detectors,to sug-
gest any other.
The apparent inevitabilityof this explanation for the perfect
correlationsin case a formsthe basis for the conundrum posed by
the device. For the explanation is quite incompatible with what
happens in case b.
If thehypothesisof instructionsetswerecorrect,thenboth parti-
cles in any given run would have to carryidentical instructionsets
whetheror not the switcheson the detectorswere set the same. At
the moment the particles are produced thereis no way to know
how the switches are going to be set. For one thing, thereis no
communication between the detectorsand the particle-emitting
box, but in any event the switchesneed not be set to theirrandom
positions until afterthe particles have gone offin opposite direc-
tions fromthe box. To ensure thatthedetectorsinvariablyflashthe
same color everytime the switchesend up with the same settings,

This content downloaded from [Link] on Sat, 11 Jan 2014 [Link] PM

All use subject to JSTOR Terms and Conditions
404 JOURNAL OF PHILOSOPHY

the particles leaving the box in each run must carrythe same in-
structionseven in those runs (case b) in which the switchesend up
with different settings.
Let us now consider the totalityof all case b runs. In none of
themdo we everlearn what the full instructionsetswere,since the
data reveal only the colors assigned to two of the threesettings.
(The case a runs are even less informative.)Neverthelesswe can
draw some nontrivialconclusions by examining the implications
of each of the eight possible instructionsets for those runs in
which the switchesend up with different [Link], forex-
ample, that both particles carrythe instructionset RRG. Then out
of the six possible case b settings,12 and 21 will resultin both de-
tectorsflashing the same color (red), and the remaining four set-
tings, 13, 31, 23, and 32, will resultin one red flash and one green.
Thus both detectorswill flash the same color for two of the six
possible case b [Link] the switch settingsare completely
random, the various case b settingsoccur with equal frequency.
Both detectorswill therefore flash the same color in a thirdof those
case b runs in which the particlescarrythe instructionsets RRG.
The same is trueforcase b runs where theinstructionset is RGR,
GRR, GGR, GRG, or RGG, since the conclusion restsonly on the fact
that one color appears in the instructionset once and the other
color, twice. In a thirdof the case b runs in which the particles
carryany of theseinstructionsets,thedetectorswill flash the same
color.
The only remaining instructionsets are RRR and GGG; for these
sets both detectorswill evidentlyflash the same color in every
case b run.
Thus, regardless of how the instruction sets are distributed
among the different runs, in the case b runs both detectorsmust
flash thesame color at least a thirdof the time.(This is a bare min-
imum; the same color will flash more than a thirdof the time,un-
less the instructionsets RRR and GGG neveroccur.) As emphasized
earlier,however,when the device actually operates the same color
is flashedonly a quarterof the timein the case b runs.
Thus the observedfactsin case b are incompatible with the only
apparent explanation of the observed facts in case a, leaving us
with theprofoundproblem of how else to account forthe behavior
in both cases. This is theconundrumposed by thedevice,forthere
is no other obvious explanation of why the same colors always
flashwhen theswitchesare set the same. It would appear thatthere
must, afterall, be connectionsbetween the detectors-connections
of no known descriptionwhich serveno purpose otherthan reliev-

This content downloaded from [Link] on Sat, 11 Jan 2014 [Link] PM

All use subject to JSTOR Terms and Conditions
 QUANTUM MYSTERIES FOR ANYONE 405

ing us of the task of accounting for the behavior of the device in

theirabsence.
I shall not pursue this line of thought,since my aim is only to
state the conundrum of the device, not to resolve it. The lecture
demonstrationis over. I shall only add a fewremarkson thedevice
as a parable.
One of the historic exchanges between Einstein and Bohr,3,4
which found its surprisingdenouement in the work of J. S. Bell
nearly threedecades later,5can be statedquite clearly in termsof
thedevice. I stressthat the transcriptioninto the contextof thede-
vice is only to simplifytheparticularphysical arrangementused to
raise theissues. The device is a directdescendantof therathermore
intricatebut conceptually similar gedanken experimentproposed
in 1935 by Einstein, Podolsky, and Rosen. We are still talking
physics,not descendingto the level of analogy.
The Einstein, Podolsky, Rosen experimentamounts to running
the device under restrictedconditions in which both switchesare
required to have the same setting(case a). Einsteinwould argue (as
was argued above) that the perfectcorrelationsin each run (RR or
GG but neverRG or GR) can be explained only if instructionsetsex-
ist, each particle in a run carryingthe same [Link] the
Einstein,Podolsky,Rosen versionof the argumenttheanalogue of
case b was not evident,and its fatalimplicationsforthehypothesis
of instructionsets went unnoticeduntil Bell's paper.
The gedanken experimentwas designed to challenge theprevail-
ing interpretationof the quantum theory,which einphatically
denied theexistenceof instructionsets,insistingthatcertainphysi-
cal properties(said to be complementary)had no meaning inde-
pendent of the experimental procedure by which they were
measured. Such measurements,far fromrevealing the value of a
preexistingproperty,had to be regardedas an inseparable part of
the veryattributetheywere designed to measure. Propertiesof this
kind have no independentrealityoutside the contextof a specific
experimentarranged to observe them: the moon is not therewhen
nobody looks.
In the case of my device, threesuch propertiesare involved for
each particle. We can call them the 1-color,2-color,and 3-colorof
the particle. The n-color of a particle is red if a detectorwith its
switch set to n flashes red when the particle arrives. The three
n-colorsof a particleare [Link] switchon
3A. Einstein,B. Podolsky,and N. Rosen, Physical Review, xiViI,777 (1935).
4N. Bohr, Physical Review, XLVIII, 696 (1935).
'J. S. Bell, Physics,i, 195 (1964).

This content downloaded from [Link] on Sat, 11 Jan 2014 [Link] PM

All use subject to JSTOR Terms and Conditions
406 JOURNAL OF PHILOSOPHY

a detectorcan be set to only one of the threepositions, and the

experimentalarrangementsformeasuringthe 1-,2-,or 3-colorof a
particle are mutually exclusive. (We may assume, to make this
point quite firm,that the particle is destroyedby theact of trigger-
ing the detector,which is, in fact,the case in many recentexperi-
mentsprobing the principles thatunderlythe device.)
To assume that instructionsets exist at all, is to assume that a
particle has a definite1-, 2-, and 3-color. Whetheror not all three
colors are known or knowable is not the point; the mere assump-
tion that all threehave values violates a fundamentalquantum-
theoreticdogma.
No basis forchallenging this dogma is evidentwhen only a sin-
gle particleand detectorare [Link] ingenuityof Einstein,
Podolsky,and Rosen lay in discoveringa situationinvolvinga pair
of particlesand detectors,where the quantum dogma continued to
deny the existence of 1-,2-, and 3-colors,while, at the same time,
quantumtheory
predicted
correlations
(RR and GG butneverRG or
GR) thatseemed to require theirexistence.
Einstein concluded that,if thequantum theorywerecorrect,i.e.,
if the correlationswere, as predicted,perfect,then the dogma on
the nonexistenceof complementaryproperties-essentially Bohr's
doctrineof complementarity-had to be rejected.
Pauli's attitudetoward this in his letterto Born is typicalof the
position taken by many physicists:since thereis no known way to
determineall threen-colorsof a particle,why waste your timear-
guing about whetheror not theyexist?To deny theirexistencehas
a certainpowerfuleconomy-why encumber the theorywith inac-
cessible entities?More importantly,the denial is supportedby the
formal-structure of the quantum theorywhich completelyfails to
allow forany considerationof the simultaneous 1-,2-,and 3-colors
of a particle. Einstein preferredto conclude thatall threen-colors
did exist, and that the quantum theorywas incomplete. I suspect
that many physicists,though not challenging the completenessof
the quantum theory,managed to live with the Einstein,Podolsky,
Rosen argumentby observingthat though therewas no way to es-
tablish the existenceof all threen-colors,therewas also no way to
establish [Link] theangels sit,even if theycan't be
counted.
Bell changed all this, by bringinginto considerationthe case b
runs,and pointing out thatthequantitativenumericalpredictions
of the quantum theory(4 vs. 3) unambiguouslyruled out the exist-
ence of all [Link] since Bell's paper con-

This content downloaded from [Link] on Sat, 11 Jan 2014 [Link] PM

All use subject to JSTOR Terms and Conditions
 QUANTUM MYSTERIES FOR ANYONE 407

firmthe quantum-theoreticpredictions.6Einstein's attack,were he

to maintain it today,would be more than an attack on the meta-
physical underpinningsof the quantum theory-more,even, than
an attack on the quantitative numerical predictionsof the quan-
tum [Link]'s position now appears to be contradictedby
nature [Link] device behaves as it behaves, and no mentionof
wave-functions,reductionhypotheses,measurementtheory,super-
position principles, wave-particleduality, incompatible observa-
bles, complementarity,or the uncertaintyprinciple, is needed to
bringhome its [Link] is not the Copenhagen interpretation
of quantum mechanics thatis strange,but the world itself.
As faras I can tell,physicistslive with theexistenceof thedevice
by implicitly (or even explicitly) denying the absence of connec-
tions betweenits pieces. Referencesare made to the "wholeness" of
nature:particles,detectors,and box can be consideredonly in their
totality;the triggeringand flashingof detectorA cannot be consi-
dered in isolation fromthe triggeringand flashingof detectorB-
both are part of a single indivisible process. This attitudeis some-
times tinged with Eastern mysticism,sometimes with Western
know-nothingism,but, common to eitherpoint of view, as well as
to the less trivialbut considerablymore obscure position of Bohr,
is the sense thatstrangeconnectionsare [Link] connectionsare
strangebecause theyplay no explicit role in the theory:theyare as-
sociated with no particlesor fieldsand cannot be used to send any
kinds of signals. They are thereforone and only one reason: to re-
lieve the perplexityengenderedby the insistencethat thereare no
connections.
Whetheror not thisis a satisfactory stateof affairsis, I suspect,a
question betteraddressedby philosophers than by physicists.
I conclude with the recipe for making the device, which, I em-
phasize again, can be ignored:
The deviceexploitsBohm'sversion7 of theEinstein,Podolsky,Rosen
experiment. The twoparticlesemerging fromthebox arespin'2parti-
cles in the [Link] two detectorscontainStern-Gerlach
magnets, and thethreeswitchpositionsdetermine whether theorien-
tationsof themagnetsare verticalor at ?120 to theverticalin the
plane perpendicular to theline of flightof [Link]

6Theoreticaland experimentalaspects of the subjectare reviewedby J. F. Clauser

and A. Shimony,Reports on Progressin Physics,XLI, 1991 (1978). For a less techni-
cal surveysee B. d'Espagnat, ScientificAmerican,CCXL, 5 (November 1979): 158.
7D. Bohm, Quantum Theory (Englewood Cliffs,N.J.: Prentice-Hall, 1951), pp.
614-619.

This content downloaded from [Link] on Sat, 11 Jan 2014 [Link] PM

All use subject to JSTOR Terms and Conditions
408 JOURNAL OF PHILOSOPHY

switcheshave the same settingsthe magnets have the same orienta-

tion. One detectorflashesred or greenaccording to whetherthemeas-
ured spin is along or opposite to the field; the otheruses theopposite
color [Link] when the same colors flash themeasuredspin
componentsare different.
It is a well-knownelementaryresult that,when the orientationsof
the magnets differby an angle 0, then the probabilityof spin meas-
urementson each particle yieldingopposite values is cos2(0/2). This
probabilityis unitywhen 0 = 0 (case a) and Xwhen 0 = ? 1200(case b).
If thesubsidiarydetectorsverifyingthepassage of theparticlesfrom
the box to the magnets are entirelynonmagnetic theywill not inter-
ferewith this behavior.
N. DAVID MERMIN
Physics, Cornell University

BOOK REVIEWS
Philosophical Perspectives in Artificial Intelligence. MARTIN D.
RINGLE, ed. New York: Humanities Press, 1979. x, 244 p. $17.50.
There is no gadget more characteristicof our age than the compu-
ter. Computers and all fields and organizations that serve them
arouse an interestthat seems insatiable. Prestigeand recognition
naturally follow. Among the alphabet soup of acronyms, how
many enjoy the recognitionaffordedIBM? So, when a discipline
appears thatpresentsitselfas the theoreticalarm of this new wave,
it is only natural to find the metatheoreticalimpulses of philos-
ophers aroused. This natural urge is further
compounded when the
discipline in question is generallyunencumberedby even thefaint-
est hint of [Link]-Al for short-has tan-
talized the philosophical community(as well as many psycholo-
gists and linguists) by announcing the discoveryof both new
principlesof human understandingand new methodsforthe study
of mind, behavior,and intelligence-new methodsleading to real
preciseunderstandingin contradistinctionto thesterileapproaches
of yesteryear.1
This book aims to introducethephilosophically inclined to both
the substanceand themetasubstanceof Al. As reviewerI have some

'Some have likened the degreeof understandingto the breakthroughsthat New-

ton and the calculus spurredin physics. Minskypublic lecture,MartinLuther King
day, M.I.T., spring 1975.
0022-362X/81/7807/0408$00.80 ? 1981 The Journal of Philosophy, Inc.

This content downloaded from [Link] on Sat, 11 Jan 2014 [Link] PM

All use subject to JSTOR Terms and Conditions