Two Thought Experiments in the Dissoi Logoi Author(s): Deborah Levine Gera Source: The American Journal of Philology, Vol. 121, No. 1 (Spring, 2000), pp. 21-45 Published by: The Johns Hopkins University Press Stable URL: http://www.jstor.org/stable/1561645 Accessed: 17/01/2009 10:24 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=jhup. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact
[email protected].
The Johns Hopkins University Press is collaborating with JSTOR to digitize, preserve and extend access to The American Journal of Philology.
http://www.jstor.org
TWO THOUGHT EXPERIMENTS IN THE DISSOI LOGOI DEBORAH LEVINE GERA
HAS STRESSEDthat it is not useful to speak of RECENTSCHOLARSHIP Greek scientific experimentation in sweeping fashion. The Greeks did perform scientific experiments, but the quantity, quality, and areas explored varied over different periods. Thus, while at certain times such testing procedures flourished, at other times very few actual experiments were performed. So, too, certain fields were more fruitful or feasible for experimentation than others.' Those who tend to play down the quantity and quality of Greek scientific experimentation usually point to the Greeks' preference for speculation and theory over observation and performance of manual tasks.2 Thought experiments allow the mind to range freely without leaving one's armchair or getting one's hands dirty, and that may be why such mental tests were favored by Greek thinkers, from Xenophanes onward. My purpose here is to study two thought experiments found in the Dissoi Logoi in conjunction with several related trials found in Herodotus, Xenophon, Plato, and Xenophanes. All these thought experiments make use of hypothetical human beings and deal with educational and ethical matters. Generally speaking, these hypothetical tests contrast two types of people, either individuals or larger groups, in an attempt to determine the roles played by heredity and environment in the acquisition of knowledge and moral values. It is notoriously difficult to arrive at a precise definition of thought experiments,3 and the following is meant to be no more than a working definition. Thought experiments are, first of all, experiments.4 In ancient
1See G. E. R. Lloyd 1991;von Staden 1975. 2See, e.g., J. 0. Thomson, quoted in G. E. R. Lloyd 1991, 75. 3Note, e.g., the wide variety of approaches found in the various essays in Horowitz and Massey 1991, and see too Souder 1998, 3-4, and the further references there. 4For thought experiments as a subset of the larger category of ordinary experiments see Sorensen 1992, ch. 9 and passim. American Journal of Philology 121 (2000) 21-45 ? 2000 by The Johns Hopkins University Press
22
DEBORAH LEVINE GERA
Greece scientificexperimentswere not used as a neutral means to decide between two competingtheoriesbut ratherwere intendedto prove or refute a hypothesis,and this is true of Greek thoughtexperimentsas well.5Experimentsmust involve a procedureof some kind, a test consisting of at least one change deliberately introduced into the initial situation.6Normally,then, an experiment is composed of (1) a thesis to be proven (or refuted), (2) a carefullycontrivedstartingpoint, (3) at least one furtheraction to be undertaken,an action affectingthe original situation,and (4) a test of the results of the action(s). In a thought experimentthese steps are carriedout in thought,in the laboratoryof the mind.7The thought experimentervisualizes a situation, mentally carriesout an operation,and then sees, in the mind'seye, the results.A is not actuallyperformed,either because there is Gedankenexperiment no need to do so-common sense, intuition,or experience is sufficient to supply the results of the procedure-or because the test is one that cannot, in fact, be executed in the real world.Such experimentsare not intended to prove fanciful, unlikely theories, nor are they meant to yield results which are intrinsicallyimplausible.Generally,thought experimentsare carefullyimaginedscenarios,which are convincingboth because they have some basis in empirical observation and because they use specificdetails, often colorfulones, to appeal to the audience's intuition.8These well-conceived tests teach us new things,even though their execution does not supply us with new empirical data of any kind. Let us begin by looking at one such hypotheticaltrialin the Dissoi Logoi, an anonymoussophistic composition,written mainly in Doric, which is generally dated to the beginning of the fourth century.9This 5See G. E. R. Lloyd (1991, 71): "The role of the data obtained from many tests is not so much to decide between theories judged antecedently to be of equal standing, as either to corroborate the author's own view or to refute that of an opponent" (italics his). He is describing here experiments actually performed by the Greeks, but his words apply to Greek thought experiments as well. 6Cf. Sorensen 1992, 211: "Experiments are procedures. Procedures must have a number of steps." 7The phrase is that of Brown (1991). This is not to say that thought experiments cannot be executed at other times-see below, discussion connected with notes 29-31. 8See Norton 1991 on such particulars as comprising one of the main characteristics of thought experiments. 9This date is not without its problems: see Conley 1985 and cf. Robinson 1979, 3441, which includes a useful survey of earlier scholarship on the Dissoi Logoi.
TWO THOUGHT EXPERIMENTSIN THE DISSOI LOGOI
23
brief work is comprised of nine chapters, often presented in the form of double (or opposing) arguments, on subjects such as good and evil, justice and injustice, truth and falsehood, etc. Of these, chapter 6 discusses whether wisdom and virtue can be taught, and here the anonymous author touches upon the acquisition of language, while contrasting systematic teaching with informal instruction. He argues that we learn language and are not born knowing it, and offers an imaginary skeptical opponent the following proof of his claim (ai 6E TwOL [T] tL lTOV EoTl... yvcxTco ex TCOv6e,6.12): if one were to send a newborn Greek infant off to Persia and raise him there, without his hearing Greek, he would speak Persian; and if one were to bring a Persian baby to Greece, he would speak Greek (c4YTLgei9Og yevo6[tvov jrati6ov eg IIeQa(C aJroxco)bov 'EEXkbog ()cvdg, JTEtiPal Xal TTqVElTQ(c4qOL,
TEQcoloL XWc'A'LTLg
TlVOv6evTl6tb xo[ticaL, eXXUavitoxc, 6.12).
Brief as it is, this test points to the various stages of a thought experiment. After stating his initial thesis-that language is acquired, not innate-the author imagines a hypothetical situation in which a Greek child is transported immediately (erOvg) after birth to Persia. The next stage involves raising this infant. The hypothetical child is to be brought up in Persian-speaking surroundings, with the added proviso-a control for the experiment-that he hear no Greek speech (xw(6v 'EXXa6og (wovdg). The original thesis can now be put to the test, by determining which language the child uses. The result, states the writer, is clear: the Greek-born child will speak Persian. Interestingly, our author includes an additional control: a second experiment precisely parallel and complementary to the first, in which a Persian baby is to be sent to Greece and will then end up speaking Greek. Each of these two experiments acts as a check for the other, strengthening and confirming the results of the parallel trial. It is also worth noting that the experiment can be performed by anyone (TLg) and, we might add, at any time on any random, hypothetical child, male or female. Such randomness and repeatability is a sine qua non of a thought experiment and of actual, scientific experiments as well. The experiment is phrased as a less vivid future condition, with ei and the optative in the protasis, and the optative and xa (the Doric dv) in the apodosis (a'l TLg... &jroxtp4uctt xal ... TQdc0oL... jEQQi~LOLxa' ca TS ... xodi[aM, eXUcaviLo xa), and this is the syntax regularly used in
thought experiments. While the test is clearly meant to be hypothetical and our author is not suggesting that babies actually be sent off to foreign environments, the experiment could, in fact, be performed. This is
24
DEBORAH LEVINE GERA
an instanceof a test whichneed not be carriedout in actuality.Thinking about it is enough. This experimentyields even richerresultsthan those noted by the authorhimself.His immediatepurposeis destructive,for he uses the experimentto refute the contentionthat wordsare innate (cf. eJoTca[tevog a[ta yLveo0aL, 6.12), but the actual test proves a great deal more and leads to constructiveresultsas well.10Thusthe experimentdemonstrates that the languageone speaks has nothing to do with race or blood, for speech is acquiredfrom one's surroundings:changinga child'senvironment will lead to a changein his nativespeech.11Languagedoes not arise from within, nor is it taught by a specificteacher;it is impartedby the surroundingcommunityas a whole. Indeed, as the author notes, it is oux impossibleto singleout specificteachersof speech (To'g6bLacoxdtXW iCeCteg).
The Dissoi Logoi betraysthe influenceof variousfifth-centuryauthors and sophists of the firstgeneration.Scholarsare generallyagreed that Protagorasinfluencedthis sophisticwork both in structure-that is, in the use of opposing,antitheticalarguments-and in content.12There are also several specificresemblances,includingverbalechoes, between this section (chapter6) of the Dissoi Logoi, which deals with the teachability of virtue, and Protagoras'views as presentedin Plato's Protagoras. In fact, the argumentthat we learn languagewithoutknowingwho our teachersare is used by the sophistin that dialogue.Protagoraspoints to the analogy between learning political virtue and learning Greek: both are acquiredfromthe communityat large(Prot.327e).13The echoes of Plato's Protagorasfound in the (presumably)earlier Dissoi Logoi would seem to indicatethat Plato'ssophistis expressingthe views of the actual Protagoras,who influencedour anonymousauthor.14Could our l?In the terminology used by Brown (1991, 33-34, 43-45) it is a "platonic" thought experiment, simultaneously destructive and constructive. "See Desbordes 1987, 35-36. 12See Gomperz 1912, ch. 8, for a detailed discussion, and compare Guthrie 1969, 316-19; Robinson 1979, 54-59. Kerferd (1981, 54, 84-85) argues that the technique of opposed arguments was part of the general sophistic movement and not confined to Protagoras, so that there need not be a direct link between Protagoras and the Dissoi Logoi. '3See Robinson 1979, 57, 216 (ad Diss. Log. 6.11), on the verbal echoes between PI. Prot. 327b-328d and Diss. Log. 6.11-12: both use the words evFvIJNg, Xkcavi,eiv,ixavog. There are parallels between Diss. Log. ch. 6 and Plato's Meno as well: see Taylor 1911, 114-19. 14See Guthrie 1969, 64 with n. 1.
TWO THOUGHT EXPERIMENTS IN THE DISSOI LOGOI
25
thought experiment derive from Protagorasas well? It is tempting to think so, but there is no hard evidence to support this supposition.In any event, the Platonic Alcibiades draws a similar parallel between learninglanguageand learningjustice.He states that fromearliestchildhood onwardhe learnedto speak Greekjust as he learnedof justice and injustice,from the many.He, too, cannotpoint to any one person as being his teacher of language(Alc. 1.llOd-llla). A particularlyinterestinginfluenceon our thought experimentis an experimentreported by Herodotus (2.2): the notorious attempt by the Egyptian king Psammetichusto determine humankind'sfirst language. Psammetichushanded over to a shepherdtwo newly born children, chosen at random(Jna8ia 60ovcoyv&aa&v0(d)tov TrOvETLTVxovTOW,2.2.2).15The herdsmanwas orderedto raise them by themselves,in an isolated hut (ev oteyn etr' eorUTbv,2.2.2), bringingin goats k[nFbe to supply them with milk and otherwisetending to their needs. No one was to utter a sound in their presence.Psammetichuswished to know which languagethe childrenwould firstspeak, once they had done with incoherentbabblings(XTcv&ofl[tWvxvvrq[tdTwv, 2.2.3).Whentwo years had gone by, the shepherdopened the door of the hut one day,and both the children greeted him with outstretched arms, uttering the word bekos ({exog). Psammetichusdiscovered that bekos was the Phrygian word for "bread,"and consequentlyhe and his fellow Egyptiansconcluded on the basis of this trial (cf. TOLOUTooTa0ftrtod[tevo0LJrTyXcLaTL, 2.2.5) that the Phrygianswere the oldest people. The suppositionunderlyingthe Egyptianking'sexperimentis that language is innate.16More specifically,Psammetichusseems to have thoughtthat the firstor earliestlanguageis innatein childrenand would be spoken by them when they were placed in the same conditions as those of the earliestpeople. This firstlanguagethen points to the oldest people:the most ancientlanguageis spoken by the most ancientpeople. The anonymousauthorof the Dissoi Logoi clearlydisagreeswith Psammetichus' assumptionthat language is innate and argues that children 15Salmon (1956, 321-22 n. 3) rightly remarks on the consistent use of the neuter Ta jzta6La to describe the children-their gender is random as well. This is also true in the Dissoi Logoi experiment. 16Robinson 1979, 217 (ad Diss. Log. 6.12), calls it "the clearest extant example of an 'innateness' theory of this sort." Compare Borst 1957, 39-40, and see Benardete 1969, 32-35, for an interesting analysis of the various assumptions behind Psammetichus' experiment.
26
DEBORAH LEVINE GERA
learn their speech from their surroundings.His thought experiment seems to be a reaction,in part,to the Egyptian'stest,17which-whatever its historicityor credibility-is presentedby Herodotusas an actual,executed experiment. In almost every respect, our author's thought experiment is the more satisfyingof the two trials.For one thing,the resultsof his experiment are far more convincingthan Psammetichus'problematicfindings. It is highlyunlikelythat two childrenplacedin suchconditionswould actually have learned to speak.18Ancient authorshad alreadynoted the possibilitythat the sound bek was the resultof the children'sattemptto imitate the bleating of the goats, who supplied them with milk.19Perhaps that is why other versions of the experimentdo not include goats but state instead that the childrenwere fed by women who attendedto them in silence.20Modernscholarspoint to a furtherdifficultyin Psammetichus'results.The word bekos can be taken for an Egyptianword,21 and the king would presumablyhave been more than happy to fasten upon such a resemblancein orderto prove the primacyof the Egyptians. Our author'sthought experimentnot only achieves more plausible results than Psammetichus'actual test, it is also more elegant and more humane,preciselybecause it does not have to be executed in re17The author of the Dissoi Logoi does not refute Psammetichus' experiment directly but rather argues against the idea of language being innate. If the author were to perform the king's trial in thought, he would, I think, contend that if the children heard no language they would not learn to speak, whereas being exposed to goats would lead them to bleat. '8Crystal (1997, 291) lists nearly fifty recorded cases of children of the wild, who were isolated from social contact, and notes that normally such children did not learn to speak. Several more recent inquisitive monarchs are said to have conducted experiments similar to that of Psammetichus. In the trial conducted by Frederick II of Hohenstaufen (13th cent.) the children died; the infants put to the test by Akbar the Great of India (16th cent.) remained mute; in the case of James IV of Scotland (15th-16th cent.) the children were reported by some to have spoken good Hebrew! See, e.g., Sulek 1989, 647-48; Crystal 1997, 230, 290. '9Suda 3 229 s.v. P3exeoeXive;schol. Tzetz. Ar. Nub. 398a; schol. ad Apoll. Rhod. 4.257-62c. Salmon (1956, 326) notes that even in Herodotus' version the goats are very much present. The historian tells us that they were brought into the children's hut (2.2), which makes us wonder why they were not milked outside, with the youngsters then drinking their milk from a cup. 20See below, discussion leading to note 25. 21A. B. Lloyd 1976, 10 (ad Hdt. 2.2), refers to one of the Egyptian names for Egypt and the Egyptian word for bread. See also Salmon 1956, 323, and compare Hipponax fr. 125 West.
TWO THOUGHT EXPERIMENTS IN THE DISSOI LOGOI
27
ality.22A few minutes of thought,based perhapson an actual acquaintance with children of foreigners who had learned to speak Greek as theirnativetongue,23sufficeto underminethe purportedresultsof a trial whichis said to have lasted two years or more.24No real-life infantsare torn away from their mothers to be raised in silence and solitude by a shepherd;the young children are simply dispatchedto farawayplaces in our thoughts.All the variousversions of Psammetichus'experiment involve crueltyof one form or another.Herodotushimself mentions an alternate account accordingto which the children were cared for by women whose tongues were cut out: both the caretakersand the young childrenwere damagedby Psammetichus'curiosity.Elsewhere,in other variants,goats are said to have suckled and raised the children,which means that the childrenhad no human contact whatsoever.In another variation,just a single child is broughtup in speechless solitude, without the solace of the companyof anotherhumanbeing. Even the most humaneaccount,accordingto whichthe infants'motherswere the ones to raise them, but were ordered to do so in silence, entails mental anguish for the mothers and deprivationfor the children.25A thoughtex22It is intriguing to note how closely the Dissoi Logoi trial is echoed in a recent study of modern linguistic thought experiments, in a very similar context. Thomason (1991, 248-49) first mentions Psammetichus' experiment and the ethical problems raised by the king's test and then goes on to suggest a thought experiment which could be used by a linguist to test the theory that children are genetically predisposed to learn a specific language, not just language in general: "Take an infant born to parents of completely homogeneous monolingual linguistic background going back, say, ten generations; remove the child at birth and place him/her with adoptive parents whose own language, and that of the entire surrounding community, is (as far as linguists can tell) completely unrelated to the language of the child's biological parents and ancestors. Linguists will predict that the child will learn the language of his/her adoptive community as fast and as easily as s/he would have learned the language of the biological parents." 231 owe this suggestion to the editor of this Journal, who also notes Herodotus' observation at 2.104.4 in this context. 24In some accounts the experiment is said to have lasted even longer, with the children first pronouncing the word bekos at the age of three or even four: schol. Tzetz. Ar. Nub. 398a; Suda P 229 s.v. P13exc^XvE (and schol. vet. Ar. Nub. 398b); schol. rec. Ar. Nub. 398e. 25Tongueless wet nurses: Hdt. 2.2.5; Suda ,3 229 s.v. P3xeoeXrivC; schol. vet. Ar. Nub. 398b, 398c; schol. rec. Ar. Nub. 398f. Tongueless mothers: schol. vet. Ar. Nub. 398d. Goats: schol. Apoll. Rhod. 4.257-62c (p. 274 Wendel); Suda (3229 s.v. PexeoY.Xve (and schol. vet. Ar. Nub. 398b). Single child: Claudian In Eutropium 2.251-54; schol. Thomas/Triclinius Ar. Nub. 398b; schol. rec. Ar. Nub. 398f; schol. vet. Ar. Nub. 398c. Silent mother(s): schol. rec. Ar. Nub. 398e; schol. Tzetz. Ar. Nub. 398a; schol. Thomas/Triclinius Ar. Nub. 398b.
28
DEBORAH LEVINE GERA
periment is infinitelykinder.26Herodotus' Psammetichus,incidentally, again dispatchedchildrento farawayplaces, for linguisticpurposes,in an entirely differentcontext. The king, we are told, sent Egyptianchildren to his Ionian and Carianmercenariesin order to have them learn Greek. This was the origin of the Egyptianclass of translators(2.154.2). Psammetichusis of a scientific, inquiringbent, and we hear in Herodotus' History of another experiment conducted by him, an attempt to sound the depthsof the springsof the Nile.27He is said to have fashioned a rope thousandsof fathoms long, but his soundingline did not reach bottom, leading him to conclude that the springswere bottomless (2.28). Herodotus doubts that this experiment actually took place, but he points out that powerfulwhirlpoolsand eddies could have prevented the line from reachingbottom. Here, too, the Egyptianking performs an open-ended experiment in order to satisfy his curiosity, but the inference he draws from the experiment'sresults is again unconvincing.
Interestingly,this experimentby Psammetichuscan be contrasted with a similarbut more successfulexperimentsuggestedby Herodotus himself.28When discussingsilt depositedby the Nile, the historiannotes that a soundingline let down one day's sail from land will drawup mud at a depth of eleven fathoms.Herodotus' wordinghere, his use of the generalizingsecond-person singularand the future indicative,"if you sail... and let down a line, you will draw up mud" (tQOoJrXVc0v... ajrecov ... xaTEigXCaTaXtETLQTq@iTV jrqX6vTE &voioetg, 2.5.2), does not allow us to determinewhether such a trial was in fact undertaken.He may simplybe reportinga hypotheticaltest that might be carriedout.29
26Cf.Tzetzes (schol. Tzetz. Ar. Nub. 398a), who calls the tongue-cutting version in& human (or lav yXc0ooonr6tToexcs0' 'HQ66o0TovaTdvOQtov yQ TOlTO),and compare schol. rec. Ar. Nub. 398e TOOTOcai 0et!QL(o6cg. 27Herodotus uses what seems to be a technical word for experiment, dtrteiect, when describing Psammetichus' two trials (2.28.4, 2.15.2); see Christ 1994, 182 with n. 40. Benardete (1969, 41) sees both experiments as an attempt to go back to beginnings. 28Christ(1994, esp. 172, 183) compares the inquiring historian and the experimenting king. He discusses at length the love of experiment demonstrated by Herodotean kings; see also Flory 1987, 78, 174 n. 34. 29See Christ (1994, 183 with n. 42), who terms the experiment "hypothetical." He compares two further experiments conceived by Herodotus, thought experiments which could never actually be executed: the diversion of the Nile into the Red Sea (2.11.4) and the reversal of the position of the north and south winds (2.26). The first thought experiment has earned Herodotus high marks from modern commentators: see Gould 1989,
TWO THOUGHT EXPERIMENTS IN THE DISSOI LOGOI
29
Actual Greek scientificexperimentswere often phrasedas future conditions, taking the form "If conditions of type A are realized, then an event of sort B will occur."30While such future conditionalswere perhaps the normal means used by Greek scientists to design or record their experiments,the conditionalsalso leave open the possibilitythat some trials were hypotheticaland the experimentsrecorded were not actuallyexecuted. In other words, the same languageand syntax could be used by Greek writersfor three different kinds of tests: (1) actual, executed experiments,(2) thought experiments,and (3) proposals involving a concrete test which has not yet been performed.Such a procedure, if performed,would (unlike a thought experiment)yield new, empiricalresults.The identicalsyntaxused for all three kinds of experiments means that seemingly scientificexperimentsrecordedin Greek writingsmay simply be hypotheticalsuggestions.31At the same time, it seems clear that at least some scientificexperimentswere performedin the late fifth centuryand early fourth,the traditionaldate of the Dissoi Logoi.32
Returningto Psammetichus'linguisticexperiment,it is worth noting that Herodotus' tale is a study in contrasts:the king's trial and results are incredible, but his methodology is scientific and serious.33 He
86-87, and Heidel 1933, 161 (attributing the passage to Hecataeus of Miletus). The second thought experiment, while a vivid analogy, has found much less favor: see Heidel, 160, and cf. Gould, 90. 30Thusvon Staden (1975, 186), who notes the use of three different kinds of future conditions in the experiments he discusses: (1) the protasis has Ei with the optative, and the apodosis the future indicative; (2) the protasis has ec with the optative, and the apodosis the optative with aiv;(3) the protasis has Edv with the subjunctive, and the apodosis the optative with aiv. 31See, e.g., the experiment described by the author of Sacred Disease (11.4 Grensemann; ch. 14 Loeb edition), on performing a postmortem on a goat: "If you cut open the head, you will find that the brain is wet" (i]v 6LaxouVBTflv xeqctXakv,eQfioetLgToyvEYEG. E. R. Lloyd (1979, 23-24) points out that it is not certain that the qc)aXovVyQ6OV ova). writer actually carried out this procedure. Even in more recent times it is not always easy to distinguish proposed experiments from executed ones. Some of Galileo's experiments are interesting cases in point; see Naylor 1989. 32See Ar. Clouds 148ff. and Dover 1968, xl-xli. Compare too the Hippocratic Airs Waters Places (ch. 8). G. E. R. Lloyd (1979, 27 n. 91) dates both AWP and Sacred Disease (above, note 31) to "within about 20 years of the turn of the fifth and fourth centuries." 33This contrast has led (e.g.) Salmon (1956) to argue that the scientific aura surrounding Psammetichus is intended to make the king look ridiculous.
30
DEBORAH LEVINE GERA
attempts to answer a complex question scientifically,by conductinga controlled,open-ended experiment.He does not know which language the childrenwill speak, and when the youngstersunexpectedlyend up talkingPhrygian,the king immediatelyacceptsthe resultsof the experiment and relinquishesall claims to Egyptianprimacy"underthe influence of a single empirical counter argument."34 The king's linguistic trial-however unfoundedthe assumptionsbehind it, and however unlikely the reported results-is arrangedwith the utmost care. When performinghis test, Psammetichususes a series of controls, choosing the two childrenat random,takingthem away at a very young age, and carefullyisolating them from the sound of speech.35The author of the Dissoi Logoi makes use of these very same factors in his thought experiment,for he too (mentally)removes two infantsfrom their natural habitat,places them in an artificialenvironment,and isolates them from the sound of certaintypes of speech, that is, Greek and Persianrespectively.At the same time, the two experimentsmake use of the two children in quite differentways. In Psammetichus'trial, the youngstersare raisedtogetherand are presumablymeant to stir one anotherto speech, whereas in the Dissoi Logoi they are sent to separatedestinationsand each serves to prove one-half of the complementarytwo-part thesis. Even in Psammetichus'experimentthe two childrendo not, in fact, begin to speak by communicatingwith one anotherbut turn to the shepherd instead. In other, non-Herodotean versions of the trial, the children first speak when they addressa special emissarysent by the king, after havingbeen kept in isolationfor years.The emissaryis orderedto enter their hut and observe them silently-yet anothercareful control. Indeed, in some accountsof the experimentthe king isolates only one child, and this does not affect the end result.(Nor should it, if language is thoughtto be innate.)36
34Sulek (1989, 650), who admires the Egyptian king's astounding intellectual honesty. 35See Salmon 1956, 326; A. B. Lloyd 1976, 5-6 (ad Hdt. 2.2). 36Special silent emissary: Suda ,3 229 s.v. [FexeoXiTvve (and schol. vet. Ar. Nub. 398b) JT nmaQOei0v); see also schol. vet. Ar. Nub. 398d. One child: see above, note 25. (note OLWor Contrast the use the eighteenth-century linguistic thinker Condillac makes of two such hypothetical isolated children (in part 2 of his Essai sur l'origine des connoissances humaines), to demonstrate how the two children, left to their own devices, might gradually have developed a language; see Stam 1976, 45-50; Harris and Taylor 1989, ch. 10, esp. 125-28.
TWO THOUGHT EXPERIMENTS IN THE DISSOI LOGOI
31
Whereas the two children in Psammetichus'experiment were raisedtogether,and the two childrenin the Dissoi Logoi are to be raised in distinctlyopposite but complementaryways, Xenophon, in another thoughtexperiment,suggeststhat two youngstersbe raisedtogether,but accordingto very differentprinciples,in orderto trainthem for two contrastinglifestyles(Mem.2.1.1-7).Xenophon'sSocrateswishesto demonstrate that Aristippus'self-indulgent lifestyle categorizeshim as someone who is fit to be ruledsubmissivelyby others,ratherthan be a leader. The philosopherdevises an armchairexperimentsuch that he and Aristippus are to train two hypotheticalyoung men, one of whom must be capableof ruling,while the other is to be a willingsubject.Socratesruns through a series of qualities involvingenduranceand self-control and in each case elicits Aristippus'agreementthat of the two youngstersit is the prospectiverulerwho should be trainedin these traits. Here, too, an alleged actual experiment may be lurking in the background.The Spartan leader Lycurgusis said to have raised two pups from the same litter to lead two very differentkinds of lives, spoiling the one and trainingthe other to hunt. He then called together the Spartansand used the puppies in a live demonstrationto show how trainingand disciplineform character.When he put a dish of food and a hare in front of the two dogs, the spoiled one ate from the dish, while the traineddog chased after the hare.37Just as Xenophon uses the two hypotheticalyoung men to bringAristippusto think about his own way of life, Lycurgusmade use of a vivid physicaltrial to stimulatethe Spartans to deliberate about their own practices.38Such use of visual presentation is thought to be a particularlySpartancharacteristic.39 Herodotus' tale (9.82) of Pausanias'ordering that a Persian meal and a Spartanone be placed side by side, after the Greek victory at Plataea, is another such graphicdemonstrationof two contrastingways of life: the two meals do not actuallyconstitute an experiment,but Cyrusthe Great, who has had the Persianswork all of one day and then feast the
37See Nicolaus of Damascus FGrH 90 F56; Plut. Mor. 3a-b, 225e-226b; cf. Gigon 1956, 14-15. Nicolaus, our earliest source on Lycurgus' trial, is of course much later than Xenophon. 38Interestingly,Lycurgus used two dogs from the same litter in his trial (Nicolaus of Damascus FGrH 90 F56.3; Plut. Mor. 3a-b; compare Mor. 226a) to stress the effect of training versus nature, whereas Xenophon does not stipulate that the imaginary two young men be brothers. 39See Powell 1988; Flory 1987, 105-6.
32
DEBORAH LEVINE GERA
next, in order to see if they are willing to revolt against the Medes (1.125-26), is conductinga trial of sorts.40Such colorful,concrete demonstrationsof two ways of life are perhapsmore compellingthan Xenophon's theoretical trial, but another thought experiment on the same theme, that of Gyges' ring, indicates that hypothetical trials can be quite memorableas well. The experimentinvolvingGyges' ring, formulatedby Glaucon in Plato's Republic(359b6-362c8;cf. 612b),is one of the most famous ethical thought experimentsof antiquityand presents another interesting instanceof the relationbetween an alleged actualexperimentand a hypothetical trial. Glaucon contends that men are just only unwillingly, simply because they are unable to get away with being unjust.He tells Socratesthat the truth of his claim will be more easily perceivedif they performthe following actions in their imagination(tktXLorcaiv cdtoOoia 359b6-cl). He suggestsgrant[iCea, ei Tol6ve& AotLo[tqev Tf 6tavoig, freedom to do as they like and an man a and complete ing just unjust where desire leads them. Each of the two them to see then following will be red-handed men, states Glaucon, caught proceedingin the same of way, in his own self-interest.41At this stage the experiment,the observers, Socrates and Glaucon, are unseen, so to speak, while the just and unjustman performtheir actionsopenly. Glauconthen proposesrefiningthe trial, allowingthe just and unjust men complete license by grantingeach one a magic ring which will make him invisible,a ring akin to Gyges'.Glaucontells the story of Gyges' discoveryand use of his magic ring at considerablelength,presenting the shepherdas an investigatorand experimenterin his own right.42 After Gyges finds the ring on a larger-than-life corpse, he discoversits magic powers by accident,noting that he disappearswhen he turnsthe ring inward and reappearswhen he turns the ring outward.He then carefullytests the power of the ring (xai TOTOevvovoCavTlCa a&totrteQoOatTO 6&axTv)Love T)xTlqv eXolt Tiv bvva(ktv, 360a4-5), experimenting by trial and error as he repeatedlyturns the ring in and out to become visible and invisibleby turns.Once he has establishedto his satis40Flory (1987, 174 n. 34) sees this as an instance of the Herodotean motif of the "king's research": see above, note 28. 41The experiment is set up as a less vivid future condition with el and the optative in the protasis, and the optative and dv in the apodosis, and conditional sentences of this form recur throughout the passage. 42Cf.Heidel 1933, 162-63.
TWO THOUGHT EXPERIMENTS IN THE DISSOI LOGOI
33
faction the ring'smagicaleffect, he uses it to seduce the queen, murder the king, and take over the Lydianthrone.GlauconstressesGyges' role as an inquisitiveexperimenter,perhapsbecause he is about to suggest to Socratesthat they continue their trial and turn the ring round,so to speak. When Glaucon now grantshis hypotheticaljust man and unjust man their magic rings, they are permittedto wander around invisible, while the observers,Glaucon and Socrates,turn visible, as it were, and watch the two figures.What will be the result of this armchairtrial? Both the just and the unjustman, Glaucon argues,will behave equally badly,stealing,murdering,and so on. Glaucon'suse of magic ringshere adds color and life to his hypothetical trial and makes it all the more memorable.43We are strucknot only by the ruthlessand immoralbehaviorof the two men, but by their magic invisibility,which allows them to behave in this fashion.Suchpicturesque,telling details which linger in the mind are part and parcel of the more stimulatingthought experiments,even if such colorful bits are not necessarilyan essential part of the actual trial. So, in modern thought experiments, for example, we remember Maxwell's demon, Schrodinger'scat, JudithThomson'sunconsciousviolinist,John Searle's Chineseroom, etc.44Glauconwill continueto contrastand comparethe just and unjustman, investigatingthe two opposingways of life in their purestform, but his furtherremarkslack the vivacityand immediacyof Gyges' experiment.Glaucon next asks Socrates to imagine two men: one completely just but with a bad name, and another unjust with a good reputation.The just man, he says, will end up torturedand reviled, while the unjust one will have wealth, influence, and power (360e362c). Socrateslaughs at Glaucon for devisingthe hypotheticalcircumstances so deliberately,polishing and purifyingeach of the imaginary men as carefully as if they were statues (361d4-6). Although the creation of such a contrivedstartingpoint is an importantelement in setting up a thought experiment,Glaucon does not go beyond this initial stage here. Each of his two men is simplyassigneda certainway of life with its concomitantqualities and then allotted the fate which logically results from that lifestyle. The two men do not of themselves do any43Wilkes (1988, 11) argues against the validity of Glaucon's thought experiment and claims that not enough information is supplied. She asks, for instance, whether the ring makes the men intangible as well as invisible. 44See Sorensen 1992 passim for details of these experiments. Norton (1991) points out that these particulars are not really essential.
34
DEBORAH LEVINE GERA
thing, and it is up to the reader to project the intermediate stages of this thought experiment.45 Both the Gyges thought experiment and Xenophon's hypothetical education of two young men are presented in dialogues, so that in each instance the experimenter is outlining his trial to an actual audience. In the Republic Socrates simply listens to Glaucon describing Gyges' test, but in the Memorabilia Aristippus joins Socrates in executing the thought experiment firsthand, agreeing as to the manner in which a future ruler should be trained. Thought experiments can be performed at any time, by anyone who is willing to use his powers of reflection and imagination, and often an audience is addressed-either implicitly or explicitly-by the person who formulates such a test. The linguistic experiment in the Dissoi Logoi, for instance, is phrased as a response to a hypothetical doubter who is virtually invited to think for himself what the result would be if someone were to send a young child abroad (cf. cai 6? To()
JTL t(TOVeoTLT& ov
t... atcTcTa ![tavOdvcLv&[e
yvcWToex TrOVb6,
6.12). We often find an appeal to the reader or listener-who is addressed, at times, in the second person-to play a part and collaborate in the experiment.46 "Thought experiments are intended to persuade by reflection on the experimental plan,"47and contrivers of such trials frequently use rhetorical means to convince their audience. We have already noted the similarity, in presentation and syntax, between thought experiments and the scientific experiments recorded in early Hippocratic works, and the latter include persuasive addresses to an audience as well. An early experiment on freezing water in Airs Waters Places (ch. 8) is introduced in a fashion similar to our experiment in the Dissoi Logoi, with the reader gently bidden to try things for himself (yvoiiq 6' &v c)6e' ei ya 3oVXe.... eyXtUg i6UcoQ0eval... a eTET... eoEVEyxcv
.. .(aVa[ExTQeLV
... EVQqCEl g...
TOVTOTEXRLOV OTL.. .).48
4iFor this section of Glaucon's exposition (PI. Rep. 360e-362c) as a thought experiment see, e.g., Williams 1993, 98-101; Sansone 1996, 53-55. The latter argues that Glaucon's picture here is inspired, in part, by Euripides' Helen, in which the real Helen lives virtuously in Egypt, maintaining her chastity and uprightness but nonetheless earning a disgraceful reputation. Euripides' play, Sansone claims, serves as a model, a "real-life" experiment, as it were, underlying Glaucon's hypothetical scenario. 46See Souder (1998) on the use of the second person in philosophical thought experiments as an invitation to the reader to step into someone else's shoes. He contends that examples framed in the third person also implicitly ask the audience to assume a role. 47Sorensen 1992, 225. 48Cf.Sacred Disease 11.4 Grensemann; ch. 14 Loeb edition; and above, notes 29 and 31. Compare the attempt by Robert Boyle to turn his readers into virtual witnesses of his
TWO THOUGHT EXPERIMENTSIN THE DISSOI LOGOI
35
Let us turn now to a second thought experiment in the Dissoi Logoi, a hypothetical reallocation of moral and ethical values. This experiment features in the second chapter, which presents the contrasting views held on the seemly (T6 xackov) and the unseemly (T6 aoXQO6v). Our author outlines the experiment twice: once when arguing that a[oXQov and xcakov are relative terms (2.18), and a second time when refuting the contention that the two qualities are identical (2.26-28). The thought experiment is first mentioned in an ethnographical context: when attempting to demonstrate that seemliness and unseemliness are relative concepts, the writer describes a series of practices which are thought honorable by one ethnic group and disgraceful by another. Thus, for example, Persians are said to consider incest seemly, while Greeks find such behavior shameful and unlawful (2.15). The author then suggests a hypothetical trial to demonstrate the variety and relativity of values: suppose someone were to order all of mankind to bring tot'L TCg ioxQt gether into a single heap what each thinks shameful (cal Ev xe.kUOl OL vveVxalc a EXaoTOL 5g jtdvag tcvOQcbOJtcOg voi[dovTL) and then to have them take away from this same collection what each considers seemly (xat JcdckLv &096cowvTOUTcoVTx&xakca Xaev, & EXcaoTOL aylvTxal); not even one thing, he says, would be left (oiV6f ev xa xakXetL)0f[tev). All would divide up everything, he concludes, for not everyone holds the same views (&kka&jdtVTg jrTdVT 6takCaC3v. ov yade JTdvTFegTavTa VO[ii,OVTL, 2.18). As with the previous thought experiment in the Dissoi Logoi, this trial is phrased as a less vivid future condition, with eLand the optative in the protasis, and the optative and xa in the apodosis: a'LTLg... xeXeOl o vvELVxat... xaci Mktv ...
a[3ev ... o?6e ev xaXaXC.Xetq0ftev
(2.18). Here too we find that the author first contrives an initial hypothetical situation, next suggests a further action to be undertaken which affects the original circumstances, and then turns to examine whether the result of this action, that is, the result of the experiment, confirms his original thesis. Thus we are first asked to imagine that each and every individual is commanded to bring to a common pile those qualities personally considered to be most shameful. Next, in the second stage of the experiment, each is asked to take away from the newly formed pile qualities which are personally regarded as seemly. We can now test the hypothesis that moral judgments are not absolute by ex-
experiments in the laboratory, by providing detailed written descriptions of his equipment and procedures; see Cantor 1989, 163.
36
DEBORAH LEVINE GERA
aminingour hypotheticalheap to see if any qualitieshave been left behind.Will there be any unsavorypracticewhichno one on earthconsiders positive and wishes to take away?The answer,states our author,is no, for every conceivableform of behaviorhas its supporters. This thoughtexperiment,unlike the earlierlinguisticone, must remain a hypotheticaltrial conductedin the laboratoryof the mind and cannot be executed in reality.Even if everyone in the world could be gathered together in one spot,49those gathered would not be able to place intangible qualities-their ideas of what constitutes a shameful act-into a heap.Yet there is somethingcolorfuland vivid in this image of people firstpiling up their incorporeal"wares"at a common market or barteringplace, with each person then carefullyexaminingthe merchandise before choosing the new commodity to take away from the stockpile.We can readilyvisualize,for instance,a Greek firstdepositing incest in the pile, only to be followed by a Persianwho finds the practice seemly and bears it away. The notion of displayingqualities,customs, or lifestyles as goods to be chosen in a marketplaceappearsin other, nonexperimentalcontexts as well. One instance is Plato's myth of Er,50in which souls of the dead are given lots in the otherworldand are then, each in turn as the lots are drawn,allowed to choose for themselvesnew lives from the "marketplace"of various types of lifestyles (Rep. 617e). The origins of such heaps of intangiblequalitiesmay well be the famous image of the two urns of Zeus described in the Iliad (24.527-33). There it is Zeus, ratherthan humanbeings,who drawsgood and evil fates, and he draws from two jars.51Here, rather than jars, we have one heap which contains at one and the same time both good and bad qualities-which is, in fact, the point of the experiment:each qualityconsideredbad by one person is consideredgood by someone else. The thesis behind this trial, that v6O[to are relative,can be traced back to Pindar(daca 6' aXkoLoLv 6' aivl 6tixav &v66b)vexaoTog, fr. 215 = P. Oxy. vo6[MLa,oMETE@av 49Inthe second versionof the experiment,only a limitednumberof people collect attitudesfromthe variousnationsall over the world(ex TcOvE0OvcWv Diss. Log. JdtvxoOev, 2.26). 50See Immerwahr 1966, 320-21 with n. 37. 51Forthe ancient debate as to whether Zeus has two jars (one evil, one good) or three (two evil, one good) see Richardson's commentary ad loc. (1993, 330-31). Pandora's notorious jar (Hes. W&D 90-104) should also be noted in this context.
TWO THOUGHT EXPERIMENTS IN THE DISSOI LOGOI
37
2448) and was perhapsa commonplaceby the time that the Dissoi Logoi was written.52 Yet at the same time, here too Herodotus casts his shadow.The context in which this thought experiment is introducedin the Dissoi Logoi certainlyseems Herodotean.We have seen that the experiment follows upon descriptionsof a series of unusual non-Greek customs (2.9-17). Many of these practices are mentioned in Herodotus' History,53 and it is possible that the historianwas also the inspirationfor the theoreticaltrialfound in our sophistictext.54Indeed, Herodotusdescribes a similarhypotheticalsituationof men who are commandedto choose for themselves the finest customs, from a collection of all the world'spractices(vo6ovg To; xcakkioTovg;?x TOrvncVTov v6oi[tv,Hdt. 3.38.1).The historianconcludesthat all men, aftercarefulreflection,will choose their own (6LaSoxeVpdtevoL&v FoiaxTo CxaoToL
ToUg; EoVUTCV),for
everyone thinks his own nomoi superior by far. Herodotus' test, presented in the form of a future-less-vivid condition (se yd6Q tLgJTQo0eil ...
ExXFtaoGat xeXseO?v...
&v Ekotaro), is a compressed thought ex-
periment,with the different stages-collecting the customs, examining them, then choosing one-sketched in outline. His thesis is also somewhatsimplerthan that found in the Dissoi Logoi, for he simplycontends that each man will thinkhis own nomoi best. He does not stressthe relativity of values, where one man's most distastefulnomos might be another man'sfavorite.(And note that all of the world'scustoms are collected, not just the unseemlyones.) Immediatelyafterward,however,he tells of Darius' famous comparisonof the different burial practicesof Indiansand Greeks, which is an exceptionallyvivid demonstrationthat very different kinds of nomoi are practicedand respected by different peoples (3.38).The context of the hypotheticaltrialsin the Dissoi Logoi and in Herodotus is, then, virtuallyidentical.Darius' flamboyantpresentationof the differentnomoi practicedby Indiansand Greeks fright-
52See Guthrie 1969, 131-34; cf. Heinimann 1945, 71-72, 78-89. 53Compare Diss. Log. 2.13 and Hdt. 5.6, 4.64-65; Diss. Log. 2.14 and Hdt. 1.216, 3.38; Diss. Log. 2.15 and Hdt. 3.31; Diss. Log. 2.16 and Hdt. 1.93; Diss. Log. 2.17 and Hdt. 2.35. Cf. Conley (1985, 60, 64 n. 10), who points out that the parallels between the two texts are not always exact; Robinson (1979, 165-66 ad 2.9), who is skeptical about Herodotus' being a direct source for the ethnological lore of Diss. Log. 2.9-17. S4Gomperz (1912, 163-64) suggests that both the Dissoi Logoi passage and the Herodotus parallels stem from Protagoras; see also Heinimann 1945, 80-81.
38
DEBORAH LEVINE GERA
ens his Indiansubjects,and perhapsboth our anonymousauthorand the historianwere influencedby the Persianking'sautocraticbehaviorwhen formulatingtheir hypotheticaltrials.Both imaginemankindbeing commandedor biddento choose the favoredcustomsand qualities.55 We also find in Herodotus' History two actual communal heaps created at the behest of a king. The Scythianruler Ariantascommands each of his subjects,on pain of death, to bringhim a bronze arrowhead (for census purposes);then he transformsthe arrowheadsinto a huge bronze bowl which serves as a monumentto himself (Hdt. 4.81). When Darius leads his army againstScythia,he orderseach soldier to place a stone near the river Artescus and then leaves the great piles of stones behind (4.92);the stones could then have been used to count the Persian host, but Darius alreadyknows their number(4.87) and the collection simplyserves as a monumentto him and his great army.56Perhaps these actual,physicalheaps, formed from the contributionsof one and all, lie behind the abstract,imaginaryones. In any event, Herodotus elsewhere (7.152.2)mentions yet another hypothetical trial which involves the communalcollection of incorporealqualities,and in this case his languageis similarto that used by the authorof the Dissoi Logoi.57 The historianimaginesall men depositingtheir misdeeds (xax&)58into a common pile, with the intention of then exchangingthem with those of their neighbors.Each man, states Herodotus,will happilycarryback his own transgressions,after taking a close look (eyxvpUavTeg)at his neighbor'swrongs.Interestingly,in the Platonictale of a marketplaceof lives mentioned above, Er similarlynotes that at the marketplaceof lifestyles,the souls generallychose lives in accordancewith their previous existence (Rep. 620a). The souls, too, prefer the familiar and known.
55Hdt.3.38.1, Ei y6Q TLgJTQo0i al TLg... x?Ve6oL...
Jrat a aL vOQorotiO... xpXwvov,and Diss. Log. 2.18, jVTcag avOQcbjrcog. Cf. 2.26, ca' LVEg... ovJy7XcaEoaVTE xeXeiotLEv.
See also Christ 1994, 188 with n. 60. Significantly, Herodotus' parallel test at 7.152.2, discussed immediately below, is not worded in this way. 56See Christ 1994, 173-75 and n. 20 (references on the use of tokens in Persian census-taking). 57Compareesp. Diss. Log. 2.18,arlTLg T& aloXQ& eg evXEVE6OL oUvevexCtXL rdtvtag oVveveLXaLeV. avOcbjrwg, to Hdt. 7.152.2, ?e JrdvT;g a`v0Qw0rol xT oLXla Ixaxa eg iooV 58See Immerwahr 1966, 321 n. 37, for this meaning and compare xTg ETvXtag in the parallel anecdote attributed by Plutarch (Mor. 106b) to Socrates; mala sua is used in Valerius Maximus' version (7.2 ext. 2), where the saying is assigned to Solon.
TWO THOUGHT EXPERIMENTSIN THE DISSOI LOGOI
39
Let us return to the thought experiment outlined in the second chapter of the Dissoi Logoi and compare it for a moment with the linguistic trial in chapter 6. In his language experiment, the anonymous writer suggests sending a Persian child to Greece, in addition to dispatching a Greek child to Persia, as a parallel and complementary step. Each half of the two-part experiment then serves to reaffirm and reinforce the other. In the thought experiment on unseemly customs the author could have easily devised a similar companion piece to investigate the converse, that is, by having everyone bring honorable qualities to a joint pile and then take away dishonorable characteristics, again leaving no qualities behind on the heap. He does not do so, and the experiment thus has no control. And it lacks not only a control, but any sense of rigorousness or any means to prove-or disprove-the "results." While the trial is a lively and colorful presentation of the author's thesis, its outcome is neither wholly intuitive nor demonstrably true. In fact, the author subsequently attempts to overturn these results by outlining the experiment a second time, in order to argue for an outcome which is the very reverse. The Dissoi Logoi contains a whole series of double or twofold arguments, presenting both sides of various theses, and the author feels free to include the other side, as it were, of this thought experiment as well. When he tries to refute the first results, he does not argue against the original contention behind the trial, that ctaoXQ6vand xakov are relative terms which vary according to the context, so that what is unseemly for one person may well be seemly for another. Instead, wittingly or unwittingly, he substitutes a different thesis to be refuted, namely that c(LoXQ6vand xackoXv are one and the same thing, and then counters this revised argument by stating that the two qualities are distinct opposites.59 In this second account of the thought experiment, he first describes the original test as being presented by others (X70ovTL6e, dOgta' al Ta TLVEgTa a(oxQa ... ovvEvELxcatLv, 2.26; compare o4't 6'a'(Lo xeXeuoL ?v and ovevelxal, then its 2.18) argues against supposed xQa?g results. This time we are asked to imagine several people gathering together unseemly qualities from all nations, and then bidding each person to take away what is seemly. The author now argues that people would not take away shameful qualities from such a heap, because
590n this argumentation see Robinson 1979, 172 ad 2.21 (cf. 149-51, 162-63) and compare Barnes 1982, 519-22.
40
DEBORAH LEVINE GERA
shameful matters cannot be transformedor transmutedinto seemly ones simply as a result of being placed on a pile. He "proves"this contention, that what is unseemlyor ugly will not turn beautiful,by using a series of hypotheticalconcrete instances.If people had broughthorses, sheep, goats, or other people to such a collective heap, they would not have taken away something else in their stead. Nor would they have taken brass instead of gold, or lead instead of silver.Come now ((eQe 68, 2.28), he chides his reader, if someone had brought an ugly man, would he then take him away again beautiful?60If the linguisticexperiment proposed in the sixth chapter of the Dissoi Logoi elegantly discredits an actualexperimentperformedby Psammetichus,here we find the reverse situation:concrete counterexamplesare used as a control, as it were, to gainsayor refute the results of a thought experiment.Although these counterargumentsrefer to actual, materialobjects-corporeal animals,metals, and men-which can, in fact, be brought to a communalpile, unlike the acioxQd,the intangible,unseemlyqualitiesin the thought experiment,these tests are nonetheless phrased as unreal, "
counterfactual conditions (e.g., cd yowv I'tcog; j3)g oi 6g if &vO@dr)c0g aycayov, ovx akikoTi xa a&jdyov,2.27). These concrete counterinstances
are, it seems, even further removed from actual execution than the thought experiment,which is expressed as a hypotheticalfuture condition. The relation between armchairtrials and actual, performed experimentsis complex. At times, thought experimentslead to real-life tests, while on other occasions,actualinvestigationsinspirethoughtexperiments.61An interestinginstance of the dependence of a hypothetical test upon actualexperiencecan perhapsbe reconstructedfrom fragments of the works of the poet-philosopher Xenophanesof Colophon, a pioneer in the use of thought experiments.62Xenophanes objects to anthropomorphicdivine beings.He does not like the thieving,adultery, and deceptions which Homer and Hesiod ascribe to the gods (DK 21 Bll; cf. B12), and he criticizesthe belief that gods were born or wear clothing or have a voice and shape like humans (B14). In addition,he notes that Ethiopians say that their gods are black and snub-nosed, 60My translation and interpretation of Diss. Log. 2.26-28 is based on that of Robinson (1979, 113-15, 174-76). 61See Souder 1998, 5; Sorensen 1992, 11-12, 192-97; Naylor 1989. 62See Lesher (1992, 75-77), who includes DK 21 B2.15-19, B15, B30, B34, and B38 in this category.
TWO THOUGHT EXPERIMENTS IN THE DISSOI LOGOI
41
while Thracianssee the gods as blue-eyed and red-haired.He uses this fact to demonstratethat representationsof gods are not only anthropomorphicbut ethnomorphic,varyingaccordingto the differentkinds of worshipers(B16). Having moved from mankindin general to the different races, such as Ethiopiansand Thracians,he apparentlytakes his argumentone step further,to various species of the animal kingdom. Here he can no longerpoint to tangibleevidence such as poetry or plastic art,63and he turns to an imaginarytest, a proto-thought experiment,64to back up his claim that all species portraytheir gods in their own image (B15). He first establishes a hypotheticalsituation:we are asked to imaginethat cows, horses, and lions have hands and are capable of drawingwith them. These imaginaryanimalscan also make the things that men make (xctaeQyacTEXev areQ av6Qeg)and so presumably can create representationsof the gods. We now perform our experimentaltest and investigatehow each animalspecies would portray its gods. Xenophanes replies that horses would draw the forms of the gods like horses, and cows like cows, with bodies similarto their own: animals,too, would portraygods in their own image. Xenophanes' test has all the virtues of a well-conceived thought experiment.His vivid image of cows, horses, and lions drawingthe gods with their hands pulls us into his imaginaryworld and convincesby its very singularity.The use of three different types of animals,where in fact one would have sufficed to prove the point that even nonhuman species would create gods resemblingthemselves, serves as a control. Each of the three animal species behaves in precisely the same way, thus demonstratingXenophanes' thesis three times over. At the same time, the diverse shapes of the animals'gods remindus of the tremendous variety of deities fashioned by humans.We may laugh at first at the idea of a leonine, equine, or bovine god, but we then realize that the cows, horses, and lions are the precise counterpartof Xenophanes' Ethiopians and Thracians.These imaginaryanimals are good to think with, for they teach us about ourselves,humans.
63Clement (Strom. 7.22), the source for Diels's reconstruction of Xenophanes' remarks on the Ethiopians and the Thracians (DK 21 B16), refers to their painting (cf. 6LawoyQca()ovoc) the gods in their own image. For Xenophanes' likely acquaintance with Ethiopian mores see Snowden 1970, 104 (with notes, 281-82). 64Xenophanes' statement about animals' drawing gods in their image is not quite a thought experiment according to the criteria outlined above, because it is phrased as an unreal past condition, rather than a future hypothetical one.
42
DEBORAH LEVINE GERA
Another fragment from Xenophanes similarly demonstrates reasoning which is both colorful and hypothetical. If the gods had not made honey, the poet-philosopher states, people would consider figs far sweeter (DK 21 B38). This statement, phrased as an unreal condition, could easily have been cast in the form of an imaginary trial, with the inhabitants of a honeyless world first declaring figs to be the sweetest food and then changing their minds after being asked to taste honey for the first time. Another early philosopher, Heraclitus, supplies several other instances of hypothetical arguments framed in this manner, and these arguments point the way to full-fledged thought experiments.65 Xenophanes' skill in formulating armchair trials is not surprising, for elsewhere he shows a genuinely scientific bent, arguing on the basis of the physical evidence of shells and fossils that long ago everything was covered with mud (DK 21 A33). Xenophanes was among the first to use thought experiments, but the presence of two such tests in the Dissoi Logoi points to the widespread use of these trials by the beginning of the fourth century B.C.The two experiments found there illustrate both the best and the worst features of thought experiments. The linguistic test in chapter 6 is well conceived and carefully formulated, with a companion piece that serves as a control. It shrewdly casts doubt on the results supposedly obtained in an actual experiment, while at the same time illuminating the process by which children learn to speak a language. The second experiment we have considered, that of the heaps of unseemly qualities (in chapter 2), lends the sophist's argument color and immediacy but exposes the Achilles' heel of thought experiments: their lack of rigor and absence of concrete findings. One cannot, of course, easily overturn the results of an actual experiment, but our author has no difficulty in reversing the outcome of his hypothetical experiment. Indeed he may have deliberately constructed a test which was less than rigorous precisely because he wanted to argue both sides of a question. Since thought experiments do not provide tangible, empirical results, they are particularly suscepti-
65See G. E. R. Lloyd (1979, 68-69), who refers to Heraclitus frr. 7, 23, 99; one should, perhaps, add to this list as well Heraclitus' famous complaint against Xenophanes (DK 22 B40), that much learning does not teach intelligence; see also Lesher 1992, 75 n. 2. Rescher (1991) includes all of these instances (and more) in his discussion of the Presocratics and thought experiments, but his definition of thought experiments is quite broad and encompasses virtually any kind of hypothetical reasoning. Cf. Irvine 1991, esp. 153-56.
TWO THOUGHT EXPERIMENTSIN THE DISSOI LOGOI
43
ble to the sophistic stratagems favored by our writer, who frequently and blithely changes arguments midstream. Most interesting of all, perhaps, is that he finds thought experiments such a useful means of argument. These hypothetical tests are clearly one of the tools of the sophist's trade, an instrument he readily deploys. Scholars are agreed that the anonymous author of the Dissoi Logoi is not an original thinker but rather a compiler who summarizes the debates and controversies of the first generation of sophists. It seems likely that his use of thought experiments comes from them as well.66 The thought experiments he presents-whatever their sources-and the analogous trials found in Herodotus, Xenophon, and Plato indicate that in the realms of education, psychology, and ethics, thought experiments were a popular means of argument in classical Greece.67 HEBREW
UNIVERSITY
OF JERUSALEM
e-mail:
[email protected]
BIBLIOGRAPHY Barnes, J. 1982. The Presocratic Philosophers. London: Routledge & Kegan Paul. Benardete, S. 1969. Herodotean Inquiries. The Hague: Nijhoff. Borst, A. 1957. Der Turmbau von Babel. Vol. I. Stuttgart: Hiersemann. Brown, J. R. 1991. The Laboratory of the Mind: Thought Experiments in the Natural Sciences. London and New York: Routledge. Cantor, G. 1989. "The Rhetoric of Experiment." In Gooding, Pinch, and Schaffer 1989, 159-80. Christ, M. R. 1994. "Herodotean Kings and Historical Inquiry." CA 13:167-202. Conley, T. M. 1985. "Dating the So-Called Dissoi Logoi: A Cautionary Note." Ancient Philosophy 5:59-65. Crystal, D. 1997. The Cambridge Encyclopedia of Language. 2d ed. Cambridge: Cambridge University Press. Desbordes, F. 1987. "Aux origines de la linguistique: L'exemple des Dissoi Logoi." In Etudes de linguistique generale et de linguistique latine ... en hommage d Guy Serbat, 33-43. Paris: Soci6te pour l'Information Grammaticale.
66Itis possible that Protagoras was a direct influence on our author: see above, discussion leading to notes 12-14. 671 thankan anonymousrefereeof AJP for illuminatingcriticismsof an earlierversion of this essay.
44
DEBORAH LEVINE GERA
Dover, K. J., ed. 1968. Aristophanes: Clouds. With introduction and commentary. Oxford: Clarendon Press. Flory, S. 1987. The Archaic Smile of Herodotus. Detroit: Wayne State University Press. Gigon, 0. 1956. Kommentar zum zweiten Buch von Xenophons Memorabilien. Basel: Reinhardt. Gomperz, H. 1912. Sophistik und Rhetorik. Leipzig. Reprinted Stuttgart: Teubner, 1965. Gooding, D., T. Pinch, and S. Schaffer, eds. 1989. The Uses of Experiment. Cambridge: Cambridge University Press. Gould, J. 1989. Herodotus. London: Weidenfeld & Nicolson. Guthrie, W. K. C. 1969. A History of Greek Philosophy. Vol. III. Cambridge: Cambridge University Press. Harris, R., and T. J. Taylor. 1989. Landmarks in Linguistic Thought: The Western Traditionfrom Socrates to Saussure. London and New York: Routledge. Heidel, W. A. 1933. The Heroic Age of Science. Baltimore: Williams & Wilkins for the Carnegie Institution of Washington. Heinimann, F. 1945. Nomos und Physis. Diss. Basel. Reprinted Basel: Darmstadt, 1965. Horowitz, T., and G. J. Massey, eds. 1991. Thought Experiments in Science and Philosophy. Savage, Md.: Rowman & Littlefield. Immerwahr, H. R. 1966. Form and Thought in Herodotus. Cleveland: Press of Western Reserve University for the American Philological Association. Irvine, A. D. 1991. "Thought Experiments in Scientific Reasoning." In Horowitz and Massey 1991, 149-65. Kerferd, G. B. 1981. The Sophistic Movement. Cambridge: Cambridge University Press. Lesher, J. H., ed. 1992. Xenophanes of Colophon: Fragments. Toronto: University of Toronto Press. Lloyd, A. B. 1976. Herodotus Book II: Commentary 1-98. Leiden: Brill. Lloyd, G. E. R. 1979. Magic, Reason and Experience. Cambridge: Cambridge University Press. . 1991. "Experiment in Early Greek Philosophy and Medicine." In Methods and Problems in Greek Science, Selected Papers, 70-99. Cambridge: Cambridge University Press. Naylor, R. H. 1989. "Galileo's Experimental Discourse." In Gooding, Pinch, and Schaffer 1989, 117-34. Norton, J. 1991. "Thought Experiments in Einstein's Work." In Horowitz and Massey 1991, 129-48. Powell, A. 1988. "Mendacity and Sparta's Use of the Visual." In Classical Sparta: Techniques behind Her Success, edited by A. Powell, 173-92. Norman and London: University of Oklahoma Press.
TWO THOUGHT EXPERIMENTSIN THE DISSOI LOGOI
45
Rescher, N. 1991. "Thought Experimentation in Presocratic Philosophy." In Horowitz and Massey 1991, 31-41. Richardson, N. 1993. The Iliad: A Commentary. Vol. VI, Books 21-24. Cambridge: Cambridge University Press. Robinson, T. M., ed. 1979. Contrasting Arguments: An Edition of the Dissoi Logoi. New York: Arno. Reprinted Salem, N.H.: Ayer, 1984. Salmon, A. 1956. "L'exp6rience de Psamm6tique (H6rodote II, ii)." LEC 24: 321-29. Sansone, D. 1996. "Plato and Euripides." ICS 21:35-67. Snowden, F. M. 1970. Blacks in Antiquity: Ethiopians in the Greco-Roman Experience. Cambridge and London: Harvard University Press. Sorensen, R. A. 1992. Thought Experiments. New York and Oxford: Oxford University Press. Souder, L. 1998. 'A Way to Describe and Evaluate Thought Experiments (A Dissertation Proposal)." Internet home page for Thought Experiments. http://astro.ocis.temple.edu/-souder/thought. Stam, J. H. 1976. Inquiries into the Origin of Language. New York: Harper & Row. Sulek, A. 1989. "The Experiment of Psammetichus: Fact, Fiction, and Model to Follow." JHI 50:645-51. Taylor, A. E. 1911. Varia Socratica. Oxford: J. Parker. Reprinted New York and London: Garland, 1987. Thomason, S. G. 1991. "Thought Experiments in Linguistics." In Horowitz and Massey 1991, 247-57. Von Staden, H. 1975. "Experiment and Experience in Hellenistic Medicine." BICS 22:178-99. Wilkes, K. V. 1988. Real People: Personal Identity without Thought Experiments. Oxford: Oxford University Press. Williams, B. 1993. Shame and Necessity. Berkeley and Los Angeles: University of California Press.