Unique events: The Underdetermination of Explanation

AVIEZER TUCKER

UNIQUE EVENTS: THE UNDERDETERMINATION OFEXPLANATION

ABSTRACT. The paper explicates ‘unique events’ and investigates their epistemology.Explications of ‘unique events’ asindividuated, different, andemergentare philosophicallyuninteresting. Unique events are topics of why-questions that radically underdetermine alltheir potential explanations. Uniqueness that isrelative to a level of scientific develop-ment is differentiated fromabsoluteuniqueness. Science eliminates relative uniquenessby discovery of recurrence of events and properties, falsification of assumptions ofwhy-questions, and methodological simplification e.g. by explanatory methodological re-duction. Finally, an overview of contemporary philosophical disputes that hinge on issuesof uniqueness emphasizes its philosophical significance.

KEY WORDS: Events, Explanation, Unique, Underdetermination.

0. UNIQUE EVENTS: THE UNDERDETERMINATION OF EXPLANATION

Unique events raise five philosophical issues:

1. A philosophical explication of ‘unique events’.2. The epistemology of unique events.3. How can unique events be identified as such?4. Which events are unique?5. How can we know what can be known about unique events?

Previously philosophers debated whetherunique events are in somesense beyond the purview of science. Most philosophers skipped over is-sues 1–4 to answer 5, offering or rejecting alternative methods to whatphilosophers considered to be the scientific method, e.g. ideographicmethodologies, empathic understanding, colligation, narrative understand-ing & etc. This essay concentrates on the neglected 1–4 issues, postponinga discussion of 5. First, previous attempts to explicate ‘unique events’ asindividuated, different, andemergentevents are rejected as philosophicallyuninteresting: They do not raise issues that are of concern to philoso-phers such as the nature and limits of knowledge, the constitution ofthe universe, the content and limits of the scientific method etc. Second,

Erkenntnis48: 59–80, 1998.© 1998Kluwer Academic Publishers. Printed in the Netherlands.

60 AVIEZER TUCKER

I explicate ‘unique events’ as explananda that underdetermine all theirpotential explanations. Then, relative uniqueness is differentiated fromabsolute uniqueness. Scientists eliminate relative uniqueness by discov-ering recurrent events and properties, falsifying background assumptionsof why-questions, and utilizing simplifying methodologies such as ex-planatory methodological reduction. Finally, an overview of contemporaryphilosophical disputes that hinge on issues of uniqueness emphasizes itsphilosophical significance.

‘Unique events’ entered philosophic discourse at the later part of thenineteenth century following the establishment of psychology as a science.Psychology collapsed the traditional correspondences between scientificmethods and the realm of nature, and non-scientific methods and humansubject matter. Psychology seemed to share the methods of the naturalsciences and yet have a human subject matter. Windelband suggested toreestablish the distinction among the sciences on their value-laden method-ologies, on whether they are interested in describing events asuniqueor asfollowing universal laws, rather than on the domains that constitute theirsubject matters.1 The problem that Windelband and his South-WesternSchool of Neo-Kantianism were facing is just as contemporary today:Does science have any limits beyond the limits of our interest in applyingthe scientific method?

Within the context of analytical philosophy, the Neo-Kantian argu-ment was revived, though the meanings of ‘unique’ evolved. The possibleuniqueness of historical events was suggested as aprima-faciechallengeto Hempel’s (1965) obsolete covering law model or to the possibilityof scientific historiography.2 Recently, Murphey suggested that scientificknowledge of unique events is impossible or at least problematic.

. . . there are events that do appear to be unique in the sense that no presently imaginableclassification could bring them under a law-like generalization. The Big Bang in whichour universe began is a unique event, and certainly it is obvious that none of the Humeancriteria [of causation] can be applied to the Big Bang, at least as it is currently understood.Whether there are other such events is not clear, . . . (Murphey 1994, 106)

According to the general theory of relativity, when matter in space-time is compressed below a critical radius, the “Schwarzschild radius”(2GM/C2, where M=mass, G=gravitational constant, C=speed of light),it becomes a singularity, a region of infinite space-time curvature, wherethe general theory of relativity does not hold. The surface of a Schwarz-schild radius is anevent horizon, defining the border between unobservableunique naked singularity from which neither mass nor radiation canescape, and the non-unique universe described by science. Two astro-physical regions are singularities: Black Holes and the Big Bang.

UNIQUE EVENTS: THE UNDERDETERMINATION OF EXPLANATION 61

‘Black hole’ is a metaphor (introduced by John Wheeler in 1969 af-ter it had been defined theoretically) based on the gravitational pull thatprevents the escape of light and therefore anything else beyond the eventhorizon. Another singularity existed before a fraction of a second after thebeginning of the universe, referred to by the metaphor: “the Big Bang.” Atthe exact beginning, the size of the universe was zero (infinite density) andits heat was infinitely high. Under such conditions, all the laws of naturebreak (unique singularity).

It seems that at least in the case of singularities in relation to the gen-eral theory of relativity, there are events beyond the purview of science.We should look for an explication of uniqueness that explains how andwhy unique events are beyond the purview of the scientific method. Thisexplication is philosophically interesting because it concerns the limits ofknowledge and the applicability of scientific methods.

1.

In ordinary language, ‘unique’ is used in different contexts as synonymouswith ‘one and only one,’ ‘incomparable,’ ‘unmatched,’ ‘without equal,’‘unusual,’ ‘very special’ etc. But there is nothing to preclude a scien-tific account of objects or events that are referred to by subjects with thepredicate ‘unique’ in its ordinary language senses. The ordinary languagemeanings of ‘unique’ are philosophically uninteresting, they are irrelevantfor assessing the limits of knowledge or science.

Critics of Hempel’s (1965) covering law model of historiographicexplanation (I differentiatehistory, past events, from historiography, his-torians’ representations of history; the distinction parallels that of naturefrom science) claimed vaguely that if historical events are unique, theycannot be subjected to generalizing laws. As a criticism of Hempel thisattempt was ill-conceived. Hempel had a (bad) argument about the formalstructure of explanation, not about events and their relation with the al-leged covering laws. Still, the outdated dispute over Hempel’s covering lawmodel has been useful in focusing the discussions of uniqueness as a philo-sophical problem on scientific explanation, thereby connecting uniquenesswith changes in the philosophical understanding of explanation.

Dray differentiated what he termed a “trivial” meaning of‘unique’ froman “ordinary” one. Trivially:

any actual thing or occurrence you care to select for study isunique in the sensethat there is nothing else exactly like it.(Dray 1957, 45; my emphasis).

(U1)

62 AVIEZER TUCKER

Fetzer (1975) agreed that philosophers (unlike historians) use ‘unique’as a synonym of individuated. In his opinion, all events that are discerniblefrom each other are unique. There is nothing to prevent the scientific ex-planation of individuated events (Dray 1957; Hempel 1965; Fetzer 1975).Events may be individuated by their space-time characteristics and sciencemay abstract them, or by their location on a causal net, and science maydefine theoretically the causal relations. If ‘uniqueness’ is philosophicallyinteresting it must mean more than individuation of concrete particulars(Murphey 1994, 106).

Dray calls a second sense of ‘unique’, which he endorses, “ordinary.”As we noticed above, ‘unique’ has many senses in ordinary language; but Iam not interested here in discussing the pragmatics of ‘unique’. I evaluatethe merits of Dray’s endorsed “ordinary” sense of ‘unique’ irrespective ofwhether it actually represents its pragmatics:

. . . we ordinarily call persons and things unique, meaning that they are particular in certainrespects. Historical events and conditions are often unique simply in the sense of beingdifferent from others withwhich it would be natural to group them under a classificationterm(Dray 1957, 47).

Dray’s endorsed sense of ‘unique’ is:

Different in relation to a comparison group.(U2)

(U2) is vague. It is unclear whether Dray holds that there are natural groupsand ‘uniqueness’ means “difference from a natural group”; or whetherthe groups are conventional or pragmatic and ‘uniqueness’ is context de-pendent. Either way, Dray’s conclusion that (U2) events cannot have ascientific explanation is false. On the contrary, difference from a compar-ison group is a precondition for posing and answering a scientific whyquestion about any topic. A Why-question differentiates itstopic froma contrast class, a set of alternative propositions to the topic. Contrastclasses are necessary for clarifying the contextual meanings of questions.Accordingly, a scientific explanation suggests the reason(s) for the topic,rather than the contrast class, to give a contextual meaning to the answer(Van Fraassen 1980, 141–145). Since difference from a contrast classis necessary for contextually meaningful scientific explanations, (U2) isphilosophically uninteresting, it raises no special issues.

Fales revived the forlorn attempt to disprove Hempel’s covering lawmodel by arguing that historical events are unique:

. . . an interesting sense in which historical events are unique . . . is . . . . Like works ofart, historical events are culturally emergent. The criteria for identifying and relevantlycharacterizing them depend upon the cultural milieu in which they occur; and the traditions,

UNIQUE EVENTS: THE UNDERDETERMINATION OF EXPLANATION 63

social institutions, and legal systems which provide the context required to make sense ofhistory are the variable products, intended and unintended, of human cultural activity, notreducible to some uniform set of causal or even non-causal psychological laws (Fales 1980,275).

Unique events are:

emergent, cannot be reduced to the laws of a lower system3(U3)

Emergence does not preclude the possibility of laws describing the reg-ularities of emergent events and their micro-reduction (Kuipers 1990, 253following Looijen 1987). There are knowablesui generislaws of a highersystem that are irreducible to those of a lower level. Biological events donot have to be explained by physical theories to be explained scientifically.Social explanations need not to be reducible to psychology (Jackson &Pettit 1992). Repetitive (U3) events are not necessarily beyond the purviewof science and therefore are not interesting.

A philosophically interesting explication of uniqueness is likely to fitthe evolving philosophical understanding of science. A naiveinductiveconcept of science implies that it is impossible to infer general scientificlaws from events that occuronce and only once(Cf. Murphey 1994, 106).This concept of unique:

Occurring once and only once(U4)

accounts for the etymology of ‘unique’ from the latinunicus, related tounus, one. When ‘unique’entered English, at the beginning of the seven-teenth century, it meant single or sole. But properties of (U4) events may beshared by other events; science may study them separately and use themin inductive reasoning. Still, this early naive concept of uniqueness con-tains a basic insight. It should be adapted to a contemporary, sophisticated,understanding of science.

2.

A century old tradition suggests that an interesting explication of uniqueevents should demonstrate that they cannot be explained scientifically. It isnecessary then to outline an analysis of scientific explanation. I follow vanFraassen’s account (1980, 141–157) because I find it useful in stressing thecontextual conditions of explanation:

An explanation is an answer to a why-question, and a scientific expla-nation is an answer to a why-question that draws on scientific theories.Every why-question has three components:

64 AVIEZER TUCKER

1. Thetopic of the question (a proposition)4.2. Thecontrast class, a set of alternatives.3. Explanatory relevance, the respect in which an explanation is sought.

Accordingly, the question assumes that:

1. The topic is true.2. In the contrast class, only the topic is true.3. At least one of the propositions that bears its relevance relation

to its topic and contrast-class, is also true. Why-questions arise inthe context of background information or theories that imply theseassumptions.

Evaluations of answers are guided by three criteria:

1. The answer is acceptable or likely to be true according to backgroundtheory and information.

2. The answer with background theory and information increase theprobabilistic gap in favor of the topic in relation to the other membersof the contrast class.

3. The answer is better than alternative answers in three respects:

3.1 The background information and theory increase the probabilisticgap between the answer and its alternatives;

3.2 the answer favors the topic to a greater extent than other answers;3.3 the answer is not made wholly or partially irrelevant by other

answers that “screen” it.

Some why-questions do not have scientific answers because there isno available scientific theory that favors an answer in comparison to al-ternative answers, and with the answer increases the probabilistic gap infavor of the topic. Van Fraassen’s example is of paresis of a group ofpatients as the topic when the contrast class is syphilitics who did notdevelop paresis; currently it has no scientific explanation because there isno scientific theory that favors the topic (paresis) in relation to the contrastclass (syphilitics).

Van Fraassen’s contextual analysis of explanation disarms the argumentthat since it is impossible to describe, let alone explain, the myriad aspectsof any event, scientists describe and explain only aspects of events that fittheir theories. Thus, scientists can offer a partial explanation of any eventby concentrating on those of its aspects that can be explained by an ac-ceptable theory (Fetzer 1975). Hull argued similarly that unique5 historicalevents can be explained scientifically.

UNIQUE EVENTS: THE UNDERDETERMINATION OF EXPLANATION 65

Caesar’s crossing the Rubicon . . . was a unique event . . . . But this event, unique thoughit may be, can be assimilated to a variety of reference classes; e.g., people crossing rivers,generals disobeying orders, ambitious men making their big moves. Certainly under someof these descriptions the event is an instance of a significant generalization, though perhapsnot a law of nature (Hull 1975, 268).

Fetzer and Hull disregarded the context of explanation. Events do not fallinto the world wearing T-shirts that exclaim “Explain me!” Somebody hasto ask a why-question whose topic is an event before an explanation canbe attempted. The context of why-questions stipulates which aspects ofevents are the topics of questions. Hull’s descriptions of aspects of Cae-sar’s crossing of the Rubicon as instances of classes are not the topicsof usual historiographic why-questions. Historians do not usually care ifCaesar crossed an existing bridge, constructed a new one, crossed the riverwith ferry-boats or at a shallow point. In ordinary historiographic contexts,Caesar’s crossing of the Rubicon was a historical-political event. Relevantanswers must explain the decision of Caesar to break Roman custom andmove his troops against the city of Rome in a bid to conquer it. Relevantcontrast classes can be previous Roman leaders who had not crossed theRubicon with their armies, or Caesar’s alternatives (withdraw, cross theRubicon without his army, negotiate & etc.); and the usual relevance cri-teria have to do with political and military factors. Hull’s explanations areirrelevant in ordinary historiographic contexts.

Van Fraassen drew an important distinction betweenhaving a scientificexplanation(the theory is acceptable and explains) andthe theory explains(but is not necessarily acceptable). If acceptable scientific theories cannotbe relevant for the explanation of certain events as topics of why-questions,and the theoretical models that are relevant for explaining these eventscannot be scientifically acceptable, thescientificbackground informationthat van Fraassen’s three criteria for the evaluation of answers/explanationsrequire is missing. I argue that there are no acceptable scientific theoriesthat can explain unique events because unique events:

Occur once and only once. Their significant properties or para-meters, specified in the topic description of the why-question,are either:

(a) Not shared by any other event, apart of spatiotemporal lo-cation and self-identity6 (or it is unknowable whether theyare shared by other events). Or

(b) Too complex or chaotic for effective comparison with otherevents.7

(U5)

66 AVIEZER TUCKER

(U5a) and (U5b) events share the exclusion of the kind of evidence thatis necessary for scientific acceptance of a theory that is relevant for theirexplanation. There are two main accounts of theory acceptance:Confirma-tion theories(Van Fraassen 1980, 57-8) and theories of inference to thebest explanation. I demonstrate next that either way, the (U5) explicationof unique events implies the underdetermination of their explanations.

Glymour and Van Fraassen agree that scientific theories are differ-entiated from non-scientific ones in their evidential confirmability andconfirmation (Glymour 1980, 122). Scientific confirmation of theories re-quires a variety of recurring events and relevant properties to ascertainthat a positive case is not coincidental anecdotical evidence, to help decideamong competing theories, and to determine which of the hypotheses thatcompose a theory does a piece of evidence test. When several theorieshave an identical empirical basis, further empirical evidence is soughtto help decide among them. Otherwise, scientific theories cannot escapeQuine’s (1975) underdeterminations. Glymour claimed that establishedscientifictheories have a sufficient variety of evidence of recurring eventsand parameters for repeated testing (Glymour 1980).

When the topic of a why-question is (U5), all possible theoretical back-grounds that are relevant for its explanation (a theory or an isolated part ofit) can have only a single positive confirming case, the (U5) event itself.(U5) events exclude the possibility of finding further evidence for testingthe relevant theoretical backgrounds that participate in explaining them.(U5) events can falsify some theories or be the single positive case ofothers. But, by definition, (U5) events exclude the possibility of otherevents that can be relevant for confirming the theoretical backgroundsthat can participate in their explanation because: Unique events are notrecurrent; their significant properties are either not shared by any otherevent or it is impossible to know whether other events share them; orthey are too chaotic and complex for effective comparison with otherevents. Explanations of unique events are radically underdetermined be-cause any theoretical background that is relevant for their explanation isunderdetermined.

The underdetermination of explanations by unique events is more rad-ical than Quine’s (1975) underdetermination of theories. Quine arguedthat any theory that accounts for afixed setof observation sentences mayhave an alternative theory, logically incompatible and irreconcilable withit that implies that set. Quine postulates the fixed set of observation sen-tences as the empirical basis of theories. If Quine’s observation sentencesexist, a unique (U5) event can generate a potentially infinite set of irrel-evant observation sentences of its various observable aspects. Still, such

UNIQUE EVENTS: THE UNDERDETERMINATION OF EXPLANATION 67

an infinite set does not suffice for deciding among competing incompat-ible theories that explain it because the observation sentences could notpossibly be describing regularities, nor can such regularities be deducedthrough conjunction with observation sentences of other events becauserelevant properties of the unique event are non-repetitive or too complex.In other words, observation sentences are not as significant as their objectsof observation.

(U5) explication of uniqueness holds just as well in accounts of the-ory acceptance that rely oninference to best explanation. A version ofthis position suggests that the best theories are selected according to theirconsilience, simplicity and analogy (Thagard 1988, 75–99). Consilienceexpresses the number of classes a theory participates in explaining. A the-ory is more consilient than another if it explains more classes of facts. “Ininferring the best explanation, what matters is not the sheer number of factsexplained, but their variety and relative importance” (Thagard 1988, 81).Simplicity is inverse of the number of ad hoc auxiliary assumptions neces-sary for an explanation. Ad hoc assumptions participate in the explanationof a narrow range of cases (assumptions shared by competing theoriesare not ad-hoc). Theories can become more consilient on the expense oftheir simplicity, if they add auxiliary assumptions to explain more cases;and vice-versa, become more simple on the expense of their consilience.Thagard’s basic formula for computing the values of theories is:

Simplicity× Consilience = Value

Simplicity is measured by the inverse of the number of assumptions (cohy-potheses) that must be added to the theory/hypotheses in order to explain.“If a hypothesis has more cohypotheses than facts explained, its simplicityis judged to be 0, since it needs a special assumption for everything itexplains . . . ” (Thagard1988, 90). This seems to describe the explanationsof some events in historiography and psychoanalysis (Tucker, 1993). Eachof the main theories (Freud, Marx, Braudel, Jung, etc.) requires revisionsand added ad hoc assumptions to be able to explain. For example, psycho-analysts add to their core beliefs different ad hoc assumptions to explainclinical cases. The ad hoc theory:

is based on a mixture of theory and private language seasoned with a sprinkling of bodywisdom, some vague memories of clinical moments, and perhaps a mixture of other sen-sations aroused by the ambience of the consulting room. . . . Some of this mixture can bedistilled into such well-known concepts . . . which can then be exchanged at meetings andargued over in discussion groups . . . (Spence 1994, 200–201; cf Tucker 1995).

Simple theories cannot explain (U5) events because they require ad hocadditions to explain events that do not occur elsewhere and whose proper-ties are not known to occur elsewhere, or are too complex for comparison.

68 AVIEZER TUCKER

Non-simple consilient theories are possible in the case of unique events.Certain core beliefs, theoretical assumptions, (e.g. Freudianism) may par-ticipate in the explanations of several (U5) events, but they need a host ofdifferent and even inconsistent ad hoc assumptions for each explanation. Itis impossible to choose among non-simple consilient theories that explain(U5) events because according to Thagard’s criteria they are underdeter-mined, their value in Thagard’s computation is 0 because they need moread hoc assumptions than the number of cases they explain. Theories thatexplain (U5) events are scientifically unacceptable because: “. . . we clearlydo not want toaccepta theory merely on the basis that there is somethingthat it explains” (Thagard 1988, 75).8

(U5) events are surrounded by a “cloud” of underdetermined explana-tions. This “cloud” indicates the possibility of a unique event within anepistemic “black hole.” For example: Suppose the topic of a why questionis the Big Bang. The contrast class is nothingness. The relevance conditionis physical and causal: The question is interested in the kind of causes thatphysics deals with (as distinct of teleological reasons and divine creation).Scientists have discovered only of a single case of a Big Bang (if it happensregularly as theories of the contracting universe suggest we cannot possiblyreceive any information about previous “Big Bangs”). The basic physicalproperties of the Big Bang such as infinite heat and density are not sharedby any other event because infinity is not a degree of heat or density thatcan be expressed by valued parameters. Consider an explanation of the BigBang that is based on the theory that “Nothingness is unstable and tendsto create bubbles of matter.” It may explain the Big Bang. But how can thetheory be confirmed? There is only a single confirming case, the Big Bangitself! Any competing theory of the Big Bang will be underdetermined aswell (van Inwagen 1993, 115–118).

Another example: The topic of the why-question is the French Revo-lution. The contrast class can be Britain in 1789 or France in 1750 (i.e.the question asks for relevant causes of the French Revolution that werepresent in France and not in Britain or in France only after 1750); andthe relevance condition asks for factors that historians are interested inlike institutional structures, economic changes, ideological developments,etc. The French Revolution happened only once. Many of its historio-graphically relevant properties, perhaps all, can be found elsewhere, buttheir unique combination in 1789 may be so intertwined, complex andchaotic that it is impossible to combineacceptabletheories to produce,or deduce from, a theory that can explain it. There are many competinghistoriographic theories that explain the French Revolution with no singleexplanation emerging as better than the others.

UNIQUE EVENTS: THE UNDERDETERMINATION OF EXPLANATION 69

3.

Historically, many events that had been considered (U5) have eitherdisappeared or received scientific explanation, either by refuting the as-sumptions of the why-questions in the context of which the unique eventsappeared as topics, or by discovering that the events that appeared (U5) arenot unique after all.

For example, the appearance of individual comets had appeared to be a(U5) event for most of history.

In 1686, when Newton published his theory of gravitation, there were . . . two currenttheories concerning comets. The more popular one regarded comets as signals from anangry God warning that He will strike and bring disaster. A little known theory of Kepler’sheld that comets were celestial bodies moving along straight lines. Now according toNewtonian theory, some of them moved in hyperbolas or parabolas never to return; othersmoved in ordinary ellipses. Halley, working in Newton’s program, calculated on the basisof observing a brief stretch of a comet’s path that it would return in seventy-two years’time; he calculated to the minute when it would be seen again at a well defined point in thesky. This was incredible. But seventy-two years later, when both Newton and Halley werelong dead, Halley’s comet returned exactly as Halley predicted (Lakatos 1978, 5).

Halley’s discovery involves the major methods by which uniqueness ofevents is dissipated: Refutation of the assumptions of why-questions andscientific discovery: Every why question assumes that a proposition aboutthe topic is true, the contrast class is not true, and at least one of thepropositions that bears its relevance relation to its topic and contrast-classis also true. If one of these assumptions is proved false, the why-questioncollapses. For example, suppose a contemporary of Halley’s asks a why-question whose topic is the last appearance of the cometary harbinger ofill-tidings, the contrast class is the planetary astrological influences onhuman lives, and the relevance conditions have to do with astrologicaltheories of bad omens. The question is: Why did the cometary harbingerof ill tidings appear while the planets appear regularly? Possible relevantanswers would relate the topic to previous bad omens or to human sinsthat caused supernatural retribution & etc. Halley would have to answerthat the topical proposition is not true, the comet is not a harbinger of ill-tidings, nor is one of the propositions that bears a relevance relation to thetopic true.9

Scientists discover that events that were considered unique are notunique and introduces explanations that answer the original why-questionabout them by discovering that the events are recurrent, by discoveringthe properties of events that were considered (U5a) elsewhere; and byinventing techniques for managing the complexity of (U5b) events.

70 AVIEZER TUCKER

It is useful to introduce herea distinction between absolutely uniqueevents (U5) and relatively uniqueevents (U6):

All events that seem (U5) relative to a given context of scientificdevelopment and theoretical perspective, but may be discoveredas non-unique relative to a more advanced level of science.

(U6)

The discovery of the recurrence of erstwhile (U6) events creates the bodyof evidence that can confirm theories that explain them, e.g. the discov-ery of the recurrent appearances of Halley’s comet. Similarly, scientificmedicine discovered that many groups of clinical symptoms that wereconsidered unique by traditional medicine were actually tokens of diseasetypes, and consequently could be studied scientifically. The discovery offurther occurrences of events may be guided by a scientific theory, as inthe case of Halley’s comet and Newtonian physics. The repeated effortsof scientists to replicate events can be interpreted as either the creation ofa sufficient body of evidence for the confirmation of theories that in turndecide among competing explanations, or as increasing the value of theo-ries by increasing their consilience. Some classification efforts in sciencesas diverse as botany, biology, astronomy, medicine and psychology do justthat. A scientific theory such as evolution may guide the classification ef-fort. Classifications may change with scientific theories, e.g.nebulaewereclassified as early stages in the development of solar systems such as ourown.

Scientists attempt to find further occurrences of the properties of (U6a)events (that are considered (U5a) relative to a level of scientific develop-ment). For example, the first discovery of pulsar radio emissions generatedunderdetermined ad hoc theories and explanations (including inter-stellar“light-houses” designed by intelligent life-forms). But then astronomerssearched the heavens and found other stellar objects with these properties,thus determining an explanation of pulsar radio-emissions as generated bycollapsing neutron stars.

Post-general theory of relativity astro-physics has attempted to discoversome universal properties of black holes. Astro-physicists attempted todemonstrate that in certain respects black holes have the properties ofnormal physical objects. For example, it is possible to measure the gravi-tational effects of black holes on observable non-unique stellar objects anddeduce a mass thatunder normal conditionswould create such effects.Hawking (1988) exemplifies the next stage in the scientific treatment of(U6a) events. Hawking attempts to use a unified theory combining quan-tum mechanics and gravity to infer recurring properties of singularities thatcould then “bootstrap” the theory: Matter falling into black holes causes

UNIQUE EVENTS: THE UNDERDETERMINATION OF EXPLANATION 71

rotation, a magnetic field, and jets of ejected particles. The area of the eventhorizon may measure entropy. If black hole singularities have entropythey must have a temperature in inverse relation to mass, and radiation.The radiation can be created when virtual (quantum physics) particle-anti-particle pairs are created according to the principle of indeterminacy inthe space surrounding black holes. When one of the particle pairs fallsinto the black hole as negative real particle, it annihilates with its opposite,thus reducing the mass and increasing the temperature of the black hole,while the positive energy particle in the pair radiates outward of the blackhole area, creating radiation commensurable with increase in temperature.A flow of negative energy reduces the mass, the event horizon, and theentropy inside the black hole, causing an increase in entropy in the universeoutside the black hole. If Hawking’s theory is successful, black holes havethe usual physical properties and are not unique.

Hawking espouses further the creation of quantum theory of gravity toexplain the Big Bang and the earliest stages of the universe, thus elim-inating (U6) events from physics altogether. Whether or not Hawkingis successful, his attempts demonstrate the scientific approach to appar-ently unique events, the search for repetitive properties. Currently, we stillfind “around” singularities the “cloud” of underdetermined theories thattypically surrounds (U6) events.

Scientists devote much effort for the simplification and managementof (U6b) complex events (that are too complex for determining scientifictheories relative to a level of scientific knowledge). The dominance of sta-tistical methods and the rapid spread of chaos theory and system theory inthe various branches of science are due to the increased efforts of scientiststo study increasingly complex events. Technical tools such as comput-ers are helpful as well. One scientific method that is mentioned often inconnection with scientific attempts to explain (U6b) events isreduction.

Wimsatt suggested that the kind of reduction scientists use and discussmost (in contrast to philosophers) isexplanatorymethodological reductionin which the familiar properties of higher level entities are explained interms of the unfamiliar properties and interrelations of lower level en-tities. Wimsatt characterizes “level” as “local maxima of regularity andpredictability in the phase space of different modes of organization ofmatter . . . the simplest and most powerful theories will be about entities atthese levels.” (1984, 484) Explanatory reduction is valuable for:

. . . cases that are anomalies for or exceptions to the upper-level regularities. Since ananomaly does not meet the macro-regularity, the macro-regularitycannot “screen-off”the micro-level variables. If the class of macro-level cases within which exceptions occuris significantly non-homogeneous when described in micro-level terms,thengoing to alower-level description can be significantly explanatory, in that it may be possible to find

72 AVIEZER TUCKER

a micro-level description partitioning the cases into exceptional and nonexceptional onesat the macro-level. We would then have a micro-explanation for the deviant phenomenon.(Wimsatt 1984, 492)

The anomalies that explanatory methodological reduction should explainmay be (U6b) events. Explanatory reduction is a common method bywhich scientists expand science when encountering (U6b) events. Whenscientists face many complex properties of an event that occurs only once,they attempt to explain them by other properties of a more manageablenumber and complexity from a lower level (Ruse 1989). The “depth” ofthe methodological reduction of the properties of (U6b) events dependson the complexity/ simplicity of lower levels. Beyond a certain level, therelations among the reducing properties may be more complex than thosethey are supposed to reduce, and the reason formethodological(as distinctof ontological) reduction disappears. For example, Dalton’s atomic theoryexplained (U6b) events in terms of smaller and more manageable numberof recurring properties of atoms and their less complex interactions.

A sufficiently powerful reductive theory in Nagel’s (1961) (inter-theoretic) sense may determine one of the underdetermined theories thatexplain (U6b) events by deducing it. The bridge laws that should connectreducing and reduced theory may be confirmed because the properties/ pa-rameters of (U6b) events can be recurrent. But as critics of Nagel’s modelof reduction argued (Feyerabend 1962; Brittan 1970; Schaffner 1984; Ruse1984; Churchland 1985), such derivation reduction is rare in the his-tory of science. Schaffner (1984), Ruse (1984), Churchland (1985), andChurchland & Churchland (1995) suggest a weaker model of epistemicintertheoretic reduction in which reducing theories entail only a “stronganalogy” or “isomorphic approximation” of reduced theories and can ac-count for the extensions, evidence, explanatory and predictive powers ofthe reduced theory. Feyerabend (1962) and Hull (1984) claimed that thederivation of analogies or approximations of theory/concepts is actuallytheory/conceptreplacement. Replacing theories share the extensionalityand exceed the predictive powers of the preceding theories (otherwise theydo notreplace) but they are not analogous to them.

When a scientific theory succeeds underdetermined theories that ex-plain (U6) events, we are likely to discover theoryreplacementbecausethe motivation for weak intertheoretic reduction is absent. Wimsatt (1984,482) lists three reasons for wishing to establish weak reductive relationsbetween succeeding scientific theories: First, following Nickels (1973),weak reduction gives prepackaged confirmation of the reducing theory byshowing that it generates the reduced theory as a special case. Underdeter-mined theories that explain (U6) events cannot support the confirmation of

UNIQUE EVENTS: THE UNDERDETERMINATION OF EXPLANATION 73

theories that reduce them. Second, following Sklar (1967), weak reduction“explains away” the old theory, or explains why we were tempted to be-lieve in it. Underdetermined theories are not competitive enough to require“explaining away.” Third, following Nickles (1973), a weak reduction de-limits acceptable conditions for use of a reduced theory as heuristic device,by determining conditions of approximation. Underdetermined theorieshave little or no heuristic value once a proper scientific theory replacesthem. Churchland & Churchland add reasons for weak reduction: “thereduction provides us with a much deeper insight into, and thus a more ef-fective control over, the phenomena within the old theory’s domain” (1995,71); and the reduction provides us with a simpler account of its domain andexplains more phenomena than the reduced phenomena. This may be true,but these reasons are equally supportive oftheory replacement. Church-land’s (1985) interest in weak epistemic intertheoretic reduction originatedin a search for a model for the epistemic reduction of folk psychology toneurological science that preserves the ontology of folk psychology, thethings and properties it postulates. But our belief in one of the inconsistentand underdetermined theories about (U6) events is not necessarily as en-trenched as our belief in folk psychology. Scientific theories of comets anddiseases replaced rather than reduced previous theories.

4.

Two kinds of philosophical controversies are related to uniqueness as ex-plicated here. First, there are debates on whether certain events are (U5)or (U6), whether the events underdetermine any scientific explanation ofthem or whether these events do not have a scientific explanation onlypresently and this may or likely to change as science advances. Secondly,there are disputes on whether or not certain explanations are currentlyunderdetermined by (U6) events.

The possibility of methodological reduction of certain properties is de-bated in the philosophies of biology, mind, history, psychology, and thesocial sciences. Some reduction debates in philosophy are actually aboutwhether certain events are (U5b) or (U6b). Campbell (1997, 128–129)expressed optimism that all the events in nature (including black holes)may be methodologically reduced to a set of basic recurring tropes. Oppo-nents of methodological reduction in these fields claim that certain typesof properties are irreducible; or areholistic, dependent on other propertiesof the higher level; or are complex beyond human and machine computa-tional capacities. There are generic counter-arguments for some of thesearguments (Hoyningen-Huene 1989). The methodological reducibility of

74 AVIEZER TUCKER

individual mental events is debated. “It may indeed be unrealistic to expectan exhaustive global account of the neural and behavioral trajectory of aspecific person over any period of time. The complexity of the neural sys-tems we are dealing with may for ever preclude anything more than usefulapproximations to the desired ideal account” (Churchland & Churchland1995, 74). The uniqueness of historical events is debated in the philosophyof history. There have been periodic arguments on whether some histori-cal events are unique and therefore scientific historiography is impossible(Tucker 1998).

The resolutions of these methodological reduction debates as well asother debates on the relative vs. absolute uniqueness of events are likely tocome from scientific research, trial and error, rather than from philosophicspeculation: “The prospects of an empirical research program cannot besettled from an armchair” (Schwartz 1991, 204).

Arguments on whether existing explanations are underdetermined by(U6) events abound in historiography, clinical psychology and theirphilosophies. Historiographic explanations and theories proliferate aroundpossible unique historical events such as the Magna Charta; the Ameri-can Revolution; the Industrial Revolution; Imperialism; and the two WorldWars (Marwick 1989, 328–374). Psychological explanations and theoriesproliferate around mental events and personalities (Rappard, 1993).

Some argue that historiography uses concepts (states, revolution, wars,etc.) that are concrete particulars only in the sense of taking place in a defi-nite time and space to describe historical events, otherwise, historiographicconcepts refer to several historical events (Joynt & Rescher, 1961). Othersclaim that historiographies attempt to explain a variety of (U6) events(e.g. the Second World War, Caesar’s crossing of the Rubicon). Thoughhistorians have research programs like scientists, they are interested inexplaining certain “important” events and fit, adapt, interpret and add adhoc to theoretical models and conceptual frameworks they develop or bor-row for explaining these events. Historians belong to schools that havetheoretical core beliefs (Annales, Marxist, old history of ideas, sociologyof knowledge, intellectual history, internal/external history of science etc.)but each historian has to reviseand add ad hoc assumptionsto build atheory that can explain an (U6) event. Historians can have inconsistentunderdetermined theories that explain identical (U6) events (Tucker 1993,1998).

Murphey (1994) defends historiographic explanations and theoriesfrom the charge of underdetermination. He holds that in historiography,as in science, a confirmed theory increases the probability of a set of datamore than alternative theories. Further observations decide among com-

UNIQUE EVENTS: THE UNDERDETERMINATION OF EXPLANATION 75

peting theories. When a scientific theory seems empirically challenged,repeated experimentstest its least probable hypotheses. When

there is a fixed body of observational data, it is no surprise that alternative theories canbe created which are able to explain those data equally well. It may even be that manyof these alternative theories are logically incompatible with each other and that they arepairwise irreconcilable by reconstrual of predicates. But . . . it does not follow that theywill also be equivalent when it comes to predicting new and hitherto unknown phenomenaor in accounting for observations whose relevance had not been previously detected . . .(Murphey 1994, 232–233).

In Murphey’s opinion, new data and new ways to extract new data fromknown evidence are constantly discovered, creating new historiographicobservation sentences that can determine explanatory theories. But Mur-phey does not consider the possibility that some historical events may be(U6). I find it doubtful whether there is a clear resolution of some histo-riographic explanatory debates (Tucker, 1997a, 1997b). Only an empiricalexamination of historiographic practices can decide whether Murphey isright.

If there are (U5) events, there are serious doubts concerning the epis-temic usefulness of causal theories of events (Davidson 1980, 163–180)and properties (Shoemaker 1997). If there are (U5) unique events, it isimpossible to determine their causes or effects (relative to a level of scien-tific development or absolutely), and therefore it is practically impossibleto individuate events and properties. This is particularly problematic forShoemaker who holds that:

My reasons for holding this [causal] theory of properties are, broadly speaking epistemo-logical. Only if some causal theory of properties is true, I believe, can it be explained howproperties are capable of engaging our knowledge, and our language, in the way they do.

We know and recognize properties by their effects, or, more precisely, by the effects ofthe events which are the activations of the causal powers which things have in virtue ofhaving the properties . . . .

. . . The behavior of objects, that is, the displays of their powers, will reveal that theyhave certain conditional powers. Once it is discovered that certain conditional powers areconnected in a lawlike way, we can use these to ‘fix the reference’ of a property term tothe cluster containing those conditional powers and whatever other conditional powers arerelated to them in the appropriate lawlike relationship. And we can set about to determineempirically what the other conditional powers in the cluster are (Shoemaker 1997, 236,246).

True, there is a causal chain that links events with knowledge of them.But it is impossible to know the cluster of conditional powers that uniqueevents have. For example, it is trivial to note that we know of historicalevents, e.g. the economic crash of 1929, through a causal chain that linksthe event with our knowledge. But theories about the clusters of causes and

76 AVIEZER TUCKER

effects of possible unique historical events are underdetermined, e.g. thereare many inconsistent theories about the causes and effects of the 1929crash.

5.

Since uniqueness cuts across the disciplines, it is as clear today as it was acentury ago that the applicability of science to explain events does not de-pend on the domain of nature where the events happen: Psychology is ableto explain scientifically some human events, and physics may not be able toexplain singularities. The purview of science is clearly not just a matter of avoluntary axiological choice of method, as Windelband asserted. The prop-erties of events in conjunction with the context of inquiry as manifestedin the formation of the why-question decide the possibility of scientificexplanation. Further philosophical investigations of uniqueness should ex-plore comparatively unique events and the extent of underdetermination ofexplanations by unique events in the various disciplines.

Between the class of unique events and the class of scientificallyexplainable events, there may be a class of semi-unique events, whicheither havesomesignificant properties (specified in the topic of the why-question) that are non-recurring while others are recurring; or though alltheir significant properties are recurring the number of recurrences is in-sufficient for the determination of one theory that explains them. Spaceis too short for analysing here the epistemology of these “semi-unique”events. But it could be noted that scientific partial explanations or statisti-cal/probabilistic correlations may be possible in the first case. In the secondcase, the number ofactualsurviving inconsistent theories that participatein explaining such events may be less than in the case of unique events,though logically the number could still be infinite. Semi-unique eventsdeserve an article by themselves.

Further studies can compare and evaluate the various non-scientificalternative methodologies that have been mooted for the explanation ofunique events and their alternative models of knowledge: e.g. Verste-hen, hermeneutics, narrative explanation, empathy, and phenomenologicalepoché.

NOTES

1 Windelband held thatnomotheticscience is interested in eternal and unchanging laws ofnature that have many instances. Theideographicsciences are interested in: “(1) Complete

UNIQUE EVENTS: THE UNDERDETERMINATION OF EXPLANATION 77

and exhaustive description (2) of a single, more or less extensive process (3) which islocated within aunique, temporally defined domain of reality” (Windelband 1980, 174 myemphasis).

Windelband’s distinctions are not useful. There is no exhaustive description of anything.Science describes and explains short and long events and processes. Describing eventsas taking place in a unique space and time (individuation) does not preclude scientificexplanation of the events. Windelband’s examples are not helpful either. He thought thatone phenomenon may be from one perspective unchanging and subject to nomologicalexplanation, and from another perspective transitory and subject to ideographic descrip-tion. His example: Language has unchanging formal laws, but is a transitory phenomenonin the history of languages. Windelband confused conventional rules with formal laws.Linguistic change does not preclude the possibility of scientific diachronic linguistics.Windelband presented systematic organic biology as nomothetic and evolutionary biologyas ideographic. But though the evolutionary process is not known to be repetitive (e.g. onanother planet) evolutionary theory explains it scientifically.2 This attempt was probably inspired by Windelband, though he saw no contradictionbetween covering laws and unique events. Windelband foreshadowed Hempel’s coveringlaw model of explanation: “Every causal explanation of any historical occurrence presup-poses general ideas about the process of things on the whole . . . natural laws of events. . . of mental events or psychological processes” (Windelband 1980, 182–3). Windelbandheld that from the perspective of causality every event assumes a syllogism whose majorpremise is nomothetic (a law of nature) and minor premise is ideographic (conditions). Butthe description of events as minor premises in scientific syllogism must use the conceptualframework of the covering law to connect with it, so it cannot be ideographic.3 Fales did not prove that hypothetical historiographic laws cannot be reduced by the lawsof psychology with the necessary help of initial and boundary conditions. Hoyningen-Huene (1989) analyzed the generic anti-reductionist argument of which Fales’ argument isan applied case, and proved its indecisiveness.4 Erkenntnis’ anonymous referee noted that topics are distinct of propositions and only thesecond can be true or not. Yet, van Fraassen (1980, 141) equates topics of why-questionswith propositions, and holds that topics of why questions are assumed to be true (143).5 Hull (1975) held that when historians and biologists speak of ‘uniqueness’ they mean arigid designation of an individual sequence, referred to by proper names. In this sense, asHull claims, unique events can have scientific explanations.6 (U5a) should not be confused with individual properties or haecceity. The propertiesreferred to in (U5a) can be shared by other events, though they are not in this world.7 I assume here a realist ontology of properties. But it is possible to “translate” (U5)into the terminologies of some alternative ontologies. In an ontology that regardstropesas basic entities and conceives events as trope sequences (Campbell 1997),unique tropesdo not resemble any other tropes (thus excluding tropevariations) or it is impossible toabstract these tropes for effective resemblance comparison with other tropes. Trope theorycould solve Fetzer and Hull problem of relevance by making tropes the topic instead of anevent with infinite properties without resorting either to van Fraassen’s contextual accountof explanation or to an ontology of natural properties.

Modal realism (Lewis) could translate (U5) as events without this-worldly duplicates (aset of one) whose natural properties have no this-worldly duplicates (are the sole membersof their sets) or it is impossible to know that there are such duplicates or it is impossible toabstract these properties for effective comparison with other properties, so it is impossible

78 AVIEZER TUCKER

to know whether there is more than one member of their set. I pass this interpretationbecause of considerations of metaphysical economy, because Van Fraassen’s account fitsbetter actual scientific practice and because in some cases historians, for example, seem toask for explanations of extrinsic properties of events, their relations with other events.8 A version ofinference to the best explanationthat does not stipulate theoretic supportfor evaluating competing explanations may challenge the philosophic significance of (U5).Martin (1989) argued that the best historiographic explanations do not satisfy universalindependent criteria, but expose the flaws in competing explanations. Martin did not dif-ferentiate questions from answers and did not note that most historiographic arguments arenot againstcompetingexplanations (alternative answers to identical why-questions) butagainstdifferentquestions, research programs, and core beliefs (Tucker 1993). Martin (32,42) noted that meanings of the explananda he examined are different, butsociologicallysince historians ignored the variations of meaning, so did he. Martin did not provephilo-sophicallythat it is possible to choose amongcompetingexplanations without resorting totheoretical backgrounds. The existence of unique events and the commensurable method-ological specificities of historiography and psychoanalysis prove that Martin’s project isnot just empirically inadequate, but founded on an unrealistic concept of explanation.9 Why-questions appear in a context of inquiry that makes certain assumptions that canbe scientific, pragmatic, metaphysical, or superstitious. The context of inquiry influencesthe selection of topics, contrast classes, and criteria of relevance. Contexts can changeand their assumptions refuted, especially during periods of transition from a pre-scientificperspective to a scientific one. This change is connected with long term cultural changesthat e.g. made people in Halley’s social milieu forsake previous assumptions about thenature of comets and be receptive to new Newtonian explanations (such changes are notuniversal: my late grandmother told me before the last appearance of Halley’s comet that inher cultural milieu the First World War was anticipated following the previous appearanceof the comet).

REFERENCES

Brittan Jr., Gordon G.: 1970, ‘Explanation and Reduction’,The Journal of Philosophy67,446–457.

Campbell, K.: 1997, ‘The Metaphysics of Abstract Particulars’, in D. H. Mellor and A.Oliver (eds.),Properties, Oxford University Press, Oxford, pp. 125–139.

Churchland, P. M.: 1985, ‘Reduction, Qualia, and the Direct Introspection of Brain States’,Journal of Philosophy82, 8–28.

Churchland P. M. and Churchland P. S.: 1995, ‘Intertheoretic Reduction: A Neuroscien-tist’s Field Plan’, in J. Cornwell (ed.),Nature’s Imagination: The Frontiers of ScientificVision, Oxford University Press, Oxford, pp. 64–77.

Davidson, D.: 1980,Essays on Actions and Events, Oxford University Press, Oxford.Dray, W.: 1957,Laws and Explanation in History, Oxford University Press, Oxford.Fales, E.: 1980, ‘Uniqueness and Historical Laws’,Philosophy of Science47, 260–276.Fetzer, J. H.: 1975, ‘On the Historical Explanation of Unique Events’,Theory and Decision

6, 87–97.Feyerabend, P.: 1962, ‘Explanation, Reduction, and Empiricism’,Minnesota Studies in the

Philosophy of ScienceIII , 28–97.Glymour, C.: 1980,Theory and Evidence, Princeton University Press, Princeton.

UNIQUE EVENTS: THE UNDERDETERMINATION OF EXPLANATION 79

Hawking, S. W.: 1988,A Brief History of Time: from the Big Bang to Black Holes, BantamBooks, New York.

Hempel, C. G.: 1965,Aspects of Scientific Explanation, The Free Press, New York.Hoyningen-Huene, P.: 1989, ‘Epistemological Reductionism in Biology: Intuitions, Expli-

cations, and Objections’, in P. Hoyningen-Huene and F. M. Wuketits (eds.),Reduction-ism and System Theory in the Life Sciences: Some Problems and Perspectives, Kluwer,Dordrecht, pp. 29–44.

Hull, D.: 1975, ‘Central Subjects and Historical Narratives’,History and Theory14, 253–274.

Hull, D.: 1984, ‘Informal Aspects of Theory Reduction’, in I. Sober (ed.),ConceptualIssues in Evolutionary Biology, MIT Press, Cambridge MA, pp. 462–476.

Jackson, F. & Pettit, P.: 1992, ‘Structural Explanation in Social Theory’, in D. Charles andK. Lennon (eds.),Reduction Explanation and Realism, The Clarendon Press, Oxford,pp. 97–131.

Joynt, C. B. & Rescher, N.: 1961, ‘The Problem of Uniqueness in History’,History andTheory1, 150–162.

Kuipers, T. A. F.: 1990, ‘Reduction of Laws and Concepts’,Poznan Studies in thePhilosophy of the Sciences and the Humanities16, 241–276.

Lakatos, I.: 1978,The Methodology of Scientific Research Programmes, CambridgeUniversity Press, Cambridge.

Lewis, D.: 1997, ‘Modal Realism at Work: Properties’, in D. H. Mellor and A. Oliver(eds.),Properties, Oxford University Press, Oxford, pp. 173–187.

Looijen, R.: 1987, ‘Emergence and Reduction in Biology’,LMPS-VIII Abstracts2,Moscow, pp. 265–268.

Martin, R.: 1989,The Past Within Us: An Empirical Approach to Philosophy of History,Princeton University Press, Princeton.

Marwick, A.: 1989,The Nature of History, 3rd edition, MacMillan, London.Murphey, M.: 1994,Philosophical Foundations of Historical Knowledge, State University

of New York Press, Albany NY.Nagel, E.: 1961,The Structure of Science, Harcourt, Brace & World, New York.Nickels, T.: 1973, ‘Two Concepts of Inter-theoretic Reduction’,The Journal of Philosophy

70, 181–201.Quine W. V.: 1975, ‘On Empirically Equivalent Systems of the World’,Erkenntnis9, 313–

328.Rappard, H. V.: 1993, ‘History and the System’,Annals of Theoretical Psychology9, 1–40.Ruse, M.: 1984, ‘Reduction in Genetics’, in I. Sober (ed.),Conceptual Issues in Evolution-

ary Biology, MIT Press, Cambridge MA, pp. 446–461.Ruse, M.: 1989, ‘Sociobiology and Reduction’ in P. Hoyningen-Huene and F. M. Wuketits

(eds.), Reductionism and System Theory in the Life Sciences: Some Problems andPerspectives, Kluwer, Dordrecht, pp. 45–83.

Schaffner, K.: 1984, ‘Reduction in Biology: Prospects and Problems’, In I. Sober (ed.),Conceptual Issues in Evolutionary Biology, MIT Press, Cambridge MA, pp. 428–445.

Schwartz, J.: 1991, ‘Reduction, Elimination, and the Mental’,Philosophy of Science58,203–220.

Shoemaker, S.: 1997, ‘Causality and Properties’, in D. H. Mellor and A. Oliver (eds.),Properties, Oxford Univesity Press, Oxford, pp. 228–254.

Sklar, L.: 1967, ‘Types of Inter-theoretic Reduction’,British Journal for the Philosophy ofScience18, 106–124.

80 AVIEZER TUCKER

Spence, D. P.: 1994,The Rhetorical Voice of Psychoanalysis: Displacement of Evidenceby Theory, Harvard University Press, Cambridge MA.

Thagard, P.: 1988,Computational Philosophy of Science, MIT Press, Cambridge MA.Tucker, A.: 1993, ‘A Theory of Historiography as a Pre-Science’,Studies in History and

Philosophy of Science24, 633–667.Tucker, A.: 1995, ‘The Illness of Psychoanalysis’,Studies in History and Philosophy of

Science26, 657–665.Tucker, A.: 1997a, ‘Contemporary Philosophy of Historiography’,Philosophy of Social

Sciences27, 100–127.Tucker, A.: 1997b, ‘The Pragmatics of Historiographic Explanation’, in G. Schurtz et al.

(eds.),The Role of Pragmatics in Contemporary Philosophy: (Papers of the 20th Interna-tional Wittgenstein Symposium), The Austrian Ludwig Wittgenstein Society, Kirchbergam Wechsel, pp. 928–933.

Tucker, A.: 1998, ‘Scientific Historiography Reconsidered: An Essay on the Meta-physics and Epistemology of History’,Dialogue: Canadian Philosophical Quarterly(forthcoming).

Van Fraassen, B. C.: 1980,The Scientific Image, Clarendon Press, Oxford.Van Inwagen, P.: 1993,Metaphysics, Westview Press, Boulder CO.Wimsatt, W.: 1984, ‘Reductive Explanation: A Functional Account’, in I. Sober (ed.),

Conceptual Issues in Evolutionary Biology, MIT Press, Cambridge MA, pp. 478–508.Windelband, W.: 1980, ‘Rectorial Address, Strasbourg, 1894’, trans. Guy Oakes,His-

tory and Theory19, 169-185. [translation of: ‘Geschichte und Naturwissenschaften’,in PräludienVol II, Verlag von J. G. B. Mohr, Tübingen, pp. 136–160].

Manuscript submitted September 23, 1996Final version received October 3, 1997

Faculty of PhilosophyPalacky UniversityKrizkovskeho 12Olomouc 77180Czech RepublicE-mail: [email protected]