Max Black:
Wie stellen Bilder dar?Alle Rechte im Suhrkamp-Verlag |
|
97 HOW DO PICTURES REPRESENT? Alle
Rechte |
|
Einige Fragen. An der Wand hängt ein Bild: Es zeigt ein Rennpferd mit Bäumen im Hintergrund, die Buchen sein könnten, und einen Stalljungen, der im Vordergrund mit einem Eimer hantiert. Daß das Bild alle diese Dinge zeigt und daß alle diese Dinge und einiges mehr auf dem Bild sichtbar sind, steht außer Zweifel. Was aber macht dieses Gemälde zum Bild von einem Rennpferd, von Bäumen und einem Menschen? Noch allgemeiner gefragt: Was macht ein »naturalistisches« Gemälde oder eine Fotografie zu einer Darstellung des jeweiligen Sujets? Und wie — wenn überhaupt —verändert sich die Situation, wenn wir zu so »konventionellen« Darstellungen wie Karten, Diagrammen oder Modellen übergehen? In welchem Maße tragen »Konvention« oder »Interpretation« zur Konstituierung einer Beziehung zwischen einer Darstellung und ihrem Sujet bei?
die einem ganz natürlich vorkommen und die ohne ermüdende Paraphrasierung nicht vermieden werden können.
Schon Teilantworten haben Konsequenzen für so verschiedene Gegenstandsbereiche wie Wahrnehmung und Erkenntnis, für die Struktur von Symbolsystemen, die Beziehungen zwischen Denken und Empfinden sowie für die Ästhetik der bildenden Künste. Allein schon die Einsicht in die Gründe, weshalb Fragen zur Darstellung eine solch außergewöhnliche Uneinigkeit hervorrufen, wäre lohnend genug für etwas, das zu einer schwierigen, wenn nicht gar ermüdenden Untersuchung zu werden droht. Einige Gebrauchsdefinitionen. Wenn ein Gemälde— oder eine andere visuelle Darstellung, z. B. eine Fotografie — ein Gemälde von etwas, S, ist, werde ich sagen: P bildet S ab; oder anders ausgedrückt: P steht in einer Abbildungsbeziehung. zu S. An dieser Stelle könnten sich jedoch Mißverständnisse ergeben.
Nehmen wir als Gegenbeispiel ein anderes Gemälde, das Hitler bei der Überquerung des Hudson im Jahre 1950 zeigt: Hier gibt es kein tatsächliches Ereignis, das zur Überprüfung der Treue des Gemäldes dienen könnte; gleichwohl wollen wir auch hier davon sprechen, daß das Gemälde ein »Sujet» habe, das von ihm abgebildet wird.
Das gezeigte Sujet könnte als Inhalt der visuellen Darstellung aufgefaßt werden. (Anmerkungen 1) Ich werde mich insbesondere mit der Abbildung im speziellen Sinne des »Zeigens« befassen und dementsprechend mit jenem »Sujet« eines Gemäldes, das sein »Inhalt« und nicht seine , »Originalszene« ist. Wo es der Kontext erlaubt und keine anderen Hinweise gegeben werden, kann der Leser deshalb von nun an »abbilden« und »zeigen« als synonym betrachten. Es sollte daran erinnert werden, daß ich mich nicht von Anfang an darauf festgelegt habe, daß zwischen »Sujet« und «Originalszene« immer eine irreduzible Differenz besteht. Noch wichtiger ist, daß ich mich nicht— wie die Rede über die Beziehung zwischen P und S irrtümlicherweise andeuten könnte — auf die Existenz des «Sujets« als unabhängige und selbständige Entität festgelegt habe. Mit allem bisher Gesagten ist verträglich, daß S-zeigen sich als ein einheitliches Prädikat erweisen könnte, das keine Implikationen hinsichtlich der Existenz oder Nichtexistenz von S mit sich führt. Das Vorkommen von S in »S abbilden«.könnte somit »intensional« und nicht »extensional« sein. Auf der Suche nach Kriterien. Ein ehrgeiziger Leser unserer Frageliste erwartet möglicherweise eine analytische Definition von zeigen oder S-zeigen, d. h. irgendeine Formel von der Struktur: P zeigt S dann und nur dann, wenn R, wobei >R< durch einen ausführlicheren und erhellenderen Ausdruck (was wir darunter auch immer verstehen wollen) als das undifferenzierte Wort »zeigt« zu ersetzen ist. R stellt somit die notwendige und hinreichende Bedingung für P dar, S zu »zeigen«. ++++++++ Es ist wohl unrealistisch, zu erwarten, wir könnten eine Reihe von notwendigen und hinreichenden Bedingungen finden, die diesem Muster entsprechen. Wir brauchen uns jedoch auf dieses Ziel nicht festzulegen, denn es wäre bereits erhellend genug, wenn wir einige notwendige Bedingungen herausarbeiten könnten. Ein noch bescheideneres, aber immer noch schwer zu verwirklichendes Vorhaben bestünde darin, einige Kriterien für die Anwendung von Ausdrücken der Form >zeigt S< zu ermitteln, d. h. Bedingungen, die ausschließlich kraft der relevanten Bedingungen von >zeigt< für oder gegen ihre Anwendung in bestimmten Fällen sprechen. Solche Kriterien brauchen weder unveränderlich noch universell relevant zu sein für all jene Fälle, bei denen es sinnvoll ist zu sagen, etwas zeige etwas anderes oder stelle es zur Schau: Die Anwendungskriterien könnten sich von Fall zu Fall in einer systematischen und beschreibbaren Weise verändern. Das Ziel bestünde dann in einer zwar nur partiellen, dafür aber klaren Einsicht in das Muster für die Verwendungen von »abbilden« sowie den dazugehörigen Ableitungen, nicht in einer formalen Definition. (Anmerkungen 2) Obwohl dies das Programm ist, werde ich mir gleichwohl zunächst den Plan zu eigen machen, der Reihe nach eine Anzahl von wahrscheinlich notwendigen Bedingungen für »zeigen« zu untersuchen. Erst wenn wir zur Überzeugung gelangt sind, daß keiner dieser Kandidaten für sich — und auch nicht alle zusammengenommen — als Lösung unserer Fragestellung dienen kann, werde ich mit der Argumentation für einen flexibleren Antworttypus fortfahren. Jemand, der von Wittgensteins analogen Untersuchungen
grundlegender Begriffe beeinflußt ist, könnte selbst unser derart
eingeschränktes Unternehmen als sträflich leichtsinnig einschätzen, weil er
vorauszusehen glaubt, daß sich das »Gebrauchsmuster« als zu verwickelt
erweisen wird, um auf irgendeine Formel reduziert zu werden. Diesem
Pessimismus könnte man den Hinweis auf das »kreative« Vermögen der Sprache
entgegenhalten, das ebenso wichtig wie offenkundig ist und kraft dessen wir
verstehen können, was mit einer Aussage von der Form »P ist ein Gemälde von
S« gemeint ist, auch wenn sie auf ein Gemälde eines neuen, ungewohnten und
schwer verständlichen Stils gemünzt wird. Daß wir zu verstehen glauben, was
in solchen Kontexten gesagt wird, ist ein starkes Argument für die Existenz
eines zugrunde liegenden Anwendungsmusters, das nur offengelegt zu werden
braucht. Und selbst wenn dies eine Täuschung sein sollte, bedürfte sie der
Erklärung.
Ein grundsätzlicher Einwand. Die Suche nach analytischen An-wendungskriterien für >zeigen< erinnert an die vielen mißlungenen Versuche, partielle oder umfassende Analysen der Wortbedeutung zu liefern. Gewiß liegt hier möglicherweise mehr als nur eine Analogie vor. Von einem verbalen Text können wir— unter Wiederholung einiger früherer Formulierungen — mit Recht sagen, daß er eine Beschreibung oder Darstellung von einer Szene, einer Situation oder einem Sachverhalt ist, der via »Inhalt« des Texts präsentiert wird und der möglicherweise einer verifizierenden Tatsache (dem Analogon zu unserer ,,Originalszene«) entspricht. Es ergeben sich somit Fragen, die den von mir hervorgehobenen Fragen zur bildlichen Darstellung zu entsprechen scheinen. Einige Autoren versuchen tatsächlich, die beiden Fragetypen einander gleichzusetzen und beziehen dabei ihre explanatorischen und analytischen Begriffe aus dem Bereich der Wortsemantik. Gibt es nun gute Gründe für die Annahme, daß die Suche nach einer Begriffslandkarte der Wortbedeutung zwangsläufig mißlingen müsse? Der verstorbene John Austin schien diese Ansicht zu vertreten (vgl. Austin, 1961), aus Gründen, die offenbar in Publikationen nicht diskutiert wurden. Austin stellt die Suche nach der Bedeutung eines einzelnen Worts oder Ausdrucks dem gegenüber, was er als illegitime Suche nach der Bedeutung im allgemeinen auffaßt. Er erinnert uns daran, daß die erste Art von Fragen dann beantwortet ist, wenn wir »die Syntax« des betreffenden Wortes »erklären« und dessen »Semantik aufzeigen» können, d. h. wenn wir die grammatischen Bedingungen für den Ausdruck und seine ostensiven und quasi-ostensiven Verbindungen mit nichtverbalen Gegenständen und Situationen (wo dies angemessen ist) angeben können. Sodann wendet Austin freilich ein, daß die vermeintlich allgemeinere Frage »Was ist die Bedeutung eines Worts-im-allgemeinen?« eine Scheinfrage ist. »Ich kann eine Frage von der Form >Was ist die Bedeutung von >X< nur dann beantworten, wenn "x" ein bestimmtes Wort ist, nach dem ich gefragt werde. Die vermeintlich allgemeine Frage ist in Wirklichkeit nur eine Scheinfrage eines Typs, wie er gemeinhin in der Philosophie vorkommt. Wir können ihn den Fehlschluß des Fragens nach »nichts bestimmtem« nennen, eine Praxis die der einfache Mann beklagt, der Philosoph hingegen als >Verallgemeinerung< bezeichnet und mit einiger Wohlgefälligkeit betrachtet«. Wenn Austin recht hätte, wäre unsere Hauptfrage nach der Bedeutung von »P ist ein Gemälde von S« einem ähnlichen Sinnlosigkeitsverdacht ausgesetzt, da wir sie nicht für ein bestimmtes Gemälde stellen, sondern für Gemälde im allgemeinen. Soweit ich dem Gedanken Austins zu folgen vermag, scheint er zu besagen, daß die zurückgewiesene allgemeine Frage nach der Bedeutung eines Worts im allgemeinen leicht als Suche nach einer einzelnen Bedeutung aufgefaßt werden kann, die allen Wörtern gemeinsam ist. Und wenn er recht hätte wäre unser Problem ein ähnlich lächerlicher Versuch, ein einziges, allen Gemälden gemeinsames Sujet zu entdecken. So verstanden wäre die Fragestellung tatsächlich eine Scheinfrage, wenn nicht gar eine alberne Verwirrung. Interessant ist aber die Feststellung, daß Austin unmittelbar nach der Zurückweisung der »allgemeinen« Frage nach der Bedeutung die allgemeine Frage »Was ist die >Quadratwurzel< einer Zahl?« —und zwar einer beliebigen Zahl und nicht einer bestimmten — als gerechtfertigt zugibt. Wenn dies nun eine gerechtfertigte Art der Suche nach einer Definition für >Quadratwurzel< ist, weshalb sollten wir dann die »allgemeine Frage» nach der Bedeutung nicht auch mit gleichem Recht als legitime Suche nach einer Definition — oder zumindest dem, was für die Definition relevant sein kann — von >Bedeutung< betrachten?
Dieses Zugeständnis braucht jedoch nicht zu beinhalten, daß jede Suche nach Kriterien als scheinhaft abzulehnen ist. Merkwürdig genug ist, daß Austin nach dem für ihn charakteristisch vehementen Angriff auf das »allgemeine« Unternehmen einer Abgrenzung des Bedeutungsbegriffs damit fortfährt, der gleichen Aufgabe einen annehmbaren Sinn zuzuweisen, indem er sie als Versuch zur Beantwortung der Frage »Was-ist die-Bedeutung-von (dem Satz) >die-Bedeutung-von (dem Wort) x<?« auffaßt. Die entsprechende Frage für uns könnte lauten: »Was ist die Bedeutung von (dem Satz) >S-zeigen<?« Wir können dann im Anschluß an die Empfehlung Austins fortfahren, die »Syntax und Semantik« jenes Ausdrucks zu untersuchen, ohne uns auf die Existenz zweifelhafter Entitäten festzulegen und ohne Befürchtungen über die vermeintliche Sinnlosigkeit des Unternehmens zu hegen. Mag sein, daß unser Ziel unerreichbar bleibt; aber auch das muß sich erst zeigen. Wie bildet eine Fotografie ab? Ich werde nun kurz die Form betrachten, die unsere Frage nach der Natur der Abbildung dann annimmt, wenn sie auf den Spezialfall der Fotografie angewandt wird. Denn wenn Bilder überhaupt in einer »natürlichen« Beziehung zu ihren gezeigten und wiedergegebenen Sujets stehen, dann müßten unretuschierte Fotografien vorzügliche Beispiele dafür abgeben. An ihnen dürften wir gewiß die geringsten Schwierigkeiten haben, die Komplexitäten herauszufinden, welche dem Begriff der wirklichkeitstreuen >Wiedergabe<‘ zugrunde liegen. Da Fotografien überdies den »Ausdrucks"-Intentionen ihrer Hersteller nur einen minimalen Spielraum belassen, können wir Überlegungen, die mit den Ausdrucksaspekten der bildenden Kunst zusammenhängen (und deren entscheidende Bedeutung in anderen Kontexten natürlich unbestreitbar ist), ausklammern.
Einige Philosophen könnten erwidern. daß wir bei der Verletzung kausaler Gesetze auf diese oder andere Weisen »gar nicht wissen könnten, was wir sagen sollten«. Das scheint jedoch nur ein bequemer Ausweg aus einer begrifflichen Schwierigkeit zu sein. Vielleicht würde uns ein einziges derart merkwürdiges Beispiel, wie das von mir genannte, in hoffnungsloser Verwirrung zurücklassen. Wenn das Phänomen aber durch ein Standardverfahren ordnungsgemäß wiederholbar wäre, dann, so nehme ich an, könnten wir mit Recht sagen, wir hätten eine neue, wenn auch verwirrende Art der Herstellung von Darstellungen oder »Abbildungen« der Abtei und anderer Gegenstände entdeckt. Es wäre leicht, jede beliebige Anzahl weiterer Gegenbeispiele auszudenken, bei denen aus radikal unorthodoxen Verfahren Endprodukte hervorgehen würden, die von herkömmlichen Fotografien nicht zu unterscheiden waren.
124 Sie erfordert eine getrennte Diskussion. Zunächst wollen wir jedoch die Vermutung untersuchen, der
Begriff der »Information« liefere den Schlüssel, den wir suchen. Die Vorliebe für das Reden über die in Darstellungen enthaltene »Information« — und auch für die Verwendung des Begriffs in fast allen denkbaren Diskussionen — ist sicherlich geprägt durch die vermeintlichen Erfolge des Begriffs der »Information«, der gewöhnlich mit dem Namen Shannon in Verbindung gebracht wird, in verfeinerten mathematischen Theorien, (Anmerkung 7) Es kann jedoch gezeigt werden, daß die beiden hier in Frage kommenden Bedeutungen von , »Information« kaum etwas miteinander zu tun haben.
In einer groben, mit einem derartigen Kommunikationssystem zusammenhängenden Erklärung des Begriffs der »selektiven Information« würde dieser mit dem Grad der »Reduktion anfänglicher Ungewißheit« gleichgesetzt, den ein solches System erzielen kann. Nehmen wir an, daß die verschiedenen möglichen Botschaften mi bekanntermaßen mit langfristiger Häufigkeit oder Wahrscheinlichkeit pi auftreten. Wir könnten dann sagen, daß die durch den Empfang einer einzelnen Botschaft mi übermittelte Information umgekehrt zu ihrer ursprünglichen Wahrscheinlichkeit des Auftretens, pi, variiert. Je höher die anfängliche Wahrscheinlichkeit der Übertragung ist, desto »weniger lernen wir« durch den Empfang der Botschaft. Wenn die fragliche Information im Grenzfall mit absoluter Sicherheit eintreffen sollte, dann würden wir durch ihren Empfang »nichts lernen«. Die mathematische Quantität, (selektive) Information genannt, ist das Maß für den Betrag einer bestimmten Größe — grob gesprochen für die Reduktion im Betrag anfänglicher Ungewißheit über den Empfang, wie ich oben angedeutet habe. Es muß betont werden, daß dies mit der Bedeutung einer solchen Botschaft — sofern sie überhaupt eine hat —nichts zu tun hat, ebensowenig wie mit ihrem besonderen Inhalt. Fordere ich telegrafisch Antwort auf eine Frage und sind die beiden einzigen möglichen Antworten entweder Ja oder Nein und werden beide mit gleicher Anfangswahrscheinlichkeit gesendet, dann enthält jede der beiden Antworten die gleiche »(selektive) Information«. Jede Antwort verwandelt die Wahrscheinlichkeit 1/2 in Gewißheit. Ein besorgter Freier, der eine Antwort auf seinen Heiratsantrag erwartet, würde freilich sagen, die Information der einen Antwort unterscheide sich erheblich von der Information der anderen. Das ist jedoch darauf zurückzuführen, daß er »Information« im alltäglichen oder gewöhnlichen Sinne dessen gebraucht, was man als substantielle Information bezeichnen könnte. Die mathematischen Theoretiker interessieren sich nicht für die substantielle Information — was ihr gutes Recht ist und keineswegs ein Vorwurf sein soll. Hier anders zu denken wäre ebenso verfehlt, wie wenn man einer Meßtheorie den Vorwurf machte, sie sage nichts über den Geruch oder den Geschmack der von ihr erfaßten Meßgegenstände aus.
126
Hier scheint genau das vorzuliegen, was wir suchen: Das »gezeigte Sujet« einer Fotografie scheint dem nahezukommen, was der gewöhnliche Verstand als Information zu bezeichnen pflegt, die ein dafür hinreichend kompetenter Empfänger (Betrachter) verstehen könnte. (Anmerkung 11). Wenn wir jedoch den von Hintikka und anderen Pionieren der semantischen Informationstheorie bereitgestellten Konstruktionen folgen, entdecken wir zu unserer Enttäuschung, daß auch sie etwas liefern, das uns ungeachtet des sonstigen Werts bei unserer gegenwärtigen Untersuchung nicht weiterhilft. Es zeigt sich nämlich, daß die »semantische Information« einer bestimmten Aussage etwa das gleiche ist wie das Spektrum von Verifikationssituationen, die mit dieser Aussage verbunden sind —genauer gesagt, ein Maß für die »Breite» dieses Spektrums. Und was damit schließlich bereitgestellt wird, ist ein Maß für die Ausdehnung und nicht für den Inhalt. »Semantische Information» ist eine differenzierte Verfeinerung des gemeinverständlichen Begriffs für den Betrag von Information in einer Aussage. Und ebensowenig wie die Angabe der Maße eines Körpers uns etwas über die stoffliche Zusammensetzung dieses Körpers sagt, gibt uns die Angabe der semantischen Information Aufschluß, worüber die fragliche Aussage geht. Wenn wir zwei Aussagen von gleichartiger logischer Struktur haben, z. B. »Mein Name ist Schwarz« und , »Mein Name ist Weiß« (Anmerkung 12), wird jede akzeptable Definition der semantischen Information beiden Aussagen jeweils die gleiche semantische Information zuschreiben. Wenn dieser Begriff (dessen Bedeutung ich nicht bestreiten will) auf Gemälde anwendbar wäre (anmerkung 13), müßten wir unterschiedliche Gemälde mit oberflächlich vergleichbaren Sujets (z. B. zwei Gemälde von einer Herde grasender Schafe) so auffassen, als ob sie die »gleiche Information« besäßen und übermittelten Natürlich können sich die gezeigten Sujets zweier solcher Gemälde deutlich unterscheiden.
Angenommen, wir haben den Begriff der in einerAussage enthaltenen substantiellen Information ausreichend unter Kontrolle, so könnten wir erwägen, die bestimmte Fotografie P durch eine komplexe Aussage A zu ersetzen, so daß ein kompetenter Empfänger aus A ebensoviel lernen könnte wie aus P, wenn P eine wirklichkeitstreue Aufzeichnung der entsprechenden Originalszene sein sollte. Sicherlich haftet an dieser Vermutung etwas Unrealistisches. Nehmen wir an, jemand würde mit einer solchen Aussage A konfrontiert und anschließend aufgefordert, aus einer großen Menge verschiedener Fotografien die Fotografie P herauszufinden, von der A in gewissem Sinne eine Übersetzung sein soll. Gibt es einen guten Grund zu der Annahme, daß eine solche Aufgabe im Prinzip lösbar sein muß? Mir erscheint im Gegenteil die Vorstellung einer vollständigen verbalen Übersetzung einer Fotografie (und viel stärker noch die Vorstellung der verbalen Übersetzung eines Gemäldes) als ein Hirngespinst. Ein Bild zeigt mehr, als gesagt werden kann — und zwar nicht einfach deshalb, weil im Wörterbuch entsprechende Äquivalente fehlen: Es geht nicht nur um die Nichtverfügbarkeit verbaler Bezeichnungen für die Tausende von Farben und Formen, die wir unterscheiden können. Wenn dies so ist, dann ist der Begriff der Information, der seinen Ort im Zusammenhang mit verbalen Darstellungen (Aussagen) hat, dem uns interessierenden Fall immer noch unangemessen. (Anmerkung 14) Am Ende scheint es, daß das, was mit der Figur der in einer Fotografie oder in einem Gemälde enthaltenen »Information" bildhaft ausgedrückt wird, auf nichts anderes hinausläuft als auf das, was wir meinen, wenn wir über den »Inhalt« des Gemäldes sprechen oder über das, »was« es zeigt« (sein zur Schau gestelltes Sujet). Gegen die Einführung einer auf dem Informationsbegriff beruhenden Metapher oder Analogie gäbe es nichts einzuwenden, wenn sie eine Klärung bringen würde. Mir scheint, daß dies nicht der Fall ist und daß die Bezugnahme auf die »Information«, auf der Grundlage einer mehr oder weniger einsichtigen Analogie, schließlich auf die Einführung eines — überdies irreführenden — Synonyms für »Abbilden« oder »Darstellen« hinausläuft. Es ist nicht unbillig, darauf hinzuweisen, daß »die durch ein Gemälde übermittelte Information« nichts anderes bedeutet, als , » was durch dieses Gemälde gezeigt (abgebildet, zur Schau gestellt) wird«. Aus diesem irreführenden Umweg können wir gleichwohl eine nützliche Lehre ziehen. Eine von Theoretikern der statistischen Information gleichermaßen betonte Vorsichtsmaßregel besagt, daß die Messungen der von ihnen behandelten Information immer relativ sind zu einer Anzahl unterscheidbarer Faktoren in den relevanten Situationen. Im Falle der statistischen Information ist der Betrag an Information relativ zur Verteilung langfristiger Häufigkeiten des im betreffenden Kommunikationskanal übertragbaren Systems möglicher Botschaften; im Falle der semantischen Information ist der in einer Aussage verkörperte Betrag an Information relativ zur Wahl einer Sprache und in bestimmten Anwendungsfällen relativ zu Annahmen über bestimmte Gesetze, die einen vorgängigen Bestand an vorhandener Information konstituieren, wobei jede nicht aus diesen Gesetzen ableitbare Aussage einen Zusatz zu diesem Bestand darstellt. Wir könnten somit zu einer ziemlich einsichtigen Lehre ermutigt werden: Ungeachtet dessen, wie wir dazu kommen, den substantiellen Gehalt eines Gemäldes oder einer anderen visuellen Darstellung zu identifizieren oder zu beschreiben, wird die Antwort relativ zu einem vorausgesetzten Wissensbestand sein (z. B. Wissen über das gewählte Darstellungsschema, die Intentionen des Malers oder Zeichenproduzenten usw.). Die Vorstellung, daß ein Gemälde oder eine Fotografie ihren Inhalt oder ihr Sujet ebenso direkt »enthält«, wie ein Eimer Wasser enthält, ist allzu grobschlächtig, als daß sie einer Widerlegung bedürfte. Derartige Vorstellungen haben jedoch einen beträchtlichen Teil der Diskussion über unser Thema beherrscht. Die Bezugnahme auf die Intentionen des Herstellers. Ich werde mich nun mit dem Hinweis befassen, daß ein Ausweg aus unseren Schwierigkeiten in der Bezugnahme auf die Intentionen des Malers oder des Fotografen gefunden werden könnte — oder wer es auch immer war, der so handelte, daß er die visuelle Darstellung hervorbrachte, deren »Sujet« wir hier erörtern. Ich kenne keinen Theoretiker, der auf dieser Grundlage eine umfassende Theorie der Darstellung entwickelt hätte; entsprechende Theorien der verbalen Darstellung sind hingegen ziemlich häufig. So hat Grice in einem bekannten Aufsatz über »Bedeutung« (Anmerkung 15) betont, daß die Bedeutung einer Äußerung mittels komplexer Intentionen analysiert werden könnte, beim Hörer einen bestimmten Effekt zu bewirken.(Anmerkung 16) Ferner hat E. D. Hirsch in einem Buch Wortbedeutungen
definiert als »alles, was jemand durch eine bestimmte Abfolge sprachlicher
Zeichen übermitteln will und was mittels dieser Zeichen übermittelt
(mitgeteilt) werden« kann.(Anmerkung 17) Es scheint prinzipiell keine
vernünftige Begründung zu geben, weshalb dieser Ansatz nicht ebenso gültig
sein sollte im Zusammenhang mit der visuellen Darstellung oder mit der
Darstellung überhaupt. Die unbestreitbare Anziehungskraft dieser Hervorhebung der
Intention oder des »Willens« des Herstellers kann darauf zurückgeführt
werden, daß sie nachdrücklich auf die begriffliche Kluft aufmerksam macht,
die zwischen der »Interpretation« eines natürlichen Gegenstandes (etwa wenn
wir aus den Merkmalen einer Spur auf die Eigenschaften eines Etwas schließen,
das diese Spur hervorgebracht hat) und der »Interpretation« eines von
Menschen hergestellten Gegenstandes besteht, der intentional geschaffen
wurde, um Bedeutung oder »Inhalt« einer bestimmten Art zu haben, die einem
kompetenten Empfänger zugänglich ist. Es ist eine Sache, in der Terminologie
von Grice dem zuzustimmen, daß die Bedeutung eines Gegenstands »nicht-natürlich«
ist und nicht auf die Aus- Zunächst besteht der unmittelbare Einwand, daß die Intention des Herstellers, von der angenommen wird, daß sie in einer unstrittigen Weise vorhanden war, fehlschlagen kann. Nehmen wir einmal an, ich wolle ein Pferd zeichnen und bringe mangels Fertigkeit etwas hervor, das bei einfachem Hinsehen niemand von einer Kuh unterscheiden kann: Wäre dies notwendigerweise die Zeichnung eines Pferdes, nur weil ich dies intendiert habe? Könnte ich ein Pferd zeichnen, indem ich einfach einen Punkt aufs Papier setze? Wären die Antworten darauf positiv, dann müßten wir die Intentionen des Künstlers so betrachten, als ob sie den eigentümlichen Charakter der Unfehlbarkeit besäßen: Das einfache Wollen, ein Gemälde sei ein Gemälde von einem So-und-so, würde es dann notwendigerweise zu einem solchen machen. Sicherlich wäre dies zu widerspruchsvoll, um akzeptabel zu sein. Bei einer verpfuschten und unerkennbaren Zeichnung müßten wir sagen können: »Er beabsichtigte, ein Pferd zu zeichnen, aber es mißlang ihm», wie wir es unter bestimmten Umständen von jeder mißlungenen Intention sagen müßten. Die Vorstellung der Intention schließt die Vorstellung
eines möglichen Fehlschlagens ein. 131 Der Sachverhalt wäre weniger anstößig, wenn die vorgeschlagene Analyse die folgende Form annehmen würde: »P bildet S dann und nur dann ab, wenn M, der Hersteller von P, E intendierte», wobei E als ersetzbar gedacht wird durch einen komplexen Ausdruck (und nicht durch ein direktes Synonym für >P bildet S ab<).‘(Anmerkung 19) Die oben festgestellte Zirkularität würde dann wegfallen. Damit diese Analyse jedoch akzeptabel wird, müßte £ die gleiche Extension haben wie >P bildet S ab<: wir haben die richtige Intention nur dann erfaßt, wenn das, was M zu tun intendierte, für P notwendig und hinreichend war, damit dieses eine Abbildung von S ist (wenn auch nicht in diesen Worten ausgedrückt). Und wenn dies so ist, können wir die Bezugnahme auf die Intentionen von M insgesamt fallenlas-sen, weil die Formulierung »P bildet S dann und nur dann ab, wenn E» als solche bereits die gesuchte Analyse ergibt. Diese Betrachtungsweise
hätte außerdem den Vorteil, daß sie mit der oben fest-gestellten
Schwierigkeit der fehlgeschlagenen Intention fertig wird. 20
Ein Laie wird angesichts eines Gemäldes von Claudio Bravo sicherlich berichten, daß es für jedermann so aussieht, als ob sich hinter der Oberfläche ein Paket befinde, und jeder Betrachter, sei er noch so sehr mit theoretischen Überzeugungen gegen die Rolle der Illusion gewappnet, kann, wenn er einsichtig ist, nicht umhin, bei anderen Beispielen des trompe l‘oeil eine ähnliche Beschreibung zu geben. Es besteht natürlich die ernsthafte Frage, ob eine Beschreibung, die auf diesen besonderen Typus zugeschnitten zu sein scheint, ohne Verzerrung oder Tautologie so ausgedehnt werden kann, daß sie auf alle Fälle der Reaktion auf visuelle Darstellungen paßt, die partiell naturalistisch sind. Ich glaube jedoch, daß es nicht besonders schwierig ist,
die These so zu erweitern, daß sie auch auf Fälle zutrifft, bei denen das
dargestellte Sujet unbekannt ist. Dieser Auffassung zufolge bietet es keine
besondere Schwierigkeit, den Anblick eines vom Betrachter gesehenen
fliegenden Pferdes oder einer leibhaftig in der Luft schwebenden Göttin zu
erklären. Die Erklärung kann überdies bei bestimmten »abstrakten« Werken
unverändert beibehalten werden: Wenn ich auf einem Gemälde von Rothko eine
zurückweichende Ebene sehe, die von einem kontrastierenden Streifen begrenzt
ist, dann ist dieser Anblick nicht unähnlich dem, was ich durch das Hinsehen
auf Wolken gelernt habe. Ähnliches gilt auch für Manhattan Boogie Woogie von
Mondrian oder andere Abstraktionen dieser Art. (Anmerkung 24) Bei dieser
Auffassung reduzieren sich Probleme, wie ein P ein bestimmtes S abbilden
kann, auf Fragen zur gewöhnlichen Wahrnehmung, etwa von der Form: »Wie kommt
es, daß ein wirklicher Pudel wie ein Pudel aussehen kann?» Ich bin nicht
sicher, welcher brauchbare Sinn einer solchen Frage zugeschrieben werden kann
(Anmerkung 25) jedenfalls würde sie aus dem Bereich unserer gegenwärtigen
Untersuchung herausfallen. Wir wollen nun mögliche Einwände betrachten. Bei
denen des er-sten Typs wird eingewandt, daß die »Illusion» weder vollständig
sei noch als vollständige intendiert war. Wenn wir unsere Stellung zum Bild
verändern, erhalten wir nicht die systematischen Verän-derungen in der
Erscheinung, wie sie auftreten würden, wenn sich an der angegebenen Stelle
wirklich ein Pudel befinden würde: Eine bemalte Leinwand bringt nicht einmal
so viel Illusion hervor wie ein Spiegel. Außerdem ist die dargebotene
visuelle Erscheinung Der zweite Typ von Einwänden lenkt unsere Aufmerksamkeit auf
die wahrnehmbare Verzerrung, die auch auf dem ,,realistischsten" Bild
festgestellt werden kann: Der aufmerksame Betrachter wird mit Ausnahme ganz
besonderer Fälle die Pinselstriche sehen und schließlich doch gewahren, daß
das, was er sieht, dem wirkli-chen Ding ,,nicht sehr ähnlich» ist.27 Die entscheidenden Schwierigkeiten verbergen sich jedoch hinter der täuschenden Redewendung: «Wenn wir einmal gelernt haben, wie die Dinge aussehen», denn damit wird tatsächlich eingeräumt, daß es in vielen Fällen einen Sinn gibt, in dem das Sujet auf dem Gemälde nicht so aussieht, wie es aussehen würde, wenn es tatsächlich hinter der Leinwand vorhanden wäre — und derart große Abweichungen vom gewöhnlichen Sehen können nicht durch den Einwand erledigt werden, den wir gerade in Betracht gezogen haben. Wenn Picassos Frauen als Frauen gesehen werden sollen (d. h. so, als ob sie hinter der Leinwand Frauen wären), dann werden wir einen Interpretationsschlüssel erlernen müssen, zu dem es in der gewöhnlichen Wahrnehmung kein Analogon gibt. Wir werden beispielsweise lernen müssen, zwischen einem »wirklichkeits-treuen» Gemälde eines grünen Gesichts und einem grünen Gemälde eines weißen Gesichts zu unterscheiden. Sobald wir jedoch einen solchen logischen Vorgriff in die »Technik
der Darstellung» zulassen und zulassen müssen, verliert die Gleichsetzung von
Abbildung und fiktiver Darstellung oder »Illusion» ihre Anziehungskraft.
Statt zu sagen: ,,P ist eine Darstellung von S, weil das Sehen von P, mit
einigen Vorbehalten gegenüber Verzerrungen und Unvollständigkeit, dem Sehen
von S gleicht (P sehen ist als ob man S sähe)", müssen wir nun etwa
sagen: ,,Im allgemeinen ist P eine Darstellung von S, wenn P entsprechend den
in Stil und Technik ,35 des Künstlers verkörperten Konventionen wie S
aussieht.» Und nun wundert man sich darüber, wieviel die noch übriggebliebene
Bezugnahme auf die Formel »Aussehen, als ob S präsent wäre» wirklich leistet.
Nehmen wir einen Fall extremer Verzerrung an: Wir können schließlich noch festhalten, daß die hier erörterte Auffassung so betrachtet werden kann, daß sie »abbilden» auf «aussehen wie» reduziert. Denn statt zu sagen, P werde gesehen, als obS vorhanden wäre, könnte man ebensogut sagen, daßP ,,wie S aussieht» (obgleich wir wissen, daß 5 nicht vorhanden ist). Die meisten Verfechter einer Illusionstheorie haben in der Tat ange-nommen, daß ihr eine bestimmte Auffassung von der Ähnlichkeit zwischen einem Bild und seinem Sujet zugrunde liege. Dieser Punkt verdient allerdings, gesondert untersucht zu werden.
Eine typische Aussage von diesem Standpunkt aus ist die
folgende von Beardsley: >,>Die ZeichnungX bildet einen Gegenstand Y ab‘bedeutet,
>X enthält eine Fläche, die der visuellen Erscheinung von Y-Gegenständen
ähnlicher ist als Gegenständen jeder anderen Klasse. <»30 138 140 2. Beim Zusammentreffen mit Zwillingsbrüdern sagst du:
,,Tom sieht tatsächlich wie Henry aus.» 142 nicht über eine vermeintliche Beziehung zu Wölfen. Wenn
dies so ist, dann widersetzt sich der Sinn, in dem ein realistisches Gemälde
»aussieht wie» sein Sujet, weiterhin der Analyse. Man könnte sogar zu sagen
geneigt sein, dieser Ausdruck müsse vermieden werden, da er leicht zu
Mißverständnissen führe. i 44 145 146 Anmerkungen Es besteht hier eine offensichtliche Analogie zur Unterscheidung zwischen Bedeutung und Referenz bei verbalen Beschreibungen. Im Falle einer solchen Beschreibung entspricht die Originalszene — falls es eine gibt — der Entität oder dem Ereignis, das durch die Beschreibung identifiziert wird. Das vom Gemälde gezeigte Sujei ist der ,47 148 rung dessen erforderlich sein, was Sich-Verheiraten heißt,
wenn man den Gedankeninhalt wiedergeben soll, falls man sich verheiratet?.
(5. 61) Wird Gedanke durch Intention ersetzt, entspricht die Struktur meinem
oben angeführten Argument. 150 151 Bibliographie
Anscombe,
G. E. M., On promising and its justice, and wheeher it needs he respected ,in
foro interno<, in: Critica 3 (April/Mai 1969). 153 |
|
Some
questions. There on the wall is a painting: it plainly shows some racehorse
or other, with trees that might be beeches in the background and a stableboy
doing something or other with a pail in the foreground. That the picture
shows all these things, that all these things and more can be seen in the
painting, is beyond doubt. But what makes that painting a picture of a horse,
trees, and a man? More generally, what makes any ,,naturalistic“ painting or
photograph a representation of its subject? And how, if at all, does the
situation change when we pass to such ,,conventional“ representations as
maps, diagrams, or models? To what extent do ,,convention“ or
,,interpretation“ help to constitute the relation between any representation
and its subject? plainly einfach beeches Buchen beyond
doubt zweifellos These are
the sorts of questions that I would like to consider (though not all of them
in this essay), more in the hope of clarifying the questions themselves than
in the hope of finding acceptable answers. For the main difficulties in the
inquiry arise from lack of clarity in such words as ,representation‘,
,subject‘, and ,convention‘ that spring naturally to mind and cannot be
avoided without tiresome paraphrase. consider erwägen though. obwohl inquiry. Abfrage arise entspringen tiresome. ermüdend paraphrase umschreiben Preliminary
qualms.
Are questions as imprecise
and confused as these worth raising? Or are they perhaps merely symptoms of
the philosopher‘s itch to puzzle himself about what seems unproblematic to
everybody else? Well, the itch is infectious, even to laymen. The
disconcerting thing, as we shall soon see, is that the most plausible answers
that suggest themselves are open to grave and perhaps fatal objections. Questions Fragen imprecise ungenau merely lediglich itch jucken grave. Grab Yet even partial answers have
consequences for such varied topics as perception, cognition, the structure
of symbol systems, the relations between thought and feeling, and the
aesthetics of the visual arts. To understand the reasons for the excessive
discord produced by questions about representation would be sufficient reward
for what threatens to be an arduous, even a somewhat tiresome, investigation. discord Uneinigkeit sufficient zureichend threatens bedroht arduous anstrengend Some
working definitions. When a painting—or some other visual representation,
such as a photograph—is a painting of something,
S, I shall say that P depicts S;
alternatively, that P stands in a depicting
relation to S. There is, however, room for misunderstanding here. Suppose P shows Washington crossing the Delaware. Then P is related to
Washington‘s actual crossing of the Delaware, in 1776, in such a way that it
can be judged to be a more or less faithful, or a more or less inaccurate,
painting of that historical episode. I do not want to count that historical
event as a special case of the subject, S, introduced in the last paragraph. judged beurteilte faithful treu to count zählen Consider for contrast the case of another
painting that shows Hitler crossing the Hudson in 1950: here there is no
actual event to serve as a control of the painting‘s fidelity; yet we still
want to say that the painting has a ,,subject“ that is depicted by it. Consider. erwägen Let us call Washington‘s crossing of the
Delaware the original scene to which
the painting refers; and for the sake of precision let us say that the
painting does not merely depict that scene but rather portrays it. Then the painting of Hitler‘s imaginary river
crossing will have no original scene to portray; but it may be said to display a certain subject. Thus
portraying and displaying will count as special cases of depicting. the sake der seinetwegen refers verweist merely lediglich display Anzeige count. zählen The displayed subject might be conceived of as the content of the
visual representation.1 I shall normally be concerned with depicting in the
special sense of ,,displaying,“ and correspondingly with that ,,subject“ of a
painting that is its ,,content,“ not its ,,original scene.“ Where the context
is suitable and no other indication is given, the reader may take ,,depict“
and ,,display“ henceforward as synonymous. conceived ersann concerned betraf content Inhalt suitable passend, verwendbares displaying zeigend henceforward von da an It should be borne in mind that I am not
initially committed to there always being an irreducible difference between
,,subject“ and ,,original scene.“ More importantly still, I am not committed—
as talk about the relation between P and S might misleadingly suggest—to the
existence of the ,,subject“ S as an independent entity in its own right. It
is compatible with everything that has so far been said that displaying-S might turn out to be a
unitary predicate, carrying no implication as to the existence or
non-existence of S. Thus, the occurrence of S in ,depicting-S‘ might be
,,intensional,“ not ,,extensional.“ borne in mind getragen
im Sinn committed verpflichtete misleadingly irreführend
suggest vorschlagen independent unabhängig displaying zeigend might turn out Macht
Gewalt hinauswerfen unitary=einheitlich einheitlich=einheitlich occurrence Vorkommnis In
search of criteria. An ambitious investigator of our syllabus of
questions might hope to discover an analytical definition of displaying or displaying-S, that is to say, some formula having the structure: P
displays S if and only if R, where ,R‘
is to be replaced by some expression that is more detailed and more
illuminating (whatever we take that to mean) than the unelaborated word,
,,displays.“ R, then, will
constitute the necessary and sufficient condition for P to ,,display“ S. discover entdecken investigator Erforscher (whatever we take that to mean) (was
auch immer nehmen wir damit zu meinen) unelaborated unelaborated It is probably unrealistic to expect that
we can find a set of necessary and sufficient conditions conforming to this
pattern. But we need not be committed to this goal, for it would already be
somewhat illuminating if we could only isolate some necessary conditions. A
still more modest goal, difficult enough to attain, would be to exhibit some criteria for the application of
expressions of the form ,displays S‘, that
is to say conditions that count, in virtue of the relevant meaning of
,displays‘ and nothing else, for or against its applicability
to given instances. 1 There is an obvious analogy
here with the sense/reference distinction connected with verbal descriptions.
In the case of such a description, the ,,original scene,“ if any, corresponds
to the entity or event identified by the description. The displayed subject
of the painting is analogous to the description‘s sense or meaning, which
attaches to it whether or not it identifies any actual entity or even t. attain erreichen pattern gestalten applicability Anwendbarkeit instances Instanzen attain erreichen in virtue im Jahre virtue Such
criteria need not be invariably or universally relevant, always in point
whenever it makes sense to speak of something showing or displaying something
else: the criteria of application might vary from case to case in some
systematic and describable fashion. Vary abwandeln describable zu beschreibend fashion Mode The goal
would then be partial but explicit insight into the pattern of uses of
,depicts‘ and its paronyms, rather than a formal definition.2 Nevertheless,
although this is the program, I shall first adopt the plan of examining
seriatim a number of plausibly necessary conditions for ,,displaying.“ Only
after we have become convinced that none of these candidates separately—and
not even all of them jointly—can serve as an analysis of our quaesitum shall
I proceed to argue for a more flexible type of answer. seriatim seriatim paronyms paronyms although obgleich convinced überzeugte jointly gemeinschaftlich quaesitum=Fragestellung proceed fortsetzen to argue zu argumentieren Somebody
influenced by Wittgenstein‘s parallel investigations of fundamental concepts
might regard our enterprise, even thus circumscribed, as reprehensibly
quixotic, expecting to find the ,,patterns of use“ too ravelled for reduction
to any formula. Such pessimism might be countered by recalling that
,,creative power“ of language, as important as it is truistic, in virtue of
which we can understand what is meant by something of the form ,,P is a
painting of S“ even when asserted of some painting in a new, unfamiliar or
recondite style. That we think we understand what is said in such contexts
argues powerfully for the existence of an underlying pattern of application
waiting to be exhibited. Even if this were an illusion, it would be one in
need of explanation. enterprise regard Hinblick reprehensibly
tadelnswert quixotic phantastisch ravelled ravelled truistic truistic virtue Tugend asserted behauptete unfamiliar ungewohnt recondite dunkel A principled
objection. A
search for analytical criteria of application for ,displaying‘ naturally
recalls the many abortive efforts that have been made to provide partial or
complete analyses of verbal meaning. Indeed, there may be more than analogy
here. Of a verbal text we can properly say, echoing some of our earlier
formulations, that it is a description or representation of some scene,
situation, or state of affairs, presented via the ,,content“ of the text and
possibly corresponding to some verifying fact (the analogue of our ,,original
scene“). Thus questions arise that seem to parallel those about pictorial
representation which I have em- 2 Cf. the corresponding analytical task for ,,the concept of cause,“ where, in my opinion, the best to be hoped for is a
similar mapping of variations of sense, based upon an exhibition of the
relevant criteria of application underlying such Variation. abortive vorzeitig efforts Bemühungen echoing widerhallend arise entspringen phasized.
Some writers, indeed, try to assimilate the latter to the former, drawing
their explanatory and analytical concepts from the domain of verbal
semantics. latter Now are
there any good reasons to think that a search for a conceptual map of verbal
meaning is bound to be abortive? The late John Austin seems to have thought
so (cf. Austin, 1961), for reasons that do not seem to have been discussed in
print. abortive seem Austin
contrasts a search for the meaning of a particular word or expression with
what he takes to be the illegitimate search for meaning in general. He
reminds us that the first kind of question is answered when we can ,,explain
the syntactics“ and ,,demonstrate the semantics“ of the word or expression in
question. That is to say, when we can state the grammatical constraints on
the expression and its ostensive or quasi-ostensive links with non-verbal
objects and situations (where this is appropriate). constraints Nötigungen state Staat constraints Nötigungen But then Austin
objects. that the supposedly more general question ,,What is the meaning of a
word-in-general?“ is a spurious one. ,,I can only answer a question of the
form ,What is the meaning of ,,x“?‘ if ,x‘ is some particular word you are
asking about. This supposed general question is really just a spurious
question of a type which commonly arises in philosophy. We may call it the
fallacy of asking about ,Nothing-in-particular‘ which is a practice decried
by the plain man, but by the philosopher called ,generalizing‘ and regarded
with some complacency“ (p. 25). spurious spornig, unechte spurious spornig, unechte commonly häufig fallacy Trugschluß decried in Verruf gebracht complacency=
herabsetzte If Austin
were right, a similar absurdity should infect our main question about the
meaning of ,,P is a painting of S,“ since we are not raising it about any
given painting, but about no painting in particular. Austin‘s idea, so far as
I have been able to follow it, seems to have been that the rejected general
question about the meaning of a word in general is apt to be taken as a
search for a single meaning, common to all words. And similarly, if he were
right, our own problem would be the ridiculous one of trying to find a single
common subject for all paintings. So conceived, the quest would indeed be
spurious, not to say preposterously confused. raising apt ridiculous conceived spurious It is
interesting to notice, however, that immediately after rejecting the
,,general question“ about meaning, Austin admits as legitimate the general
question ,,What is the ,square root‘ of a number?“—of any number, not any
particular number. admits square
root But if
that is a legitimate way of looking for a definition of ,square root‘, why
should we not regard the ,,general question“ about meaning, with equal
justice, as a legitimate way of looking for a definition—or at least
something relevant to the definition—of ,meaning‘? Austin‘s
reason for distinguishing the two cases apparently arises from his conviction
(which I share) that ,, ,the meaning of p‘ is not a
definite description of any entity“ (p. 26), whereas ,square root of n‘ is
(when the variable is replaced by a constant). I cannot see why this should make
a relevant difference. It would indeed be naive to think of the search for an
analysis of ,displaying S‘ as presupposing the existence of some entity
indifferently displayed by all paintings. distinguishing apparently arises conviction whereas root presupposing But this
concession need not imply the rejection of any search for general criteria as
spurious. Oddly enough, after Austin‘s characteristically energetic attack
upon the ,,general“ enterprise of delineating the concept of meaning, he
proceeds to assign that same task an acceptable sense by taking it to be an
attempt to answer the question ,,What-is-the-meaning-of (the phrase) “the meaning-of (the word) ,,x“‘?“. spurious Oddly upon delineating proceeds assign attempt The
corresponding question for us might be ,,What is the meaning of (the phrase)
,displaying-S‘?“. We can then proceed, as Austin recommends, to investigate
the ,,syntactics and semantics“ of that expression, without commitment to the
existence of dubious entities and without nagging anxieties about the
supposed spuriousness of the enterprise. It may prove impossible to reach our
goal, but that remains to be seen. recommends commitment nagging anxieties spuriousness How does
a photograph depict? I shall now consider for a while the form taken by our
basic question about the nature of depiction, when applied to the special
case of photographs. For if any pictures stand in some ,,natural“ relation to
their displayed or portrayed subjects, unretouched photographs ought to
provide prime examples. There, surely, we ought to be able to discern
whatever complexities underlie the notion of faithful ,,copying“ at its least
problematic. Since photographs also provide minimal scope for the
,,expressive“ intentions of their producers, we shall be able to bracket
considerations connected with the expressive aspects of visual art (whose
crucial importance in other contexts is, of course, undeniable). depict depiction ought discern provide underlie scope considerations crucial undeniable What,
then, is it about a given photograph, P, that entitles us to say that it is a
photograph of a certain S? Here is a picture postcard labelled ,,Westminster
Abbey.“ What justifies us and enables us to say that it is a photograph of a
certain famous building in London—or, at least, for somebody who has never
heard of the Abbey, of a certain building having certain presented properties
(twin towers, an ornamented front, and so on)?3 entitles labelled properties Appeal
to a causal history.
The first answer to be considered seeks to analyze the imputed depicting
relation in terms of a certain causal sequence between some ,,original scene“
(what the camera was originally pointed at) and the photograph, considered as
an end-term in that sequence. The photograph, P, that is to say a certain
piece of shiny paper, showing a distribution of light and dark patches,
resulted—so the story goes—from a camera‘s being pointed upon a certain
occasion at Westminster Abbey, thus allowing a certain sheaf of light rays to
fall upon a photo-sensitive film, which was subsequently subjected to various
chemical and optical processes (,,developing“ and ,,printing“), so that at
last this object—the photograph in our hands—resulted. In short, the
aetiology of the representing vehicle, P, is supposed to furnish our desired
answer. Reducing
the causal narrative to essentials, we get the following account: P portrays
S, in virtue of the fact that S was a salient cause-factor in the production
of P; and P displays S‘, in virtue of the fact that S satisfied the
description that might be inserted for ,,S“‘ (a building having towers, etc.).
On this view, P might be regarded as a trace4 of S .and the interpretation of
P is a matter of inference to an earlier term in a certain causal sequence. An
immediate obstacle to the acceptability of this account is the difficulty in
specifying the ,,portrayed subject,“ S, and that abstract of it which is the
,,displayed subject,“ S‘. For an inference from P to the circumstances of its
original generation will yield any number of facts about the camera‘s focal
aperture, its distance from the nearest prominent physical object, perhaps
the exposure time, and so on, which we should not want to count as part of
the ,,subject“ in either sense of that word. There must be some way of selecting,
out of the set of possible inferences from the physical character of P, some
smaller set of facts which are to count as relevant to P‘s content. 3 I am here
deliberately ignoring, for the time being, the distinction previously
introduced between the displayed and the portrayed subject. For to make this
distinction at once might interfere with the plausibility of the first answer
now to be considered. 4 ,,lt will be heipful
if we look at images as traces, natural or artificial ones. After all, a
photograph is nothing but such a natural trace, a series of tracks left... on
the emulsion of the film by the variously distributed lightwaves which
produced chemical changes made visible and permanent through further chemical
operations“ (Gombrich 1969, p. 36). An
associated difficulty lurks behind the lazy formula that identifies S, the
displayed subject (and our main interest), with a certain ,,abstract“ of the
properties of S. At the moment the photograph was taken, the Abbey must have
had any number of properties that might be inferred from the photograph,
though irrelevant to that photograph‘s content. (If the picture showed the
doors open, one might correctly infer that visitors were to be found in the
Abbey‘s interior on that day.) Even to limit S‘ to a specification of visual
properties will not serve: somebody might be able to infer correctly all
manner of conclusions about the Abbey‘s visual appearance (e.g., that it
looks as if it were leaning on the spectator) without such items being shown
in the picture. Some
writers have thought that such objections might be overcome by considering
the ,,information“ about the Abbey‘s visual appearance at a certain moment,
supposedly embodied in the final print. Such ,,information“ is conceived to
have been contained in the sheaf of light-rays originally impinging upon the
camera‘s lens and to have remained ,,invariant“ through all the subsequent
chemical transformations. What makes anything, A, a ,,trace“ of something
else, B, is just that A in this way presents information about B. Examination
of the immanent ,,information“ contained in P would thus presumably allow us
to distinguish between warranted and unwarranted inference to P‘s aetiology
and so to eliminate the uncertainties about the displayed and the portrayed
subjects. I shall subject this conception to criticism below. But we might as
well notice another difficulty at once. Suppose
the causal history of a certain photograph to be as outlined above (pointing
at the Abbey, chemical changes in the photo-sensitive emulsion, and so on)
while the final outcome consisted of nothing better than a uniform grey blur.
Should we then, in order to be consistent, have to maintain that we did
indeed end with a photograph of Westminster Abbey, though a highly
uninformative one?S Notice that the Abbey might indeed present the 5 Some members of my
audience at the lecture on which this essay is based were willing to take
this view, insisting that the aetiology had over-riding importance, appearance
of a grey blur if seen through eyelids almost closed: perhaps the
,,uninformative“ photograph should be regarded as yielding only an unusual
view of the Abbey? But this is surely too paradoxical to be acceptable. The moral
of this counter-example is that reliance upon the photograph‘s history of production
is insufficient to certify it as having the Abbey, or an abbey, as its
subject. No genetic narrative of the photograph‘s provenance, no matter how
detailed and accurate, can logically guarantee that photograph‘s fidelity.
(Of course, if the ,,accuracy“ of the causal account is to be determined by
some other kind of test—say ,,invariance of information,“ construed as
involving no reference to any causal history—the causal account is already
shown to be insufficient.) The
causal account I have imagined, now seen to be insufficient as an analysis of
the photograph‘s content, can also be shown not to be necessary either. Suppose
someone invents a new kind of photo-sensitive paper, a sheet of which, upon
being ,,exposed“ by simply being held up in front of the Abbey, immediately
acquires and preserves the appearance of a conventional photograph. Would we
then refuse to call it a photograph of the Abbey? We might perhaps not want
to call it a photograph, but no matter: it would still surely count as a visual
representation of the Abbey. A
defender of the causal approach might retort that the extraordinary paper I
have imagined was at least printed at the Abbey, so that the ,,essentials“ of
the imputed causal history were preserved. After all, he might add, we are
not even normally interested in the details of the particular chemical and
physical processes used in producing the final print. Well, so long as we are
indulging in fantasy, let us suppose that the extraordinary imprinting effect
was producible only by pointing the sensitive paper away from the Abbey,
while the result was still indistinguishable from the conventional
photograph: would that disqualify the product as a visual representation and,
for all we know, a highly faithful one? Some
philosophers might reply that if causal laws were violated no matter how disappointing the resulting
,,trace.“ This illustrates the grip that the causal model can have—to the
point of its being accepted in the teeth of the most absurd consequences. in these
or other extraordinary ways, we ,,should not know what to think or say.“ But
that seems a lazy way with a conceptual difficulty. Perhaps a single odd
example of the kind I have imagined would leave us hopelessly puzzled. But if
the phenomenon were regularly reproducible by a standard procedure, I suppose
we should be justified in saying that we had simply discovered some new,
albeit puzzling, way of producing representations or ,,likenesses“ of the
Abbey and other objects. lt would be easy to concoct any number of other
counter-examples in which end-products indistinguishable from conventional
photographs might arise from radicalJy unorthodox procedures. One might
be inclined to draw the moral that the causal histones
of photographs—or their fantastic surrogates—are wholly irrelevant to our
warranted judgments that they are depictions of the Abbey. But to reject the causal view so
drastically may be too hasty. Suppose we found some natural object that
,,looked like“ a certain subject—say, a rock formation that from a certain
standpoint looked for all the world like Napoleon: should we then say that the
rock formation must be a representation of Napoleon? (Or suppose, for that
matter, that objects, looking for all the world like photographs,
simply rained down from the sky at certain times.) Surely not. lt looks as if the background
of a certain aetiology is at least relevant (without being either a necessary
or a sufficient condition) in ways that need to be made clearer before we are
through. An
obvious counter might be that in the case of the imagined sensitive paper,
and in the other examples that we might be inclined to concoct, something
must be deliberately positioned in order to create, if all goes well, a
representation of the subject in question. To be sure, this would exclude the
supposed counterinstances of the natural objects, or the objects of unknown
provenance that were simply indistinguishable from conventional photographs.
But in this imagined objection, there is plainly an appeal to a very
different sort of criterion, the intention that launched the causal process. This
deserves separate discussion. Let us
first, however, look more closely at the suggestion that the notion of
,,information“ provides the clue for which we are searching. Appeal to
embodied ,,information.“ As I have already said, our puzzle about the blurred
photograph would be regarded by some writers as explainable by the ,,absence
of sufficient information“ in the final print. More generally, the concept of
,,information,“ supposedly suggested by the notion thus designated in the
mathematical theory of communication, is held to be useful in resolving the
conceptual difficulties that we are trying to clarify.6 The
current vogue for speaking about ,,information“ contained in
representations—and indeed, for bringing that notion into almost any kind of
discussion—is certainly influenced by the supposed successes of the notion of
,,information“ that is prominent in the sophisticated mathematical theories
usually associated with the name of Shannon.7 Yet it is easy to show that the
two senses of ,information“ involved have very little to do with one another. Let us
recapitulate briefly what ,,information“ means in the context of the
mathematical theory. Ihe first point to be made is that in that theory we are
dealing with a statistical notion—let us call it ,,selective information“8
henceforward to avoid confusion. The typical situation to which the
mathematical theory applies is one in which some determinate stock of
possible ,,messages,“ which may be conceived as alternative characters in an
,,alphabet“ (letters, digits, or pulses of energy) to which no meaning is
necessarily attached, are encoded into ,,signals“ for transmission along a
,,communication channel“ and ultimate reception, decoding and accurate
reproduction of the original ,,message.“ Thus individual letters of the
English alphabet are converted into electrical pulses along a telegraph wire,
in order to produce at the other end a copy of the original string of letters
composing the complex message sent. 6 Gombrich, influenced
by Professor J. J.
Gibson‘s writings on perception, has suggested that
instead of speaking about ,,interpretation“ we speak instead about how ,,the
sensory system picks up and processes the information present in the energy
distribution of the environment“ (Gombrich 1969, p. 47>. He adds that he is
,,fully alive to the danger of new words, especially fashionable words,
becoming new toys of little cash value“ (ibid.>. But he relies strongly
upon what he takes to be ,,the concept of information [as] developed in the
theory of communication“ (p. 50) throughout his article. 7 See Shannon and
Weaver (1949) and Cherry (1966) for explanations of the technical theory. It
is ironic that experts in information theory have repeatedly protested,
apparently without success, about the misleading consequences of identifying
what is called ,,information“ in the technical theory with the meaning of
that word in ordinary language. 8 See Cherry (1966),
p. 308. A rough
explanation of the notion of the ,,selective information“ associated with
such a communication system would identify it with the amount of ,,reduction of initial uncertainty“ that such a system can achieve. Suppose the various possible messages m i are
known to occur with long-run frequencies or probabilities p i. We might say
that the ,,information“ conveyed by the receipt of a particular message, m i,
varies inversely as its initial probability of occurrence, p i. For, the
higher the initial probability of transmission, the ,,less we learn“ by
receiving the message. If the message in question were, in the limiting case,
certain to arrive, we should ,,learn nothing“ by receiving it. The mathematical
quantity called the (selective) information is the measure of the amount of a
certain magnitude—roughly speaking, the reduction in the amount of initial
uncertainty of reception, as I have suggested above. It is important to
stress that this has nothing to do with the meaning, if any, of such a
message, and nothing to do even with its specific content. If I seek
an answer to a question by means of a telegram, the only two possible answers
being either Yes or No, and both being antecedently equally likely to be
sent, then either answer contains the same ,,(selective) information.“ Each
answer transforms a probability of 1/2 into certainty. An anxious
suitor, awaiting an answer to his proposal of marriage, would of course say
that the information received in the one case would be interestingly
different from the information he would receive in the other. But that is
because he is using ,,information“ in the common or garden sense of what
might be called substantive information. The theorists of the mathematical
theory have no interest in substantive information—which is, of course, their
privilege, and not a reproach. To think otherwise would be as misguided as to
make it a reproach to a theory of measurement that it tells us nothing about
the smell or taste of the masses discussed in that theory. It would
obviously be pointless to think of adapting this model to the case of
representation. In place of the messages we should have to think of the
original ,,scenes“ corresponding to the representations, to which the
long-term frequency of occurrence would have no sensible application. Even in
some special case, say that in which male and female entrants to a college
were photographed with stable long-range frequencies, the selective or
statistical information attached to any photograph would tell us absolutely
nothing about the photograph‘s subject or content—which is our present
interest. A few
writers who have clearly seen the limited applicability of the statistical concept
here called ,,selective information“ have undertaken studies of what they
call ,,semantic information,“ that might seem more
useful for our purpose.9 For it would seem that ,,semantic information,“
unlike selective (statistical) information, is concerned with the ,,content“ or meaning
of verbal representations (statements, texts). The theory of semantic
information is presented as a rational reconstruction of what Hintikka calls
,,information in the most important sense of the word, viz., the sense in
which it is used of whatever it is that meaningful sentences and other
comparable combinations of symbols convey to one who understands them.“10 Now this
seems to be just what we are looking for: the ,,displayed subject“ of a
photograph does seem dose to what common sense would call the information
that could be understood by a suitably competent receiver (viewer).“ However,
if we follow the constructions provided by Hintikka and other pioneers of
,,semantic information theory“ we shall discover, to our disappointment, that
they, too, provide something, of whatever interest, that will not help us in
our present investigation. 9 See especially
Bar-Hillel (1964), Chapters 15—17, and Hintikka <1970>. Bar-Hillel
says: ,,lt must be perfectly clear that there is no logical connection
whatsoever between these two measures, i. e., the amount of (semantic)
information conveyed by a statement and the measure of rarity of kinds of
symbol sequences [our ,,seleetive information“], even if these symbol sequences
are typographically identical with this statement“ (p. 286, italics in
original). And again, ,,The concept of semantic information has intrinsically
nothing to do with communication“ (p. 287, original italics)—and hence it has
nothing to do with the concept of information that is defined relative to
communication systems. On the other hand, Hintikka says he has ,,become
increasingly sceptical concerning the possibility of drawing a hard-and-fast
boundary between statistical information theory and the theory of semantic
information“ (p. 263). us suggestion that semantic information theory start
with ,,the general idea that information equals elimination of uncertainty“
(p. 264), a formula that would, as explained above, fit the case of selective
(statistical) information, shows the link that Hintikka relies upon—in spite
of Bar-Hillel‘s vigorous attempt to separate the two concepts. 10 Hintikka 1970, p. 3. 11 There is, to be
sure, some violence done to ordinary language here. Common sense would
reserve the use of ,,information“ for what is conveyed by the photograph
about the original scene. lt would be paradoxical to think of the painting of
an imaginary scene as providing information to anybody—just as paradoxical as
supposing that The Pickwick Papers contains information about Mr. Pickwick. For it turns
out that the ,,semantic information“ of a given statement is roughly the same
as the range of verifying situations associated with that statement—or, more
accurately, some measure of the ,,breadth“ of that range. And here, what is
finally provided is a measure of extent and not of content. ,,Semantic
information“ is a sophisticated refinement of the common sense notion of the
amount of information in a statement. Just as a report of a body‘s mass tells
us nothing about what stuif that body is composed of, so a report of semantic
information would tell us nothing about what the statement in question is
about. If we
have two statements of parallel logical structure, say, ,,My name is Black“
and ,,My name is White,“12 any acceptable definition of semantic information
will assign the same semantic information to each statement. If this concept
(whose interest I do not wish to deny) were applicable to paintings,13 we
should have to count distinct paintings with roughly comparable subjects
(say, two paintings of a flock of grazing sheep) as having and conveying the
,,same information.“ But of course the displayed subjects of two such
paintings might be manifestly different. We need a
term to distinguish what Hintikka, as we have seen, called ,,the most
important sense of information,“ i.e., what we mean by that word in ordinary
life: let us call it substantive information. And let us stretch the word to
apply to false statements as well as to true ones (so that what would
ordinarily be called ,misinformation“ also counts as substantive, but
incorrect, information). What then could we mean by talking about the
(substantive) information contained in a given photograph? 12 On the plausible assumption
that the two surnames occur with equal frequency in the populations in
question. 13 I do not think that anybody
has yet tried to apply the concept of semantic information to visual representations.
One difficulty, and perhaps not the most serious one, would be that of
,,articulating“ a given verbal representation in a way to correspond to the articulation
of statements in a given language into an ordered array of
phonemes. Unless we can regard a photograph or a painting, by analogy, as
composed of atomic characters, corresponding to phonemes, the desired analogy
will hardly find a handhold. Ihen, of course, there is the lurking difficulty
behind any attempt to assimilate paintings or other verbal representations to
assertions with potential truth-value. This might serve for blueprints,
graphs, and other representations designed to convey purported facts
(,,information“ in the ordinary sense>, but would hardly fit, without
inordinate distortion, our prime case of paintings. On the
assumption that we have a sufficiently firm grip upon the notion of the
substantive information contained in a statement, one might think of
replacing the given photograph, P, by some complex statement, A, such that a
competent receiver might learn just as much from A as from P, if P were to be
a faithful record of the corresponding original scene. But surely there is
something fanciful about this suggestion. Suppose somebody were to be
presented with such a statement, A, and asked then to retrieve the
photograph, P, of which it is supposed to be in some sense a translation,
from a large set of different photographs. Is there good reason to think that
such a task must be performable in principle? lt seems to me, on the
contrary, that the notion of a complete verbal translation of a photograph
(and still more, the notion of a verbal translation of a painting) is a
chimaera. A picture
shows more than can be said—and not simply because the verbal lexicon is
short of corresponding equivalents: it is not just a matter of the nonavailability
of verbal names for the thousands of colours and forms that we can
distinguish. But if so, the notion of. information that has its habitat in
connection with verbal representations (statements) will still fail to apply
to the case that interests us.14 In the end, it seems that what is
picturesquely expressed by means of the figure of the ,,information“
contained in a photograph or a painting comes to nothing else than what we
mean when we talk about the ,,content“ of the painting or ,,what it shows“
(its displayed subject). There would be no objection to the introduction of a
metaphor or analogy based upon information, if that provided any
illumination. lt seems to me, however, that this is not the case, and that
reliance upon ,,information“ on the basis of a more or less plausible analogy
amounts in the end only to the introduction of a synonym—and a misleading one
at that—for ,depicting‘ or ,representing.‘ It is not unfair to suggest that
,,the information conveyed by a painting“ means nothing more than ,,what is
shown (depicted, displayed) by that painting.“ 14 Of course, I do not wish to deny that we
can put into words some of the things that we
can learn from a faithful photograph. If anybody wants to express this by
saying that information can be gleaned from such a photograph, there can be
no harm in it. But we shall never in this way be able to identify the subject
of the photograph. One is at this point strongly inclined to say that in some
sense the visual subject can only be shown. We may,
nevertheless, draw a useful lesson from this abortive digression. One
caution, stressed repeatedly by theorists of statistical and semantic
information alike, is that the measures of information they discuss are
always relative to a number of distinguishable factors in the relevant
situations. In the case of statistical information, the amount of information
is relative to the distribution of long-term frequencies of the system of
possible messages transmissible in the communication channel in question; in
the semantic case, the amount of information embodied in a statement is
relative to the choice of a language and, on some treatments, to assumptions
about given laws constituting an antecedent stock of given ,,information“ to
which any statement not inferrible from those laws makes an additional
contribution. We might,
therefore, be encouraged to draw a somewhat obvious moral: that however we
come to identify or describe the substantive content of a painting or other
visual representation, the answer will be relative to some postulated body of
knowledge (concerning for instance the chosen schema of representation, the
intentions of the painter or sign-producer, and so on). The idea that a
painting or a photograph ,,contains“ its content or subject as
straightforwardly as a bucket contains water is too crude to deserve
refutation. But ideas as crude as this have in the past controlled some of
the discussions of our present topic. Appeal to
the producer‘s intentions. I shall now consider the suggestion that a way out
of our difficulties might be found by invoking the intentions of the painter,
photographer, or whoever it was that acted in such a way as to generate the
visual representation whose ,,subject“ we are canvassing. I do not know of
any theorist who has based a full-fledged theory of representation on this
idea, but corresponding theories of verbal representation are fairly common.
Thus Professor Grice, in a well-known paper on ,,meaning,“15 has urged that
the meaning of an utterance can be analysed in terms of certain complex
intentions to produce a certain effect in the hearer.16 15 See Grice (1957) and
(1969) for the elaboration and modification of his position in reply to
criticism. 16 Ihe details do not concern US here. Ihe novelty in Grice‘s account consists
in differentiating between a primary intention on the speaker‘s part to
produce a certain belief or action in the hearer, and a secondary intention
that recognition of that primary intention shall function as a reason for the
hearer to comply with the primary intention. Again,
Professor E. D. Hirsch, in a well-known book, has defined verbal meanings as
,,whatever someone has willed to convey by a particular sequence of
linguistic signs and which can be conveyed (shared) by means of those
linguistic signs.“17 Ihere seems to be no reason in principle why this kind
of approach should not be equally valid in connection with visual
representation—or, indeed, in connection with any sort of representation at
all. The
undeniable attraction of this kind of emphasis upon the producer‘s intention
or ,,will“ can be attributed to its tendency to remind us forcibly about the
conceptual gap between the ,,interpretation“ of some natural object (as when
we infer from the characters of some trace to the properties of something
that produced that trace), and the ,,interpretation“ of a man-made object,
intentionally created to have meaning or ,,content“ of a sort that is
accessible to a competent receiver. But to agree, in Grice‘s terminology,
that the import of a painting is ,,non-natural“ and not reducible to the
termini of factual inferences from the vehicle is one thing; to suppose that
the determinable subject of such a nonnatural object can be defined in terms
of features of the producer‘s intentions is something else that is far more
problematic. There is,
to begin with, the immediate objection that the producer‘s intention,
supposing it to have existed in some uncontroversial way, may misfire.
Suppose I set out to draw a horse and, in my lack of skill, produce something
that nobody could distinguish from a cow by simply looking; would it then
necessarily be a drawing of a horse, just because that was what I had intended?
Could I draw a horse by simply putting a dot on paper? If the answers were to
be affirmative, we should have to regard the artist‘s intentions as having
the peculiar character of infallibility: simply wanting a painting to be a
painting of such-and-such would necessarily make it so. Surely this is too
paradoxical to accept. Of a botched and unrecognizable drawing we should want
to say ,,He intended to draw a horse, but failed“ as we should say, in
certain circumstances, of any failed intention. 17 Hirsch (1967), p.
31. The reference to meaning as something that can be conveyed to and shared
by others shows that Hirsch is not simplistically identifying the content of
an utterance with the content of the speaker‘s intention. For elaboration of
his views about this, see pp. 49—50 of his book. I have no quarrel with
Hirsch‘s vigorous case for the need to refer to an author‘s intentions in
providing adequate interpretations of a text. The
notion of intention involves the notion of possible failure. A still
more serious difficulty and one, if I am not mistaken, that is fatal to this
approach, is that there is no way of identifying the relevant intention
except by invoking the very notion of a subject of a possible painting that
such reference to intention is supposed to clarify. Let us take the envisaged
analysis in its crudest and least defensible form. Suppose the proposed
analysis of ,,P depicts 5“ were to be ,,M, the producer of P, intended S to
be a depiction of S.“ In this form, the logical circularity is patent: we
could not understand the proposed analysis of depicting without already
having a clear notion of that relation at our disposal. Nor could the invoked
producer properly have any explicit intention to produce P as a depiction of
S unless he independently understood what it would be like for the resulting
P in fact to be a painting of S: to refer back to the intention that he would
have if he were to be trying to make P depict S would enmesh him in hopeless
circularity.18 The
situation would be less objectionable, if the proposed analysis were to take
the form: ,,P depicts S if and only if M, the producer of P, intended E,“
where E is imagined replaced by some complex expression (and not a
straightforward synonym of ,P depicts S‘).19
Then the circularity
noted above would be absent. Only in order for this kind of analysis to be
acceptable, E must have the same extension as ,P depicts S‘: we have captured
the right intention only if what M intended to do was necessary and
sufficient for P being a depiction of S (though not expressed in those
words). And if so, we can then dispense with the reference to M‘s intention
altogether, since ,,P depicts S if and only if E“ will, by itseif, constitute
the analysis we were seeking. 18 A similar point has been well made by Miss
Anscombe (see Anscombe 1969). ,,If thinking you are getting married is essential
to getting married, then mention of thinking you are getting married belongs in
an explanation of what getting married is; but then won‘t an explanation of
what getting married is be required if we are to give the content of thought
when one is getting married?“ (p. 61). With intention replacing thought, this
is the structure of my own argument above. 19 This is, in fact,
the structure of Grice‘s analysis of non-natural meaning, which is therefore
not open to the charge of immediate circularity. This way
of looking at the matter would also have the advantage of meeting the
difficulty about the failed intention that we noticed above.20 I
conclude that in spite of its attractions, the appeal to the producer‘s
intention accomplishes nothing at all to our purpose. Depiction
as illusion. We have now considered three types of answers to our prime
questions as to the analysis of a statement of the form ,P depicts S‘. Of
these, one, the reliance upon the ,,information“ supposedly embodied in the
representation, P, seemed empty, and the two others, in appealing to a causal
history and the producer‘s intention respectively, invoked temporal
antecedents that seemed only contingently connected with the final outcome.21
We still need, it seems, to isolate something about Ihe representation itseif
that will, in favorable circumstances, permit a qualified and competent
viewer to perceive in the art object, without dubious inferences to
antecedent provenance or partially fulfilled intentions,22 something about P
that makes it a painting of S and nothing else. The
reader may be surprised that I have waited until now to consider a famous answer
that has behind it the authority of Aristotle and a thousand other theorists
who have, in one form or another, endorsed his conception of art as mimesis.
Let us try to formulate the conception of art as an ,,imitation of reality“
in a way that will commit us to as little presupposed theory as possible. 20 lt might be an interesting feature of
die logical grammar of ,depicts‘ if it true that ,P depicts S‘ entailed ,Ihe
producer of P intended P to depict S‘. <Cf. ,,That counts as a move in the game
only if die player intended to make diat move“—which need not be circular and
may indeed be informative.) Unfortunately, even diat is not true: die
photograph may show much that its producer did not intend and would not even
retroactively assimilate to his intention. Ihere is such a thing as
unintentional showing—as there is such a thing as unintentional speaking. 21 This is not strictly
true for the case of intention: what the artist successfully succeeded in
achieving, in accordance with an embodied intention, does usually determine
the intrinsic character of the product. 22 We might say that
die only intention diat is relevant is the intention diat die artist
succeeded in embodying in the painting. Of course, a knowledge of the
background—the tradition within which the artist was working, die purpose he
had in mmd, and odier things—may well help us to ,,read“ his painting, but
the satisfactory reading must in the end be based upon what is there to be
found in die painting. Why not
say that when I look at a naturalistic painting—say of a white poodle on a
sofa—it is as if, looking through the picture frame, I actually saw an animal
having a certain appearance, rest-ing on a piece of furniture at a certain
distance from me. Of course, I know all the time that there is no such poodle
in the place where I seem to
see it; and that is what makes the experience an illusion, but not a
delusion.23 We are not really deceived, but we have had enough visual
experience to know that we see what it would be like if the poodle were
really there. Ihere is a suspension of disbelief on the viewer‘s part, as
there is when reading fiction, which describes non-existent persons as if
they really existed. We might therefore speak of ,,fictive“ or ,,illusive“
vision in such cases. The
expression ,,as if,“ which I have used in my proposed formula, with its
obtrusive reminder of ,,The Philosophy of As If,“ may smack of hocus-pocus.
But this can be held in check, I think, and the expression treated as a harmless
shorthand. To say ,,A is as if B“ is simply to say ,,If B were the case, then
A; but also not-B.“ In a case of illusion, the observer knows that not-B, in
spite of appearances; in a case of deception or delusion, he believes that B
is the case, contrary to fact. Thus the proposed analysis for our imagined
case is: If there were a poodle of a certain sort and in a certain posture on
a sofa at such and such a distance from me, I would see what I now see.
Hence, I can see, here and now, what the subject of the picture is, without
reference to the painting‘s aetiology, the artist‘s intentions, or anything
else that is not immediately present. This
account must certainly have some truth in it: a layman, in the presence of a
painting by Claudio Bravo, will certainly report that it looks for all the
world as if there were a parcel behind the surface, and a viewer, no matter
how armored by theoretical commitments against the role of illusion, cannot
avoid, if he is ingenuous, making a similar report for other cases of trompe
l‘ceil. There is, of course, a serious question whether an account that seems
to fit this special type of case can be extended, without distortion or eventual
tautology, to fit all cases of response to visual representations that are
partially naturalistic. 23 For views of this
sort, see Gombrich (1961), passim. Given Professor Gombrich‘s illuminatingly
rich and detailed discussions of ,,illusion“ in his great book and elsewhere,
I would hesitate to saddle him with any simple view about die role of
illusion in art. Ihat he seems to assign a central, though by no means
exclusive, role to such illusion seems indicated by such references as ,,the
illusion which a picture can give“ (Gombrich 1969, p. 46). But he has always
carefully distinguished, as I wish to do, ,,the difference between an
illusion and a delusion“ (op. cit., p. 60). I think,
however, that there is no serious difficulty in stretching the view to cover
cases in which the presented subject is unfamiliar. There is no particular
puzzle, on this view, in accounting for the viewer‘s sight of a flying horse
or a fleshy goddess floating in the air. And the account can even be held to
fit certain ,,abstract“ works: if I see in a Rothko painting a receding plane
bounded by a contrasting strip, and so on, that sight is not unlike what I
have learned to see by looking at clouds. Similarly for Mondrian‘s Manhattan
BoogieWoogie or other such abstractions.24 On this
view, puzzles about how a P can depict a determinate S reduce to questions
about normal perception of the form ,,How is it that a real poodle can look
like a poodle?“ I am not sure what useful sense can be ascribed to a question
of this form;25 at any rate, it would fall outside the scope of
our present inquiry. Let us
now consider possible objections. The first type consists in effect of the
objection that the ,,illusion“ is not, and is not intended to be, complete.
As we shift our position with respect to the canvas, we do not get the
systematic changes in appearance that would occur if there really were a live
poodle in the indicated position: a painted canvas does not even produce as
much ,,illusion“ as a mirror. Furthermore, the presented visual appearance is
,,frozen,“ does not show the slight but perceptible changes to be seen in
even a ,,still life,“ and so on.26 24 I am not arguing
that looking ,,through the surface“ is a proper way of look-ing at all
abstractions: in Mondrian‘s case, we know it would be contrary to his
intentions. My point is only that the conception of depiction as illusion can
cover a far wider range of cases than is sometimes assumed. 25 The undoubted
interest of this type of question for psychologists such as Professor J. J. Gibson
(see, for instance, his 1968 book) arises from the need to explain how a fiux
of radiant energy, reaching the eyes, can be so processed that the viewer can
correctly see the poodle as solid, at a certain distance from him, and so on.
But answers to this kind of question, important as they are, are not our
concern here. Cases of veridical experience are sufficiently familiar to be
used as explanations for the more problematic cases of seeing as if. 28 ,,As the eye passes
over the picture, across the frame, to the wall on which it is placed, it
cannot but become aware, however cunning the painting may be, of a
discrepancy or discontinuity which is fatal to die illusion“ (Wollheim 1963,
p. 25). This assumes, without justification, that unless an illusion is
total, it is not an illusion at all. Imagine an open aperture through which I
could see an The
second type of objection draws our attention to the perceptible distortion to
be noticed even in the most ,,realistic“ paintings: in all but special cases,
the sensitive viewer will see the brush strokes and will be aware, after all,
that what he sees is not ,,very much like“ the real thing.27 The
undoubted presence of interfering and distorting features, even in the most
,,faithful“ of paintings, is, up to a point, not serious for a defender of the
theory that identifies depiction with illusion. Illusions need not be perfect
and we have plenty of experience in genuine perception of discounting
variations in appearance28 and ignoring the effects of imperfections of the
eye (floating specks, effects of myopia, and so on). Once we have learned how
to look through the partially distorting medium of paintings and photographs,
we shall simply see the depicted subjects as if they were really present. But the
real difficulties are concealed in the deceptive phrase, ,,once we have
learned how things look,“ for this concedes in effect that in many cases
there is a sense in which the subject does not Look in the painting as it
would if it were really present behind the canvas-plane—and such large
deviations from ordinary vision cannot be written off by means of the retort
we have just envisaged. If Picasso‘s women are to be seen as women (seen as
if they were women behind the canvas) we shall have to learn a key of
interpretation for which there is no analogue in normal perception. For
instance, we shall have to learn to distinguish between a ,,faithful“
painting of a green face and a green painting of a white face. actual landscape outside: then what happened
as my eye passed over the walls and so on would not prevent me from saying
that I saw the landscape. 27 Cf. die reaction of Roy Campbell on first seeing
snow, having seen only paintings of it before: ,,From paintings I had
imagined it to be like wax, and snow fiakes to be like shavings of candle
grease“ (quoted from Gombrich 1961, p. 221). 28 Cf. the famous
,,constancy phenomenon“ (for which see, for instance, Hoch-berg 1964, p. 50).
We see a poodle—indeed the same poodle, from different angles, at different
distances, and in various lights. Ihen why should we not be able to see the
poodle through whatever distortions are due to the artistic medium and the
artist‘s handling of it? If we can recognize the poodle in moonlight, or even
in a trick mirror, then why not when it is, as it were, seen through a
painting darkly? But once
we allow, as we must, for such prior induction into the ,,tech nique of
representation,“ the equation of depiction with fictive representation or
,,illusion“ loses its attraction. Instead of saying ,,P is a representation of
S because seeing P is, with some reservations for incompleteness and
distortion, like looking at S (seeing P is as if one were seeing S),“ we now
have to say something like: ,,In general, P is a representation of S, if P
looks like S, according to the conventions embodied in the artist‘s style and
technique.“ And now, one wonders how much work the surviving reference to
,,looking as if S were present“ really does. Given a case of extreme
distortion, is it still necessary to say that we see, and are required to
see, something that looks as if it were behind the canvas? Is this not
perhaps only a misleading way of making the obvious point that if we have
learned how Picasso in his cubistic period painted a woman, we shall know
that the painting is of a woman?29 Is anything added by the insistence that
we also ,,see the painting as if“ it were something really there? I am
inclined to think that by the time the theory has been stretched so far it
has degenerated into useless mythology. Finally,
we might notice that the view under examination can be regarded as reducing
,,depicting“ to ,,looking-like.“ For instead of saying that P is seen as if S
were present, one might as well say that P ,,looks like S“ (although we know
that S is not present). Most defenders of an illusion theory have indeed
supposed some view concerning resemblance between a picture and its subject
to be at its foundation. But this deserves separate examination. Depiction
as resemblance. We
have seen that any tenable conception of depiction as involving a sort of
illusion (seeing the painting‘s subject as if it were present) must provide room
for the observable differences, ranging all the way from selection to
outright distortion, between the subject as represented and as it would
appear if actually present. Only when delusion or deception is the
controlling aim of the artist, do we get even an approximation to total
,,imitation.“ Now a favored way of allowing for the element 29 Even in a very
distorted representation, we can sometimes analyze the particular elements
that are operative, picking out one outline or color patch as the face,
another as the arm, and so on (but it does not seem necessary that we should
always be able to do this). Reliance upon such clues, if that is what they
should be regarded as being, does not fit easily into the conception we are
here examining. of
unlikeness in even the most ,,faithful“ visual picture is to invoke a notion of
resemblance: the picture is not conceived now as ,,looking as if“ the subject
were present, but rather as looking as if something like, something
resembling, the subject were present. A typical
statement of this standpoint is the following by Professor Beardsley: ,,,Ihe
design X depicts an object Y‘ means ,X contains some area that is more
similar to the visual appearance of Y‘s than to objects of any other dass.‘
,,30 On a
certain rather simplistic conception of similarity or resemblance (which I
take to be synonyms in the present context), it is easy to launch devastating
objections to any attempt to make resemblance central to the relation of
depiction. For one thing, a photograph or a painting, considered as physical
objects, are really not at all like horses or trees or oceans, and there is
something askew in supposing a ,,design“ to be more ,,similar“ to a tree than
to an ocean. (Cf. asking whether a postage stamp is more similar to a person
than to a piece of cheese.) But if we take Beardsley‘s formula as a careless
way of saying that the look or appearance of the ,,design“ has to be ,,more
similar“ to the look of a tree than to the look of an ocean, we are at once
enmeshed in all the conceptual difficulties that attend any conception of
comparisons between such dubious entities as ,,looks.‘ ,31 We need
not enter upon this controversial range of questions, since the superficial
logical structure of the verb ,to resemble‘ makes any resemblance view
excessively implausible. To take only a single point: we tend to think of the
relation of resemblance or similarity as symmetrical. If A resembles B,
then necessarily B re- 30 Beardsley (1958), p.
270. I do not wish to saddle Beardsley with some version of the depicting-as-illusion view. An
adherent of the conception that the essence of faithful naturalistic
depiction is to be found in some relation of resemblance need not be
committed to any opinions abont the resulting ,,illusion,“ although the two
conceptions fit weIl together. 31 Ihe archetypical
situation of our crudest conceptions of resemblance, as ansing from
comparisons of objects, is that of having the two objects side by side and looking at each in turn, in order to ,,perceive“
relevant resemblances. But dien the ,,comparison“ of appearances would require
a kind of second-order looking at looks. Perhaps this can sometimes be
achieved. However, it seems remote from what happens when we see something in
a painting as a horse. We do not in the mind‘s self-observing eye compare the
look of what we now see with the look that we should see if we were faced
with a horse. To say ,,Ihat looks like a horse“ refers surely to some more
primitive operation. 118 BLACK sembles
A, and both resemble one another.32 However, if we take this seriously, we
shall find ourselves committed to saying that any tree is a representation of
any naturalistic picture of a tree. And since nothing resembles a painting so
much as a reproduction of it, the absurdity lurks dose at hand of identifying
the subject of any picture with its copy.33 Further
objections to a resemblance model. The well-known objections, restated above,
to regarding resemblance as the basis of naturalistic depiction, might well
leave us unsatisfied. We might have an uneasy feeling that appeal to the
surface grammar of ,resemblance‘ is too summary a way of disposing of a
putative insight. To be sure, if we treat resemblance as symmetrical and
transitive, we shall be saddled with paradoxical consequences; but is there
not, after all, we might still think, something to the notion that a naturalistic photograph
,,resembles“ or ,,looks like“34 its subject? And if so, could we not modify
the superficial implications to preserve this insight? If ordinary language
commits us to saying, for instance, that in some sense of resemblance
a painting resembles nothing as much as itself,
is it beyond the wit of man to establish a more appropriate sense of the
crucial expression? lt
will be worth our while to probe more deeply. Our common, simplifying, conception of ,,resemblance“
is controlled, I would like to
suggest, by one or more ,,pictures‘ or 32 lt is
important that these logical features are not always exemplified in the
ordinary language uses of ,look like‘. When A looks like B, B need not look
like A, and the two need not look alike. It is this kind of point that makes
any easy reference to resemblance or similarity (as a substitute for the less
pretentious but more relevant notion of looking-like) so unsatisfactory. 33 Ihis
kind of objection has often been made before. ,,An object resembles itself to
the maximum degree but rarely represents itself; resemblance, unlike representation
[= depiction, in this context] is symmetric.... Plainly, resemblance in any
degree is no sufficient condition for representation“ (Goodman 1968, p. 4). I do not mean to imply that these two
expressions can always replace one another. Indeed, I shall soon argue that
the associated patterns of use show important differences. 35 I am
using ,this word here in somewhat the way that Wittgenstein often did in his
later wnitings. Cf. such a characteristic remark as ,,Ihe picture of a
special atmosphere forced itself upon me“ (Wittgenstein 1953, p. 158).
Wittgenstein‘s notion of a ,,picture“ deserves more attention than it has yet
received. Many of our key words are associated with what might be called
semantic myths, conceptions of exemplary and archetypical cases in which the
use of an expression 119 HOW
DO PICTURES REPRESENT? idealized prototypes of application. Gonsider
the following simple examples of clear cases of ,,resemblance“: 1. A
writer is buying a new supply of typing paper in an unfamiliar shop. He compares a
sheet that is offered him with one from his old and nearly exhausted stock. ,,That is rather like what I want; but that is better; perhaps that resembles what
I need sufficiently closely.“ 2. A
housewife goes to a shop to buy some extra material for a dress she is
making: she compares the material in imagination with what she already has.
,,That looks almost like what I need;
it resembles it very closely; I think it will do.“ 3. A
film producer needs a stand-in for his principal actor in some dangerous
sequence. He compares
the two men, deciding whether the substitute
sufficiently resembies
the star, so that the audience will not detect the
substitution. 4. A
historian compares the careers of Hitler and Stalin for ,,points of resemblance.“ 5. In
trying to sway a judge, an advocate offers a previously decided case as
a precedent, but is met with the objection, ,,I don‘t see sufficient resemblance
between the cases.“ Such
examples and the many others that could easily be produced suggest the
following reflections: a. The
notion of resemblance is closely connected with the notions of comparison and
matching (also with that of similarity, which I shall here ignore). In some
of the cases, but not in others, the ideal limit of the scale of relative
resemblance is that of indistinguishability: if the writer could not tell the new paper apart from the old, he would surely be satisfied, though he would be willing to
accept something less satisfactory. And similarly, mutatis mutandis, for cases 2 and 3. b. In other cases, the degree of
,,resemblance“ in the things - compared turns upon point-to-point correspondences,
so that there seems to be manifested in excelsis. Such an
archetypical situation is not merely a paradigm but, as it were, a paradigm
of paradigms, wherein we think we can grasp the essence of the expression‘s
meaning in a single flash of insight. lt is as if the extraordinary
complexity of the expression‘s actual use were compressed into a dramatic and
memorable fiction. To the extent that we are dominated by such a primeval
myth, we are led to procrustean conceptualization—a cramping, because over-simplified,
conception of the word‘s meaning. (Of course, Wittgenstein has said this far
better.) 120 BLACK is an
observed analogy between the things compared, while indistinguishability is not in question
(cases 4 and 5). c. What
determines the choice of specific criteria of degree of resemblance in a
particular case results from the overarching purpose of the particular
process of matching or analogical comparison: sometimes it is a question of
finding an acceptable surrogate, with respect to appearance, durability, or
other properties; sometimes a matter of finding a justification for applying
general concepts, dicta, maxims, or principles (cases 4 and 5>. In short,
what counts as a sufficient degree of resemblance, and the respects in which
features of resemblance are treated as relevant, is strongly determined by
the overall purpose of the process. To put the point negatively: in the
absence of such a purpose, any proposed process of comparison is
indeterminate and idle. If I am asked to compare A with B, or to say how much
resemblance there is between them, in the absence of any indication of what
the comparison is to be used for, I do not know how to proceed. Of course, if
politeness requires me to make some response, I may invent some purposive context,
trying to assimilate the task to some familiar case, and hence seeking for
points of color resemblance, or similarity of function, or whatever else
ingenuity may suggest. Of the
points I have singled out for emphasis, the first two serve mainly to remind
us of the great variety of procedures that are covered by the umbrella term
,,resemblance“: that compendious label covers a large variety of processes of
matching and analogydrawing, performed in indefinitely many ways and for
indefinitely many purposes, with corresponding variety in what counts as
appropriate and relevant to the comparison procedures. But the third point,
stressing the relativity of comparision with relation to some controlling
purpose, is the crucial one for the present inquiry. lt is quite opposed in
tendency to the picture we have of ,,resemblance“ being constituted by the
sharing of common properties, as if we could decide the question whether one
thing were or were not similar to another in vacuo, without any reference to
the aim of the exercise. (Cf. asking whether A is better than B, which also demands
a comparison, in the absence of further determination of the question‘s
sense.) Let us
now apply these elementary reflections to our prime case of the painting and
its subject. The first obstacle to using 121 HOW DO PICTURES REPRESENT? either
pattern of resemblance (the search for an approximate match or the search for
an analogical structure) is, as we have already seen, that the ,,subject“ is
normally not available for independent scrutiny. When the painting is
,,fictive,“ there can be no question of placing it side-by-side with its
subject in order to check off ,,points of resemblance.“ But let this pass,
though the point is far from trivial: it remains that the determining purpose
of the imputed comparison is left unstated. What is the point of my looking
first at a portrait of Queen Elizabeth and then at the queen herself, in
order to find points of resemblance? There can be no question here of the
painting being a surrogate for the person, as in some of our exemplary cases.
Nor can it be a matter of being able to make corresponding statements about
the two, though that might be the point, if the portrait were to be preserved
in some historical archive to supplement and amplify some verbal description.
We are left with nothing better than the empty formula that the painting
should ,,look like“ the sitter. But that is merely to substitute the
unanalyzed expression ,,looks like“ for our problematic expression
,,resembles.“ Here again, the point about the absence of determination of
purpose is relevant. Given that for some purposes and in some contexts the
most naturalistic trompe l‘oil portrait will look conspicuously unlike a
person, what is to count as ,,looking like“? Whatever merits the resemblance
view might have, it cannot provide answers to these questions. My chief
objection to the resemblance view, then, is that when pursued it turns out to
be uninformative, offering a trivial verbal substitution in place of insight.
(In this respect it is like the view of depiction as the expression of
,,information“ previously discussed.) The objection to saying that some
paintings resemble their subjects is not that they don‘t, but rather that so
little is said when only this has been said. ,,Looking
like“. I have agreed that stress upon ,,resemblance,“ however philosophically
uninformative in the end, does at least serve the useful purpose of reminding
us how the fact that a painting resembles something in the sense of looking
like it may be relevant. lt would be a willful violation of common sense to
say, for instance, that whether a photograph ,,looks like“ a tree, a man, or
whatever the case may be, has nothing to do with its function as a picture.
Certainly a picture may ,,look like“ its subject, but the 122 BLACK problem
is to see whether we can say anything useful about what ,,looking like“
amounts to. So it should be worth our while to ex-amine somewhat more closely
the notions connected with the expression ,,looking like“
or its grammatical variants. Here we
shall immediately find, as in the case of the words connected with
,resemblance‘, that there are paradigmatic uses that need to be
distinguished. So let us begin again with some examples. 1. We are
meeting somebody at the station. Pointing to someone approaching in the
distance, you say, ,,That looks like him.“ 2. On meeting twin brothers, you say
,,Tom does look very much like Henry, doesn‘t he?“ 3. Of a cloud: ,,Look at that: doesn‘t
it look like a bird?“ 4. We might say of a man: ,,He looks
very much like a wolf.“ Ihe first
type of case might be identified as one of seeming. lt can sharply be
distinguished from the others by the possibility of substituting the phrase
,,looks as if,“ with corresponding adjustment in the rest of the utterance.
Thus, in case 1, little if any difference would result from saying ,,That
looks as if it were him.“ Two other grammatical points may be made. If we try to
insert adverbial qualification, as in ,,That looks very much like him,“ we
may justifiably feel that we are shifting to another use: thus to the latter
remark, but not to the original one I have imagined, it might be natural to
reply, ,,I don‘t see the resemblance.“ A connected point is the difficulty of
negating the original remark: If I want to disagree with ,,That looks like
him,“ in the intended use, the best I can do is to say, ,,No, that does not
look like him“ or ,,I don‘t think so“—while ,,No, that looks unlike him“ has,
in context, the feel of playing on words. For present purposes, we may think
of this first use of ,looks like‘ as connected with qualified assertion: the whole
utterance has the force of expressing a weak truth-claim, with the
implication of lack of sufficient and conclusive reason. (Cf. the form ,,That
might be him.“) I note this use only to exclude it from further
consideration, since it obviously has no application to our prime subject:
there is normally no occasion to make qualified assertions about the subject
of a painting or picture.36 38 An exception might be
a case in which we were trying to identify the sitter of some portrait or the
actual scene of some landscape. In such special case we might say ,,lt looks
like Borgia“ or ,,lt looks like Salisbury Plain.“ That would imply the
presence of some visual evidence for the identification in question. 123 HOw DO PIGTURES REPREsENT? The
second type of case is the one already discussed, in which explicit and even
point-by-point matching is present or in the offing. Here, reference to
,,resemblance,“ in uses dose to some that we previously listed, is
appropriate. In the
third case (the cloud ,,looking like a bird“), I should want to argue that
the attempt to assimilate it with full-blooded matching would be a
distortion. For one thing, we seem here to be engaged in some kind of
indirect attribution, rather than in some implicit comparison. For instance,
a supplementary question of the form ,,Like which bird?“ would be rejected as
stupid, unless taken to be a request for further specification of the
attribute (an eagle, rather than merely a bird). Here, it is worth
emphasizing that recourse to ,points of resemblance“ will seem particularly
out of place; indeed, the form of words ,,Look at that cloud: doesn‘t it
resemble a bird?“ will feel like a shift to the previous type of use.37 One
might say that, in certain cases of this type, the speaker is more or less
indirectly describing the situation before hirn. lt is as if, given the task
of describing the cloud in terms of an animal, he were to say ,,If I had to
describe it as some kind of animal, the only one that would fit would be ,a
bird‘.“ There are some obvious analogies here to the use of metaphor, as
contrasted with the use of simile: looking-like in the context of
attribution, rather than comparison, is closer to metaphor than to simile. Finally,
there are cases like the last (,,He looks like a wolf“), where the suggestion
of comparison, which is admittedly still present, is so far suppressed as to
have almost no effect. Saying of a man that ,,He looks like a wolf“ or,
alternatively, ,,He has a wolfish look“ may be a way of recording an
immediate impression, with no thought of being able to specify points of
resemblance— or, in some cases, of being able to specify any ground for
description. In such cases we might be said to be dealing with non-exponible
metaphor or catachresis. On being challenged as to the propriety of our
description, perhaps the best we can say is that it seems to fit—which is, of
course, saying very little. 37 An
obstacle to making this kind of point persuasively is that uses of the words I am discussing are more elastic and variable
tban I may seem to be contending. I do not doubt that ,resembles‘ can sometimes
and without impropriety or ambiguity be used as a contextual synonym for
,looks like‘. Yet‘ if I am not mistaken, the differences in use that I am
trying to emphasize really exist and could be fixed more sharply in a longer and
more laborious investigation. 124 BLACK The chief
moral that I wish to draw from this brief examination of some related uses of
,looks like‘ is that if we had to position in our schema the use of such a
sentence as ,,That looks like a sheep“ (said while pointing to a picture), we
should do well to choose the last of our four types. If one says of a
painting, or part of one, ,,That looks like a man,“ one is normally not
saying that there is partial but incomplete evidence for its being a man
(which would be preposterous), nor that there are exponible points of
resemblance between that patch of painting and a man (which is highly
implausible and, in the absence of any assignable point to the comparison,
idle), nor attributing a property to that patch as one might, by way of
simile, call a cloud bird-like, but rather saying something about that very
thing before us, as we say of a man that he has a wolfish look, intending to
say something directly about him—and not about a certain imputed relation to
wolves. If so, the sense in which a realistic painting ,,looks like“ its
subject still resists analysis. One might even be inclined to say, indeed,
that that expression ought to be avoided, as tending to have misleading
suggestions. If a
child were to ask how one would learn to find out whether a canvas ,,looks
like“ a man, perhaps the best we could say is, ,,Watch a painter at work on
his canvas and then, in the end, perhaps you will really see a man when you look
at the painting.“ But if that is the best that we can say (as I believe), it
looks as if the fruits of our analytical investigation are, after all, very
meagre. A landing
place. I have now completed the task of examining the credentials of
plausible candidates for the role of a necessary condition for the holding of
the relation of ,,depiction.“ I have satisfied myself, and perhaps the reader
also, that none of the criteria examined will supply a necessary condition. Appeal to
the ,,causal history“ of a photograph or a naturalistic painting came to look
like the invocation of contingent factual circumstances that may in fact be
needed for the production of a terminal visual representation, but do not
determine its character as a picture by virtue of logical or linguistic
necessity. By considering extraordinary, but logically possible, cases in
which deviant causal histories might produce pictures indistinguishable from
our paradigms of faithful likenesses, we were able to eliminate appeal to a
causal history of a special kind as a necessary or a sufficient 125 condition.
Ihe same verdict, however disappointing, was all that emerged from our
examination of the other criteria. Reference to the imputed intentions of the
picture‘s producer seemed enmeshed in hopeless circularity, since the very
specification of such an intention required independent specification of what
would count as fulfilling the producer‘s intention. The seductive model of
,,information,“ factitiously borrowing prestige from an irrelevant
mathematical theory, proved a will-o‘-the-wisp, amounting in the end to no
more than a linguistic rechristening of the problematic concept of
,,depiction.“ Finally, reliance upon the attractive notion of ,,resemblance“
between a picture and its ,,subject“ left us, once we had unravelled the
skein of criteria concealed by the deceptive surface unity of the abstract
label of ,,resemblance,“ with nothing more than our original problem, under
the guise of questions as to what it really means to say that a picture
,,looks like“ what it represents, in the crucial cases in which ,,looking
like“ cannot properly be assimilated to point-to-point matching with some
independently given object of comparison. Are we
then left empty-handed? Should we confess that the investigation we undertook
has been a complete failure, with no hope of improvement? Such conclusions
would, in my judgement, be too hasty. For the point needs to be made, and
with emphasis, that the disqualification of some proposed condition as a
necessary and sufficient criterion by no means shows that condition to be
irrelevant to the application of the concept in question. lt would,
for instance, be quite wrong to suppose that knowledge of how photographs are
regularly produced, and of the perceptible changes that occur in the series:
displayed scene, negative, and final positive, have nothing
to do with our ultimate
judgement of the photograph‘s
representative content. On the contrary, our mastery of the skill of
interpreting or ,,reading“ photographs depends essentially upon our schematic
knowledge of how such photographs are in fact normally produced.35 It is
through our knowledge of 38 One might conjecture
that a factor in the alleged inability of members of primitive cultures to
understand photographs on first seeing them may be partly due to such
ignorance of the mode of production. If these bewildered would-be
interpreters were allowed to follow through the stages of production, with a
chance to compare the negative with the external world, they might begin to
have a clue to what Professor Stenins has called the ,,key“ of the relevant
system of representation (cf. Stenius 1960, p. 93, where the useful term
,,key“ is 126 the
photograph‘s provenance that we understand what the photograph ,,shows.“ In
cases of mysterious provenance, as when a layman looks at an X-ray
photograph, the absence of relevant factual knowledge of aetiology
obnubilates comprehension. Indeed, in disputed or ambiguous cases, specific
reference to the circumstances of production may be necessary in order to
determine what the subject is.39 Similar remarks
apply to the currently discredited appeal to the producer‘s intentions.40
Although we cannot define ,,depiction“ or ,,verbal representation“ in terms
of intention without vicious circularity,
it may be altogether proper, indeed sometimes essential, to refer back to
the producer‘s intentions in order to be able to read the very picture in
which his intentions, to the extent that they were successful, were
ultimately embodied. Here, as before, to pretend that we could ever learn to
understand photographs or paintings without repeated
reference to what
photographers and painters were trying to achieve would be unrealistic. Finally, similar points can be made about ,,resemblance“ and ,,looking like.“ Our justified qualms
about the capacity of these to provide defining conditions for the overall
concept must not be allowed to obscure the utility, at times, of relying upon
point to point comparisons or—to jump to something different—to the ,,way the
picture looks“ or simply to ,,what we inescapably see in the picture.“ however used in a somewhat more restricted
sense). For reports of the inability of primitives to ,,understand“
photographs, see for instance Segall et al. (1966), pp. 32—34. Some
interesting examples will be found in Gombrich (1969). We should, for
example, not know what to make of his illustration of the ,,Tracks of an
oyster catcher“ without the accompanying commentary (pp. 35—36)—which
explains, mter alia, that the bird shown was superimposed on a photograph by
an artist. Our knowledge of this unusual causal history materially influences
our ,,reading.“ Gonsider also the cases, discussed by Gombrich in the same
paper, in which we need to ,,interpret“ photographs of deliberately
camouflaged objects (pp. 37 if.). I agree with Gombrich that ,,knowledge, a
well-stocked mmd, is clearly the key to the practice of interpretation“ (p.
37). But I think it is also a key to the mastery of the relevant concept of
interpretation. 40 This is hardly the
place to discuss the so-called ,,Intentional Fallacy“ diat has been memorably
castigated in Wimsatt (1954). Professor Hirsch makes die useful and commonly
overlooked point that Wimsatt (and his collaboratOr Monroe Beardsley)
,,carefully distinguished between three types of intentional evidence,
acknowledging that two of them are proper and admissible“ (Hirsch 1967, p.
11). 127 The proper moral to be
drawn from the initially disconcerting outcome of our
investigation is that the
notion of ,,depicting“ is what has been called a ,,range concept“ or a ,,cluster concept. ,,4‘ The criteria we have
considered—and perhaps others we have overlooked—form a skein, none of them
being separately necessary or sufficient, but each of them relevant in the
sense of potentially counting toward the proper application of the concept of
depiction. In perfectly clear cases, all of the relevant criteria point
together toward the same judgment, whether we rely upon what we know about the method of production, the intentions of the producer, or
the sheer ,,look“ of the picture as it appears to a competent viewer who sufficiently knows the tradition within which the picture is placed.42 A reader who might agree with this kind of moral might still perhaps
wonder why, if ,,depicting“ is properly to be viewed as a ,,range-“ or
,,cluster-concept,“ just these criteria should have been ,,clustered“
together. One answer might be to invite such a questioner to undertake the
Gedankenexperiment of imagining conditions in which the criteria were
disassociated.43 The point of our concept of depiction— the ,,only“ concept
we have—might then become plainer. But such an answer, whatever its
pedagogical merits, is somewhat evasive. There is
something of the first importance lacking from our account, namely all
consideration of the purposes of the activities in the course of which, what
we, in our culture, recognize as ,,pictures“ are produced. And no account of the concept of depicting,
or of the various related concepts bundled together under that label, could be adequate without
some examination of such purposes. This
weakness in our discussion might even be felt in connec 41 For the
general methodology of handling such concepts, see Black (1954>, chapter
2. 42 Hard cases of ,,interpretation“ typically arise when there is a real or
apparent conflict between the defining criteria. When, for instance, we have
firm evidence of the producer‘s intentions and of the means for realizing his
intentions within the tradition to which he adheres, but cannot yet ,,see“
the desired embodiment in die picture itself and do not know whether to blame
the artist or ourselves. 43 For instance, by imagining, in full detail, what
the situation would be in a ,,tribe“ (that
convenient mental construct) whose members were keenly interested in ,,seen“
likenesses, in total disregard of the intentions or modes of productions
responsible for such objects. 128 tion with our account of how photographs depict. Photographs
have been talked about in this essay as if they were objects having no
identifiable uses and consequently no intelligible interest. But we are obviously keenly interested in photographs, and
for a variety of reasons.
If we focus upon one such interest, say that of identifying persons (as in
passport photographs), we shall not find it difficult to see why some of our
criteria harmonize with that purpose. Of course, it is by no means easy to
formulate, with any show of thoroughness, the many purposes that photographs
serve in our culture; and when we pass to the more difficult realm of art
objects, the difficulties multiply. But the moral to be drawn is that clarity
about the basic notion of artistic representation cannot be expected to be
reached by a process of logical analysis alone, however sophisticated in its
apparatus of ,,cluster concepts“ and ,,family resemblances,“
but will call for a
less tidy and more exacting inquiry into the production and appreciation of
art objects within ,,ways of life.“ But this is hardly the place for what is
already too long a discussion.44 44 One of the great merits of Professor Wollheim‘s stimulating little book on
aesthetics (Wollheim 1968) is that he initiates such discussion. |
|
|
|
|
|
|
|
|