When Paintings Argue

Abstract

My thesis is that certain non-verbal paintings such as Picasso's Guernica make (simple) arguments. If this is correct and the arguments are reasonably good, it would indicate one way that non-literary art can be cognitively valuable, since argument can provide the justification needed for knowledge or understanding. The focus is on painting, but my findings seem applicable to comparable visual art forms (a sculpture is also considered). My approach largely consists of identifying pertinent features of viable literary cognitivism and then showing how they or close analogues can be applied to non-verbal painting. The two main features are the requirements, first, that the relevant knowledge is provided significantly in virtue of the distinctive essential feature of literary fictions, i.e., their fictionality, and second, that the knowledge stems primarily from the content of the work, not from what the auditor brings to the work. Some ways that literary fiction has been taken to be argumentative are explained, and striking similarities are found between argumentative literary fiction and argumentative painting. Potential objections are addressed, and I examine a proposed way to express, in a schematized format, both the power of an argumentative painting and its relatively simple associated propositional content.

1. Introduction

My thesis is that certain non-verbal paintings such as Picasso's Guernica make (simple) arguments. If this is correct and the arguments are reasonably good, it would indicate one way that non-literary art can be cognitively valuable, since argument can provide the justification needed for knowledge or understanding. For topic manageability, the focus will be on painting (§4 and §5), but findings seem applicable to comparable visual art forms, notably, sculpture (in §6 an example of sculpture will be considered: Wall Street's erstwhile ‘Fearless Girl facing down Charging Bull’).

My approach will largely consist of identifying pertinent features of viable literary cognitivism (§2) – something that is relatively easier to characterize – and then showing how they or close analogues can be applied to non-verbal painting. So, much of my case has the form of an argument from analogy. Some ways that literary fiction has been taken to be argumentative will be explained (§3), and as we proceed we will see striking similarities between argumentative literary fiction and argumentative painting.

Now it may be that most agree that (non-verbal) paintings can have contents and hold that arguments must involve contents. Thus, it might seem intuitively obvious that paintings and other visual depictions should be able to figure in or even constitute arguments, because contentful things can be used in many ways, e.g., making assertions, suppositions, proposals, and directives. While I do hope that my view conforms to intuition, the case for it would still need to be made. For one thing, there are persistent objections to such a view, those that center on the ideas that, unlike for language, any association of propositional content with the features of a painting is too loose to allow it to be argumentative (§5.1), and that one cannot reliably identify and distinguish premises, conclusions, and illative relations in paintings (§5.2). I also address the concern that even if paintings did make arguments, they would be too simple to be cognitively valuable (§5.3). We will look at a possible way to express, in a schematized format, both the power of an argumentative painting – by including its image in the schema – and its relatively simple associated propositional content. It is ubiquitous in law, science, and ordinary life that images or aspects of an image are taken as providing evidence with respect to an argument. Is it just confusion to think that this means that images themselves can be premises in the argument? I will develop some reasons for thinking that this is not mere confusion.

There is a philosophical tradition that holds that all argumentative justification is ultimately formal deductive validity, and an argument is an ordered pair, where the second member is the conclusion set of propositions, and the first member is the premise set of propositions (possibly infinite, empty, or identical with the conclusion set). Yet this formalist tradition does not own the concept of argument. A common view (CV) more amenable to the approach that I will take agrees that arguments are generally composed of sequences of propositions, but adds, crucially, that an argument is an expression of an inference (of the sincere arguer). Inferences are private, psychological phenomena. In the formalist tradition, an argument need not express anyone's inference. Moreover, according to CV the support relation between premises and conclusion is a matter of degree, whereas in the formalist tradition it is all or nothing. On CV, to identify an argument, one must determine the type or strictness of the illative (consequence) relation intended in the inference, in addition to determining which propositions it relates. Determining the intended illative relation is publicly facilitated by a broad set of indications, including the use of modal terms (such as ‘must’, ‘probably’, ‘possibly’), context, and arguer behavior. In the formalist tradition it is something of a mystery how propositions become ordered or collected into an argument, particularly if bad arguments are countenanced (otherwise, the collection principle is deductive validity according to some system). Perhaps most importantly for present purposes, on CV arguments are creations; they come into existence at a certain time (when they are first an expression of a reflective inference) and have histories, as seems obvious (e.g.) with Anselm's Ontological Argument or Searle's Chinese Room argument. Considered as a sequence of propositions, which I take to include the content of the illative relation that generates the sequence, an argument is the kind of abstract object that, unlike a concrete object, may be in different places (i.e., expressed therein) at the same time, in the manner of (e.g.) musical pieces and games.¹

Although such issues as the role of diagrams in mathematical proof have a long history of study (Dove, 2002; Larvor, 2013), the modern general study of visual argument began in earnest only in the mid-1990s. Most attention since then has been devoted to partially verbal media on the order of ads, posters, and cartoons, rather than non-verbal, classic art forms like painting (Kjeldsen, 2015; Groarke, 2016). Here I attempt to address this lacuna but am not suggesting that my results are restricted to painting. Besides examining an example of sculpture, photography and film are briefly considered.

2. Two Requirements of Literary Cognitivism

‘Literary’ is a term of approbation. It is generally held that literary fiction is more nuanced than non-literary fiction; it has a greater richness and complexity of such things as character development, plot, or fine description, and also somehow shows insight into human affairs. How it might show or facilitate such insight is the central question of literary cognitivism. Literary cognitivism is usually² defined as the view that literary works can be a source of knowledge or understanding – a definition that is itself pretty vacuous since of course there is science in science fiction, history in historical novels, etc. Literary cognitivists and anti-cognitivists are all concerned with fictional literature because, tautologically, nonfictional literary works (e.g., in history or biography) may yield knowledge. So a critical requirement of literary cognitivism should be that the relevant knowledge is provided significantly in virtue of the distinctive essential feature of literary fictions, that is, their fictionality. Let us call this the fictionality requirement.³

A second requirement is that the knowledge stems primarily from the content of the work, not from the auditor or what the auditor brings to the work. This is because, as Gibson suggests (2008, p. 575), something can be learned from anything if we auditors are clever enough, e.g., what we have learned about climate change from variations in glaciation, which hardly constitutes a reason to believe in ‘glacial cognitivism’. ‘Cognitivism is, again, about what goes on in artworks and not in the mind of the consumer about art (except in a secondary, derivative sense)’ (pp. 584–85; cf. Fasnacht, 2023, esp. p. 5). However, although it is ‘secondary’, we will see that this is not meant to diminish the proper role of critical interpretation – our imaginative and appreciative engagement with artworks – in bringing about cognitive gain. As Gibson also says (2006, p. 444):

Rather than directed at the recovery of linguistic meaning, critical interpretation marks a process of articulating patterns of salience, value, and significance in the worlds literary works bring to view. That is, critical interpretation marks the moment of our engagement with the world of the work, and it has as its goal the attempt to bring to light what we find of consequence in this world.

Still, the controlling factor is the world of the work. Following Gibson (2009, §II), let us call the second requirement of literary cognitivism the textual constraint.

These two requirements are not necessarily exhaustive of requirements of viable literary cognitivism, but they are important. Moreover, meeting these two requirements does not by itself show that a literary fiction is argumentative. A non-argumentative way in which fictional literature may satisfy both requirements is the notion that fiction provides a necessary ‘safe zone’ in which to deal with striking or upsetting ideas, such as implications of senseless murder in Dostoevsky's Crime and Punishment (cf. Hakemulder, 2000, pp. 11, 150). As one might say, you can't learn if you're scared; rather, you flee the theater. Analogously, consider contemplating Guernica, which you might want to view many times, versus a purely mimetic (undoctored, un-Photoshopped) photograph of a similar scene of horror, which you might not want to view at all. Generally in contrast to reality, art allows the auditor choice in how to assimilate it, as in the case of viewing Guernica or choosing to read a novel lightly or deeply.

So what, more precisely, do the two requirements of literary cognitivism have to do with being argumentative? I will later contend (with appropriate qualification) that a necessary condition for literary narration to be argumentative is that it be fictional; otherwise, the creativity that is needed to construct an argument and express a point of view would be absent or too restricted. We will see that for the same reason an argumentative painting cannot be flatly realistic (in the general art-critical sense such that it contrasts to surrealism and abstractionism, notably). Also, argumentative painting and narration in general are related in that each depicts process or events unfolding. Furthermore, if the argument made by the work – a literary fiction or a painting – is reasonably complete, the work ipso facto meets the textual constraint or an analogue thereof.

3. How Literary Fiction and Painting Might Argue

Normally, the notion that certain literary fictions, taken as wholes, are argumentative appears as the claim that they are ‘thought experiments’ (e.g., Carroll, 2002; Swirski, 2007; Green, 2010; Mikkonen, 2013). But there are inappropriate connotations of this concept that are perhaps best revealed by considering the inverse relationship between parameters of assessment. Factors that make a thought experiment good (e.g., straightforwardness and precision, convincingness) tend to make a story bad (lack of nuance and subtlety, didacticism), and vice versa (cf. Egan, 2016, p. 147). We can, and possibly should, consider suppositional reasoning in connection with fictional literature without dressing it up as thought experiment. In a literary fiction, suppositional reasoning (or any kind of reasoning) generally can be exhibited only indirectly, that is, within the context of critical interpretation, for otherwise, the work would be overtly didactic or polemical, which undermines its status as literary fiction and makes it akin to philosophy or science. With this understood, one may take a literary fiction as – supposing P, Q – where P is the work's fictional ‘premise’, and Q is the set of consequences inferred in the work, which at least constitute conditional (on P) knowledge if the reasoning is good, thereby evidently satisfying both requirements of (viable) literary cognitivism above. For instance, Greene's The Third Man can be taken as constructing a supposed counterexample to the generalization that ‘when loyalty to a friend conflicts with loyalty to a cause, one ought to choose in favor of the friend’ (Carroll, 2002, p. 10).

This way of understanding a literary fiction as argumentative involves a substantial interpretive load. Another way of taking a literary fiction as globally argumentative, viz., as an argument from analogy (e.g., Hunt, 2009), appears to involve a much greater interpretive load insofar as, for indirectness, the auditor must fill in the second (target) case of the analogy. For example, which of far too many actual totalitarian states do you fill in for Orwell's Animal Farm? Thus, construed as an argument from analogy, it is dubious that the novel could satisfy the textual constraint of literary cognitivism, even though we are given to understand that Orwell himself intended that the specific target case for Animal Farm was the Bolshevik Revolution and the Soviet Union under Stalin (represented by the boar Napoleon). There is no necessity at all that the reader will know Orwell's intention, let alone complete the argument with that particular target case. If Orwell himself had explicitly completed the analogy, then (per above) the work would have been less a literary fiction than a philosophical tract, although presumably the textual constraint would have been met.

It seems that Animal Farm, construed as an argument from analogy, is a sufficiently clear case of not satisfying the textual constraint of literary cognitivism. But certainly, there are borderline cases for both this constraint and the fictionality requirement. I do not pretend to have rules for deciding such matters, yet with sufficiently clear or paradigm cases, I think adequate progress can be made for present purposes without dealing extensively with interpretative theory, or the theory of fictionality or its painting analogue (I claim) concerning non-realistic art.

So let us see. Each of the three examples of argumentative painting that I propose below exhibits the structure of what is called an ‘argument from negative consequences’ (Walton, Reed, & Macagno, 2008, pp. 101, 332) against a practice, with the form: practice p has undesirable consequences c_1–c_n; thus, p is bad. This is a normative, defeasible, and simplified style of modus-tollens reasoning. It is worth comparing such reasoning to, for instance, the global argument that Green (2010, p. 360) finds in Huxley's Brave New World:

1. 1 Suppose a society were organized along the lines dictated by hedonistic utilitarianism.

2. 2 In such a world, people would lack freedom of thought, freedom of expression, and the ability to cultivate the capacities for critical reflection on their surroundings.

3. 3 Therefore, in such a world, life would be intolerable to all but those who have lost the capacity for the activities mentioned in premise (2).

4. 4 Therefore, such a world would be unacceptable.

5. 5 Therefore, hedonistic utilitarianism is an incorrect theory of how to achieve happiness.

This is suppositional reasoning in the guise of a fairly loose reductio ad absurdum, and so exhibits a modus-tollens style as well. While the power (affective and otherwise) of an argumentative painting might be comparable to that of Huxley's novel (§5.3 below), I believe that you will not find in any non-verbal painting an argument that even approaches this complexity and sophistication (although perhaps you could in a non-verbal film). For one thing, Green treats statement 1 in his rendering of Huxley as a ‘counterfactual’ supposition and as thereby meeting his version of the fictionality requirement (he defines ‘Literary Cognitivism’ as the thesis that ‘literary fiction can be a source of knowledge in a way that depends crucially on its being fictional’ – p. 352). The examples of argumentative painting that I will propose do not in this manner satisfy an analogue of the fictionality requirement; rather, we will see that they do so by not being flatly realistic depictions, although the depictions represent actual practices and their actual undesirable consequences. Considering these possible cases of argument from negative consequences may make the topic cleaner and more manageable, and the thesis more convincing, but there may be other simple argument patterns that non-verbal paintings exhibit. Perhaps a painting could exhibit relatively unsophisticated reasoning from a counterfactual supposition, as for instance, given that Hell is a fiction, in the third panel of Bosch's The Garden of Earthly Delights, where the consequences of going to Hell are represented.

The difference between taking Huxley's novel to be about hedonistic utilitarianism and Orwell's to be about the Soviet Union is essentially the difference between the logical operation of generalizing from particulars (the situations in Huxley's novel) and making a comparison between the particulars of the given case (the animal society in Orwell's novel) and the particulars of one among many possible, but not given, cases (that of the Soviet Union). In this way, in Gibson's terms, the former is in ‘the world of the work’, whereas the latter is not.

4. Three Proposed Examples of Argumentative Painting

Picasso named his Guernica (Figure 1) after the town in northern Spain that was bombed by the Nazis in 1937. Aside from its title, which gives interpretive orientation, Guernica is entirely non-verbal. Nevertheless, it narratively depicts events unfolding. Its reasonably obvious message or conclusion lies along the lines that indiscriminate bombing (many of the figures face skyward) is evil, and the evidence or reason presented is the consequent destruction, suffering, and death hauntingly depicted (in more propositional terms, I think the argument is: horrible destruction, suffering, and death are inflicted by indiscriminate bombing; thus, indiscriminate bombing is evil). If this evidence and what it evokes are sufficient, knowledge of the conclusion and of the consequences of indiscriminate bombing are derivable from the work, and since it is a Cubist-Surrealist departure from pure mimesis, which allows the argument to be constructed, it seems that analogues of the textual constraint and the fictionality requirement of literary cognitivism are satisfied.

Figure 1 Pablo Picasso: Guernica, 1937. Oil on canvas, Museo Nacional Centro de Arte Reina Sofía, Madrid [photo from Museo Reina Sofia © Succession Picasso, VEGAP, 2017].

These are the basic ideas. Let me offer some explanation. There is of course interplay between the need for critical interpretation and the need to satisfy a pictorial analogue of the textual constraint. Fasnacht (2023, p. 5) nicely describes this:

… for the understanding of the story, the image needs a recipient, and we can make certain assumptions about who should count as a recipient…. The recipient further needs to be aware of certain conventions of visual language, like the concept of speech- and thought-bubbles or other conventions such as a blurry frame when a scene is meant to be a flashback or dream. This can be understood as pictorial literacy….The recipients need to interpolate and close gaps, but they always have to do so in relation to the image, not by means of free association, so that one could dispute the correct reading by pointing to elements depicted by the image (like facial expressions, body language, even the use of colour, perspective, composition, shapes, light, etc.).

Regarding my theses that the relevant analogue of fictionality in literature is non-realism in painting and that these are generally required, respectively, for argumentativeness, the main reason concerns whether the aim of the work involves veracity. Certainly, the aim of Guernica is not veracity; it is not to accurately portray the visual appearance of what happened in the manner of ‘illusionistic’ art (the illusion being that one can hardly tell the difference between seeing the painting and seeing the scene itself, i.e., what I am calling ‘pure mimesis’). Rather, Guernica is strongly expressive or stylized in technique, and its subject matter is in various places fanciful – so the painting is non-realistic in that respect as well. Correspondingly, because it is fictional, even a classic work of literary realism such as Madame Bovary does not strive for veracity in what it represents, as if it were history or biography, although its verisimilitude figures in its categorizing as literary realism. While it is realistic in both technique and subject matter, it is still fictional.

Pure mimesis in a painting, as in an unmanipulated photograph, would restrict the creativity that is needed to construct an argument and express a point of view. Similarly, for literary narration to be argumentative, it is generally required that it be fictional – that it not strictly be, e.g., history. This is not because history, biography, etc. need be any less vivid than fictional narration (the chain of thought is not: ‘vivid, therefore persuasive, therefore an argument’). Rather, it is because there is a huge theoretical obstacle standing in the way of regarding a nonfictional narration, or a realistic painting, as argumentative: As indicated, the point of nonfictional narration and realistic painting involves veracity – sticking to the facts, telling what happened or how things are – so there is no theoretical room for the creativity that is needed to construct an (indirect) argument by significantly inventing what happens or massaging reality.⁴ This is not to deny, as I have indicated (§1), that unmanipulated images are frequently taken as providing evidence with respect to an argument. But it is an issue (to be addressed in §5.3) whether this means that such images themselves can be premises, and my point now is that their unmanipulated character prevents them from expressing (an entire) argument in the manner of Guernica.

Of course, however, to some extent this theory is an idealization subject to qualification, seemingly like any theory concerning art. For example, works commonly classified as literary history or biography, etc. may make augments that are imbedded in the narrative, or they might even express an overall argument, by doing so directly, unlike literary fiction or painting. Accordingly, no substantial critical interpretation would be necessary, or at least none beyond what is typical for philosophy or science. Still, such works may also to some degree invent what happens, and to that degree, they could be indirectly argumentative. Moreover, reality is multifaceted and immensely complex. Thus, a realistic painting or a nonfictional narration that captures reality will capture only selected portions of it, and such selection can require creativity. And no doubt there will be borderline cases in determining what is to count as pure mimesis for a photograph or a painting and how great a departure from it is enough to permit indirect argumentation. It seems, for instance, that Dorothea Lange's celebrated depression-era photograph Migrant Mother is not by itself argumentative, yet it does contribute to the argument about unjust deprivation made by her series of ‘social realist’ photographs for the U. S. Farm Security Administration. All the more reason, then, to stick to what appear to be paradigm cases.

An indication that Guernica actually presents an argument is that its content is contestable: presumably (e.g.) the authors of the Allied bombing campaign against Germany would dispute its conclusion, as well as the sufficiency of the evidence – including its proscriptive nature. So it is possible that Guernica's argument is a bad argument by usual standards of argument cogency. Certainly, it seems to be a simple argument (more on this in §5.3). Relatedly, it is sometimes claimed, e.g., by Fleming, that images cannot express negation (1996, pp. 17–18) or contradiction (Champagne & Pietarinen, 2020, p. 224n32), and being able to do so is a necessary condition for being argumentative. But even Fleming considers the case of an image of a cigarette on which a red circle with a diagonal line is superimposed. He says this is ‘a visual sign reliably translated as a verbal proscription’; it is ‘merely a visual substitute for the verbal utterance “No smoking”.’ Yet it is not clear why this matters, why it is not a counterexample to Fleming's own claim.⁵ And, pardon the thought, suppose some of the faces in Guernica expressed approval. This would surely make the painting incoherent, like anything self-contradictory.

An issue in the study of visual argument has been whether such an argument need be, or even can be, entirely non-verbal in that there is no written language at all in the image (e.g., Kjeldsen, 2015, p. 124; Groarke, 2019, p. 335). If it is thought that the Picasso example is somehow cheating because it has a revealing name, look at Figure 2, a case of possibly untitled⁶ street art by NemOs painted on the side of a building. At the risk of belaboring the obvious, the message or conclusion of this surreal scene is that endlessly turning trees into building structures is bad – the structures are defecated by an unattractive and stylized, humanoid tree-eater – and the evidence presented is the nondescript, jumbled, and ever-rising nature of the pile of buildings depicted (stated more propositionally, I think the argument is: the process of endlessly turning trees into buildings is bad since it defiles the environment). Granted, Figure 2 is not as (the painting analogue for) ‘literary’ as the Picasso – it is not ‘high art’ – yet the same kinds of considerations indicating that analogues of both requirements of literary cognitivism are satisfied by Guernica also apply to Figure 2.

Figure 2 Street art by NemOs, 2010. Mural on building, Milan. Used with permission [https://www.whoisnemos.com/walls/edifeci, accessed 22 March 2024].

Contrast Goya's The Third of May 1808 (Figure 3). As compared to the other two paintings, this is realistic art; there is little or no surrealism here. As relatively realistic, the scope for the creativity that is needed to construct an argument is restricted. Correspondingly, it would be difficult to make the case that any analogue of the fictionality requirement of literary cognitivism is satisfied. Blair (1996) says about the Goya that it ‘portrays human cruelty, fear, terror, hopelessness and courage; but it gives no reasons for favoring the loyalists or opposing Napoleon’,⁷ yet I think he wrongly lumps it together with the Picasso: ‘What conclusion are we to draw? That this was a terrible, cruel, destructive act? But that is what Picasso's painting expresses; there is no argument’ (pp. 27–28).

Figure 3 Francisco Goya: The Third of May 1808, 1814. Oil on canvas, Museo Nacional del Prado, Madrid [Wikimedia Commons, https://commons.wikimedia.org/w/index.php?curid=6342196, accessed 22 March 2024].

The arguments, stated propositionally, that I have suggested that the Picasso and the NemOs express are pretty vague and general. Take the NemOs: The process of endlessly turning trees into buildings is bad since it defiles the environment. As Green's statement of the global argument he sees in Huxley's Brave New World illustrates, the argument of a literary fiction may be more specific (and Green's statement could have been even more specific). Perhaps the amount of such specificity that is appropriate is roughly proportional to the amount of artistic material, assuming here that paintings and novels are commensurate in this respect. So for example, Shannon Brick (personal correspondence) wonders whether the premises of the NemOs argument include ‘that we are creating things that we don't need; that we are creating things that are ugly and commercial; we are making things and not caring about them’. This content seems to involve specifics that are insufficiently grounded in the painting's features. Here I am, in Fasnacht's words, disputing ‘the correct reading by pointing to elements depicted by the image’.

Who argues, the painter or the painting? The preceding discussion indicates that it is primarily the painting. ‘Absolute actual intentionalism claims that the actual intentions of the author are the only determining factor in the meaning of a work’, whereas extreme anti-intentionalism holds that ‘there is no reason to add an arbitrary component like the intentions of the artist in order to form an interpretation’ (Hulbert, 2021, pp. 236–37). No doubt the more defensible position lies somewhere between these extremes (as Hulbert's survey argues). To be sure, there is no need to investigate the artist's intentions to see the argument embodied in the NemOs, assuming it is viewed from the perspective of our current Western socio-cultural context. Yet even as viewed before the Industrial Revolution, the argument would still be apparent just by looking, although the painting might engender considerable puzzlement (why is endlessly turning trees into building structures not good?). In contrast, Picasso's intentionally naming his painting Guernica vastly narrows down the field of possible interpretations.

These issues are vividly illustrated by my final proposed example of an argumentative painting, van Gogh's Head of a Skeleton with a Burning Cigarette (Figure 4). (It is not clear how the painting acquired this name, but in any case, the name offers little or no interpretive information that is not already obvious in the image itself.) Within our current Western socio-cultural context and the conventions that form part of that fabric, this surreal painting makes an argument against smoking (cigarettes), a minimal propositional statement of which is: smoking can kill you, so smoking is bad. Indeed, our conventions seem to force this interpretation. On the other hand, many think that van Gogh did not intend such a message. Indeed, the website of the Van Gogh Museum in Amsterdam, where the 1886 painting is housed, says that it ‘is a juvenile joke’. This plugs into the interpretation whereby van Gogh was making fun of the common practice of using skeletons rather than live models (which came later) to help develop an artist's anatomical knowledge. Antwerp's Royal Academy of Fine Arts, where van Gogh was studying at the time, was no exception. Accordingly, van Gogh was trying, well, to breathe a little life and movement into the skeletal form by adding the burning cigarette. The cigarette represents life, whereas from a contemporary perspective, it is a harbinger of death. This should be no more surprising than that a word can acquire, over time, a meaning opposite to the original (e.g., ‘awful’).

Figure 4 Vincent van Gogh: Head of a Skeleton with a Burning Cigarette, 1886. Oil on canvas, Van Gogh Museum, Amsterdam [Wikimedia Commons, https://commons.wikimedia.org/wiki/File:Kop_van_een_skelet_met_brandende_sigaret_-_s0083V1962_-_Van_Gogh_Museum.jpg, accessed 22 March 2024].

Thus, it seems that from van Gogh's point of view there is no argument, whereas from our point of view there is: in such a case, the painting's argument expresses an inference of the audience, not the creator of the painting.

5. Objections and Replies

5.1 Composition

Perhaps the most common point of contention revolves around the widely accepted idea that arguments are sequences of propositions, and since paintings are not such sequences, how could a painting be an argument? But, just as for a literary fiction, I am not claiming that a painting could be an argument, only that it could make, express, or embody an argument indirectly, i.e., through critical interpretation. It seems that a sentence or series of sentences cannot be an argument either; rather, their meaning or associated propositional content is the argument.

Accordingly, the issue may reduce to determining how loosely propositional content is associated with a painting such as Guernica. Sentences would appear to have a huge advantage in that they are tightly governed by semantical rules, have propositional syntax and argument-indicator terms, etc. However, it seems that resemblance relationships between aspects of (non-verbal) images and their objects, as well as the conventions governing the cognitive processing of images, can be clear or tight enough that it is not an interpretive free-for-all, allowing the cognitive import to lie primarily in the content of the image or painting, not in what the auditor brings to it. For instance, consider traditional pictograph writing systems, icons, emojis, the ‘No smoking’ sign, Venn diagrams, the representation of justice by an image of a blindfolded woman holding scales and that of death by the image of a hooded person carrying a scythe. Certainly, it is unclear that the interpretive load involved in discerning an argument in such paintings as Figures 1, 2, and 4 is greater than that involved in discerning a global argument in a literary fiction, if you put the works on par in terms of their cognitive complexity. On the other hand, the more abstract a painting is, the greater the interpretive load. In the case of Suprematism, for example, such as Malevich's Black Square, the load is extreme, and correspondingly, it is dubious that any analogue of the textual constraint of literary cognitivism could be satisfied. Similarly, it is dubious for at least some ‘experimental literature’ that the textual constraint could be satisfied.

There is an enormous amount of research (mostly recent and often technical) on the syntax, semantics, and pragmatics of non-verbal pictures and the conventions governing their cognitive processing, far too much to summarize here. For example, in the vein that ‘truth is not the exclusive property of language’ (Hyman, 2021, p. 498), a proposed ‘T(ruth)-schema for [realistic] pictures’ is ‘A picture p is true with respect to a possible world w iff there is a viewpoint v in w from where w resembles p’ (Schlöder & Altshuler, 2023, p. 701). Still, the ‘consensus’ view is that ‘in languages, a pattern of letters can, in principle, stand for anything. Not so for pictures’ and their ‘patterns of color, light, and dark’ (Kulvicki, 2021, p. 314). In any case, the modern general study of visual argument has mostly occurred not in such research but in argumentation research.

5.2 Internal Structure

A concern is that such images as Figures 1, 2, and 4 lack ‘the requisite internal differentiation’ that would ‘reliably’ permit distinguishing premises from conclusion; this distinction, which ‘is at the heart of argument…is thus collapsed’ (Fleming, 1996, p. 13). Blair says that ‘this is the main difficulty in interpreting’ Guernica ‘as an argument’ (1996, p. 28). Champagne & Pietarinen (2020, p. 232) see the problem as involving a dilemma: ‘if the conclusion is present in the image, then the visual argument risks begging the question [because it cannot be distinguished from the premises]; but if the conclusion is absent from the image, then the visual argument risks supporting any conclusion’. An example they give of the dilemma's second horn is a Nazi propaganda postcard from the early 1930s that shows the Sun on the horizon emblazed with a swastika, and a harvested wheat field in the foreground. Groarke (2015, p. 148) says that the image suggests ‘that a Nazi future will bring food and abundance’, leading to an implicit conclusion that ‘You should vote Nazi’. Yet Champagne & Pietarinen object that ‘there is no evidential basis in the image itself to infer that the sun-like Nazi emblem is rising as opposed to setting’ (p. 215).

These are reasonable concerns, but they seem overblown or overgeneralized. Take the NemOs. It is just not that hard to ‘reliably’ identify the premise material and distinguish it from the equally identifiable conclusion material. How could the conclusion not be along the lines that endlessly processing trees into buildings is bad, since in the image the buildings are defecated by a repulsive creature? Certainly, not just ‘any conclusion’ is expressed, even though this is a simple normative conclusion. The premise material or evidence presented is, as I mentioned, the nondescript, jumbled, and ever-rising nature of the pile of buildings depicted.

Nevertheless, one might wonder what in the NemOs corresponds to an argument-indicator or illative term (‘since’, therefore’, etc.), as in an argument verbally expressed. One kind of response is that of course the NemOs is in some respects enthymematic, as are most purely verbal arguments. For the NemOs, the illative relation is enthymematic, as is, to some degree, the notion that the situation indicated by the evidence is bad. No doubt one can often appeal to various dimensions of context to help fill in such enthymematic blanks, as (e.g.) Groarke, et al. argue (2016, pp. 220–21). A plausible addition is proposed by Champagne & Pietarinen. Since ‘argumentation is a sequential activity’ (2020, p. 229n40), their main idea, inspired by Peirce, is that ‘illation – the distinctive transition from premise(s) to conclusion … involves a growth of signs’ (p. 230). They discuss Alfred Wegener's landmark 1929 map, which he used as evidence in abductively concluding that Africa and South America were once part of a supercontinent but subsequently were subject to ‘continental drift’. In the map, the South Atlantic Ocean has been vastly narrowed, and one can see simply by looking at the map that the coastlines easily fit together. According to Champagne & Pietarinen, what is key is that ‘the components of the map [the coastlines] need to be moved in order to establish the relevant conclusion’, moved, that is, backwards in time from their current positions (p. 227).

A similar analysis can be applied to Figures 1, 2, and 4, although each expresses an argument from negative consequences against a practice (not abduction). The NemOs depicts the process of endlessly turning trees into buildings, and going from premise material or evidence (the nondescript, jumbled, and ever-rising nature of the pile of buildings depicted) to the conclusion (that this process is bad) involves going back in time through the process. The van Gogh indicates that the habit of smoking cigarettes can kill you, and going from this premise to the implied normative conclusion that smoking cigarettes is bad involves going backwards in time from death to smoking occasions. Guernica is more enthymematic or relies on its name, but here too, proceeding from premise material (the destruction, suffering, and death depicted) to the conclusion (indiscriminate bombing is evil) involves moving back in time through the events. The purely static and (hence) non-argumentative character of such a work as Malevich's Black Square stands in sharp contrast and confirms the Peirce-Champagne-Pietarinen theory. In the Black Square there is not so much as a hint of narrative depiction of events unfolding.

Besides, and by now you probably anticipate this point, there are many who interpret certain literary fictions, taken as wholes, as argumentative, usually as thought experiments. It does not seem any easier to identify and distinguish premises, conclusions, and illative relations in literary fictions than it is in paintings. After all, literary fiction cannot be suppositional reasoning, or any kind of reasoning, in a straightforward way; if it were, it would be overtly didactic or polemical. It generally can be argumentative only indirectly or implicitly.

5.3 Too Simple?

Are the arguments made by Figures 1, 2, and 4 too simple to be cognitively valuable? The short answer is no, because (like so many things), it depends on context, and the context these paintings provide is rich.

The propositional content associated with the paintings does appear to be simple, but that is not the whole story about cognitive value and the role argument plays. Another part of the story is that these paintings have the affective power to evoke an emotional response in the viewer, such as dismay and horror (the Picasso), disgust and shame (the NemOs), and repulsion and fear (the van Gogh). While this affective power might itself be sufficient to move one to adopt the conclusion beliefs, that does not mean that the role of the simple arguments disappears; it might only mean that these beliefs are overdetermined. In addition, there may be a deeper way that emotion and argument are connected here. A kind of position that is commonly advocated is that ‘emotions’ have a ‘cognitive dimension’ in that they ‘embody some of our most deeply rooted views about what has importance, views that could easily be lost from sight during sophisticated intellectual reasoning’ (Nussbaum, 1990, p. 42; see Todd, 2014 for a survey). Given that this is correct, the arguments of the paintings are accorded a cognitive boost – they become more genuinely plausible than they otherwise would be – by conforming to appropriate emotions.

Moreover, if there is anything that everyone agrees on about our topic, it is that images generally are ‘thick’ representations with greater meaning than ‘thin’ representations such as sentences (although no doubt the force of sentences can add up). As Kjeldsen contends, a thick representation ‘in an instant, can provide a full sense of an actual situation and an embedded narrative connected to certain lines of reasoning’ (2016, pp. 267, 279; cf. Hyman, 2021, p. 520). Trying to grasp Guernica's meaning by only considering its associated propositional content is something like trying to appreciate a musical piece by considering its written score in isolation (no sounds imagined). MacGregor (2022, p. 41) says ‘no work of art adequately conveys the horror of war. Not The Iliad, not War and Peace, not All Quiet on the Western Front. Maybe Picasso's Guernica comes closest in our own time’. If this is right, then even if knowledge furnished by Guernica is minimal, the painting provides the means for significant understanding of the evil of indiscriminate bombing and is in that way cognitively valuable. Perhaps not coincidentally, there has lately been a shift in attempts to establish the cognitive value of literature away from emphasizing propositional knowledge to emphasizing understanding (documented by Mikkonen, 2015; 2021).

It would probably be a stretch to hold that the understanding occasioned by the NemOs or the van Gogh is on par with the Picasso. Nevertheless, there are other avenues of cognitive gain. For each of these paintings, its power plus the simple argument it makes may function to forcefully remind the viewer of forgotten aspects of the issue that the painting raises, or be an effective catalyst for pondering new, related aspects. These aspects could be propositional elements of more sophisticated arguments. Given an analogue of the textual constraint, the relevant cognitive value contributed by the painting here is not constituted by such propositions, but is rather the value of being a potent reminder or catalyst for further reasoning.

Perhaps surprisingly, Stolnitz's (in)famous paper ‘On the Cognitive Triviality of Art’ (1992) says some similar things about Austen's novel Pride and Prejudice: ‘Once we divest ourselves of the diverse, singular forces at work in its psychological field, as we must, in getting from the fiction to the truth, the latter must seem, and is, distressingly impoverished’, namely, ‘Stubborn pride and ignorant prejudice keep attractive people apart’ (pp. 193–94). The points of similarity are, first, that the propositional content associated with a painting such as Guernica is simple, perhaps ‘distressingly’ so. Second, the power of Guernica lies in its surrealistic expressiveness, like the power of Austen's novel lies in ‘the diverse, singular forces at work in its psychological field’, which belong to its fictionality.

Is there a way to express the power of an argumentative image and express the associated propositional content in a schematized format? Consider a simple argument that Groarke (2019, p. 347) represents as in Figure 5. In describing the occasion of the argument, Groarke says (pp. 345–47):

my wife suggests that we should go to see Neuschwanstein Castle (the famous Bavarian castle built by King Ludwig II) on a trip to Germany. When I question her suggestion, she tries to convince me that it is something worth seeing by showing me the photograph [in Figure 5]. In doing so she provides me with a reason for concluding that we should visit Neuschwanstein, though she does so visually – appropriately so given that the issue at question is what we should go to see…Arguments about the castle which are wholly verbal can describe what it looks like, possibly in poignant ways. But they cannot provide the detail we see in the photograph and do not show us what the castle looks like.

Figure 5 A visual argument's Key Component table and diagram. Used with permission [PDF view of the file chapter13-groarke.pdf (scholarsportal.info), accessed 22 March 2024].

Groarke develops the ‘Key Components’-plus-argument-diagram format in ways that permit the representation of far more complex arguments with visual components; e.g., there is symbolic notation for implicit premises and for linked premises, and ‘ostensive’ reference to a visual when a picture is impractical. Also, a conclusion can appear as an image in the format. Given that the image of Pinocchio and his long nose has basically the ‘fixed meaning’ of liar, the Michael Ramirez political cartoon with thirteen written premises (e.g., ‘the private sector is doing fine’) connected to a long nose on Obama accused him of being a liar (Groarke, et al., 2016, pp. 221–22). In Groarke's schematic representation of the argument, a thumbnail image of a Pinocchio-like Obama head appears as the conclusion, and the ‘Explanation’ entry is ‘Visual (the extended nose indicating that Obama is a liar)’ (2019, p. 364).

For Guernica, a thumbnail of the painting could appear in the ‘Key Components’ column as a premise, and the basic associated proposition content that horrible destruction, suffering, and death are inflicted by indiscriminate bombing could be expressed in the ‘Explanation’ column. Below that, an implicit-conclusion row would say that indiscriminate bombing is evil. A similar adaptation of Groarke's approach could be applied to the NemOs and the van Gogh. Groarke's schematized format gives an affirmative answer to what may be the most difficult question in the study of visual argumentation: can images themselves, and not just propositional content associated with them, be premises and conclusions in arguments?

Although this particular issue cannot be addressed adequately in the present paper, there are several points in favor of Groarke's answer that are worth mentioning. Plainly, at least in the evidentiary use, images or aspects of images are frequently related to arguments in ordinary life (‘seeing is believing’), law, science, etc. Consider, for instance, medical imaging and visual abductive reasoning in archaeology and geology. Are pictures in the same class as other fundamental empirical evidence such as ‘sense data’ and perceptions, which can help to ground premises in arguments but not be premises themselves? Much appears to depend on context. If you know the conclusion I wish to draw, my simply presenting you with a picture and indicating that I rest my case might be enough to have an argument. This is what Groarke's wife does in the castle example. She certainly seems to be providing a reason to visit the castle, as Groarke says. Moreover, Gregory (2020, p. 165) contends it is dubious that one could

construct indexical-free [hence, no demonstrating or pointing to an image] linguistic expressions whose contents are so strongly bound to vision that they identify ways for things to look simply in terms of the subjective character that is shared by all and only the possible instances of those ways for things to look… But if nonindexical language and conceptual contents more generally are indeed limited in that manner, any propositions that are built upon distinctively visual contents will not be expressible using indexical-free linguistic means. The notion of a proposition would then not really be essentially linguistic, even though its arrival within philosophical thought first derived from reflection upon language.

If this were the case, the additional force or meaning that images contribute to Groarke-like schemata would actually be propositional (though not expressible in nonindexical language), which would undermine the most common objection to holding that an image can be a premise or conclusion. Finally, the comparable issue in the philosophy of mathematics about whether a visual can be a component of an argument might be instructive. Although the mainstream view since the early 20^th century has been the ‘formalist ideology’ that diagrams are inessential to mathematical proof (Johansen & Pallavicin, 2022), this may be changing. Criticism of formalism has recently been propounded by, e.g., Dove (2002), Larvor (2013), and De Toffoli (2021; 2023). Larvor even says (p. 247) about a diagrammatic proof in classical geometry that ‘we could render this proof into prose and deliver it as speech, but it would be pretty well impossible to follow. Anyone who could follow it would do so by creating and manipulating mental images, that is, by re-creating and acting on the diagram in imagination’.

Irrespective of all of this, the arguments made by Figures 1, 2, and 4 are simple in that they contain no hint of a dialectic, as in an extended argument that raises and answers possible objections. As various authors contend (e.g., Alcolea-Banegas, 2009; Champagne & Pietarinen, 2020), such complexity can be achieved in film – ‘moving pictures’ – as would be expected, given that (§5.2 above) the argumentativeness of a painting is tied to its indicating motion or process. Another caveat is that with the power of an argumentative image, there is risk – of (e.g.) being bamboozled – as by the Nazi postcard. The situation is the same with fictional literature; see, for example, Green (2016) ‘Learning To Be Good (or Bad) in (or Through) Literature’ or Goffin & Friend (2022) ‘Learning Implicit Biases from Fiction’.

6. An Example of Argumentative Sculpture

Regardless of its checkered corporate history, it is hard to view the sculpture in Figure 6 and not think that an argument is being made. (Not long after commissioning Fearless Girl, State Street Global Advisors agreed to pay $5,000,000 to settle claims that it had engaged in sexual and racial pay discrimination!) Located in Bowling Green Park on Broadway in New York City, the two statues faced each other for over a year and a half beginning in March 2017, until Fearless Girl was removed and now stands in front of the New York Stock Exchange. The popularity of the duo amounted to a public safety hazard because pedestrians spilled into the streets, and Charging Bull's creator complained that the stare-down reflected negatively on his work [e.g., https://www.nytimes.com/2018/04/19/nyregion/fearless-girl-wall-street-bull-statue-move.html].

Figure 6 Wall Street's ‘Fearless Girl facing down Charging Bull’, 2017. Bronze castings [photo 121308408 © David Makharashvili | Dreamstime.com, ‘editorial’ use purchased 10 March 2024].

Within our cultural context, the charging bull is the optimistic icon of established (male-dominated) financial markets, and the girl, given her youth and stance, represents women as upcoming. From the point of view of the girl, the argument (propositionally) is to the effect ‘I am now already strong enough to face you down, so in the future may overpower you’. The meanings of the two individual statues that constitute the (temporary) sculpture are conventionally entrenched (in the case of the charging bull) or naturally understood (in the case of the girl's stance), and putting the two together reliably yields a meaning along the lines I have indicated (judging by widespread commentary); thereby, an analogue of the textual constraint of literary cognitivism is satisfied. The satisfaction of an analogue of the fictionality requirement is a little less obvious, since each of the two statues is realistically executed. But with a little thought one can see that it is their face-to-face juxtaposition that is non-realistically imaginative.

7. Conclusion

Even given that Figures 1, 2, and 4 make arguments, it could be that most paintings do not. Many paintings are too abstract or too realistic for a case taking the approach I have taken (using literary cognitivism as a guide) to get off the ground. If the painting is too abstract, an analogue of the textual constraint is not satisfied. If the painting is too realistic, an analogue of the fictionality requirement is not satisfied.⁸ (Parallel remarks apply to sculpture.) This is not at all to imply that non-argumentative paintings lack cognitive value. Non-argumentative ways that a painting can have cognitive value include knowledge by acquaintance of peculiar colors and shapes, as well as phenomenal knowledge of what objects, states of affairs, and even emotions look like – allowing one to imagine what an experience or emotion feels like, thereby enabling empathy. The Goya above nicely illustrates this.

Perhaps this is the biggest disanalogy that the preceding suggests between painting and literary fiction: a literary fiction is more likely to be (globally) argumentative in some way. I would think that this makes sense because literary fiction is verbal and tends to involve more artistic material. Nevertheless, the similarities between argumentative painting and argumentative literary fiction are remarkable. Argumentative painting must narratively depict process or events unfolding. Insofar as creativity is needed to construct an argument and express a point of view, it is generally a requirement for literary narration to be argumentative that it be fictional, and likewise, an argumentative painting cannot be flatly realistic; veracity cannot be the point, though verisimilitude might be. After adjusting for cognitive complexity, the interpretive load involved in discerning an argument in a painting may be on par with that involved discerning a global argument in a literary fiction. It hardly seems easier to identify and distinguish premises, conclusions, and illative relations in literary fictions than it is in paintings. Apart from meaning imparted by their surrealism or fictionality, respectively, the verbal gist of an argumentative painting or literary fiction may equally tend to be ‘distressingly impoverished’. The cognitive value of both may lie more in imparting understanding than knowledge. Finally, the artistic power of argumentative painting, for good or ill, may be comparable to that of argumentative literary fiction.