II. Thinking Aloud

How should we conceive of the inseparability of freedom of expression and thought when considering the scope of a free speech principle? Here is a simple way to interpret this relation, as a starting point. You think a thought, fully, to completion, in the privacy of your mind, then you choose whether to verbalize it, and thus make it available to others. This picture of thought and language looms large in philosophy of mind. Consider, for instance, Richard Kimberly Heck’s sketch of what is termed the Naïve View of Communication:

The thought—in this instance, a belief—fully exists prior to the speaker’s attempt at communication. The content of the thought is externalized, intact, via an utterance. Finally, through an act of comprehending uptake, a listener re-internalizes the thought (again, intact), and thereby comes to entertain the same thought the speaker began with.

If this naïve view is true, why would restrictions on speech practically restrict possibilities for thought? One answer is that under speech-restrictive social conditions, it gets harder to persuade, compel, object, contend, agree and so on. When we are able to do fewer things with our words, this in turn has a limiting effect on which thoughts we are inclined to think. This seems like a good reason, prima facie, to suppose that legal restrictions on speech acts have a practically limiting effect on which thoughts arise in a community of thinkers.

But although this seems reasonable enough, it only takes us so far. What it fails to capture is how speech restrictions can practically impinge upon the development of a thought in itself, independently of whether and how it gets shared with others. Seana Shiffrin’s thinker-based theory of free speech goes beyond the naïve view in explaining this connection. Some of our incipient thoughts are too complicated, or too elusive, to be fully grasped via introspection. In these cases, verbalization enables us to realize the incipient thought. To paraphrase Shiffrin, speaking and writing help the thinker to externalize the relevant bits of her mind and get some observational distance on them in order to view them clearly, then reflectively affirm or amend them. In speaking or writing (or signing), the thinker gets a better grasp on what is actually there, albeit not yet fully realized, in the thoughts they are trying to think.Footnote ¹³

The achievement facilitated by this expressive process, in cases where mental content is not readily introspectively graspable, has two parts: (1) a realization of mental content; and (2) a facilitation of the thinker’s apprehension of that content. Let’s call the expressive process that aims at these achievements ‘thinking aloud’. If thinking aloud is a commonplace feature of human cognition then, plausibly, the right to speak one’s mind in public, protected against the threat of sanction for expressing views to which other people object, will generally be conducive to the realization and reflexive knowledge of people’s mental states.Footnote ¹⁴

Part of what makes this story attractive, with respect to thinking aloud in language is language’s impressive representational power. For any thought you might wish to realize or grasp, natural languages provide terms for referring to the entities, actions, processes, events, or qualities you have in mind, as well as ways of qualifying, quantifying, negating, or conditionalizing your statements. When we are trying to verbalize incipient thoughts, we may feel as if we are taking indistinct parts of our consciousness, which somehow want to be made distinct or tangible, then helping them become what they want to be. Languages provide rich and complex arrays of representational resources, which in principle seem capable of capturing almost any thinkable ideas. So it is natural that verbal expression often seems like the best tool for thinking aloud.

The question for our purposes, though, is not whether language is often the best option, but whether it always is. We don’t believe it is. That is because—and here is the crucial premise—accurately expressing a thought doesn’t always mean expressing it in a precisifying medium. While many thoughts are best realized by verbal thinking aloud, there are other thoughts whose accurate realization has to preserve a wordless fuzziness. Verbal expression isn’t ideally suited to that task, for the same reason that it is well suited to many other expressive tasks: because of its great compositional articulacy. For thoughts that are of their essence loose, imprecise, protean, or impressionistic, the process of realization and apprehension by way of external expression, as portrayed above—that is, the process of thinking aloud—can sometimes be better achieved in a nonverbal expressive medium.

It is useful here to consider a vivid description of an instance of nonverbal expression, which conveys a sense of what it is like to express a nebulous thought in a non-precisifying medium. Jenefer Robinson’s work—in which she aims to account for the role of emotions in the creation and enjoyment of the arts—is helpful to this end. Robinson wants to say that musical compositions can capture emotion just as effectively as, and sometimes more effectively than, attempts to capture emotion in words. In defending this claim, she offers us a sketch of what this process involves in the example of Shostakovich’s Tenth Symphony, a piece that, on Robinson’s interpretation, expresses an ineffable sense of hope:

In Robinson’s view, Shostakovich is using instrumental music to “articulate a specific cognitively complex emotion … roughly describable as ‘hope’.”Footnote ¹⁶ The term “articulation,” in Robinson’s description, is crucial. There are manifest differences in how words and music articulate the content of our minds. Whereas the sentence “hope can endure, even in bad times” conveys a truth-apt proposition about how the world is, Shostakovich’s expression of hope at most gestures toward a related feeling. But Shostakovich isn’t thereby getting something wrong. He isn’t using music to say—just in a fuzzier, stranger way—that “hope can endure.” Rather, he is expressing a hope-related idea that in its very essence is more impressionistic than this statement. His musical expression is an appropriately vague “capturing” of the protean mental content that is occasioned in his expressive activity. (Naturally, we won’t be able to use words, here, to precisely restate what that content is!)

In a similar vein, consider the thoughts that Picasso was trying to capture when he picked up his palette to produce Guernica. Imagine that, in trying to make his thoughts tangible, Picasso had uttered some descriptive remarks instead. What might have resulted from this? “The suffering wrought by the bombing was terrible.” “It was frenzied; there were screams and mutilated bodies.” To say that these remarks don’t capture what is conveyed in Guernica is an understatement verging on a category mistake. The only words that are remotely up to the task are words that themselves advert to the inadequacy of the verbal—something like “the bombing was an unspeakable atrocity.” The power of Picasso’s image is due to its success in conveying something of the unspeakable. When someone is trying to express such thoughts, language’s clarity and overtness can become liabilities.Footnote ¹⁷

This example shouldn’t be seen as suggesting that it is the unspeakability-qua-enormity of certain thoughts that makes language an unfit medium for expressing them. What is true for Guernica’s horror applies to the Tenth Symphony’s hopefulness as well. It isn’t morally problematic to try to put the hopefulness of Shostakovich’s composition into words. The issue is just that this would involve some infidelity to the protean content conveyed by Shostakovich’s Tenth. If you have in mind a thought of hopefulness that is more impressionistic than ordinary hope-talk would suggest, there may not be any better‚ more faithful—medium for you to try to capture this content in than a musical composition that, with its impressionistic qualities, expresses this hopefulness with its abiding traits intact.

Our interim contention is that the accurate expression of a nebulous mental state is one that respects its nebulousness. So words aren’t always the best tool for the job of thinking aloud, because for these thoughts words aren’t a very suitable tool for the job. Musical and pictorial formats are generally more suitable media in which for thinkers to externalize their thoughts, in these cases in order to get distance on their own minds (to paraphrase Shiffrin again), and thus to better—that is to say, more faithfully—realize what is present, incipiently, in the thoughts they are trying to think. In some cases, to capture what you are thinking, you need all the expressive dexterity that language affords. But in other cases, you would do better to limit your palette and paint in broader brush strokes.

The Shiffrin-style account of free speech’s foundations, which we are adopting and building on, says, roughly, that the forms of expressive activity to be included in the domain of free speech are those that facilitate thinking aloud. To be clear, this account isn’t meant to rule out appeals to other values or ideals that may be invoked to justify the protection of different forms of expression. Familiar normative theories of free speech—theories that see free speech as integral to the promotion of protection of democracy, say, or knowledge, or individual autonomy—are not meant to be invalidated by our Shiffrinian claims about the significance of thinking aloud. But, as Shiffrin rightly observes, these other normative theories of free speech seem to be premised on a common, normatively inflected conception of the person, and of our fundamental interests as persons. What underwrites and unifies speech’s importance, in relation to democracy, knowledge, and individual autonomy, is that speech facilitates the processes of self-understanding and communication upon which all of these goods rely, in different ways.Footnote ¹⁸ What we have been trying to do in this part is to explain why certain nebulous thoughts are most amenable to thinking aloud in nonverbal media, and thus to vindicate Shiffrin’s claim that nonverbal expression can facilitate these same processes, despite nonverbal expression lacking the distinctive affordances that seem to underpin language’s special utility in facilitating these processes.Footnote ¹⁹

III. Language as the Cut-off

In certain instances, thinking aloud—the process of realizing and apprehending thoughts by expressing them—is better carried out via music or visual imagery. And this gives us the kernel of a justification for including (some) nonverbal expressive acts in the scope of free speech coverage. The notion that there is a practical inseparability between people’s ability to think freely and people’s freedom to think aloud can be understood such that this relation pertains in principle to both verbal and nonverbal expressive media.

But isn’t this interim conclusion denying a plausible thesis about the relationship between language and thought? Isn’t verbalization a more fundamental or potent way of expressing thought? Isn’t it appropriate, then, to regard language as a cut-off point, when theoretically delimiting the range of media in which people are free to think aloud? As Peter Carruthers says, it is a commonsense notion that “inner verbalization is constitutive of our thinking … that we think by talking to ourselves in inner speech.”Footnote ²⁰ If this is correct—if we already think in language—then there must be something special about the verbalization of thought.Footnote ²¹ Whether or not any content gets externalized, a thinker should be able to segue between introspection and expression with no change to the content. On this picture, competence in a given language just is, as Michael Devitt says, “the ability to translate back and forth between mental representations and sounds of the language.”Footnote ²² What should our account of thinking aloud say to this challenge?

Granted, someone who sees linguistic thinking aloud as special may allow that non-linguistic expression can be roughly similar to it. Maybe advanced musical prowess involves an ability to translate back-and-forth between mental sentences and some related—associatively rich or affectively insightful—musical expressions, which have a complex connection to the original thought. But still, if thought is essentially linguistically structured, any nonverbal expression of thought must involve some kind of transformation—some reformatting of the thought’s native syntactic form. Any nonverbal thinking aloud would, at best, result in a reformatted variant of the original mental content, rather than a faithful external facsimile, of the kind that verbalization can produce. Again, this would make verbal expression a sensible cut-off point, when delimiting the range of expressive media we see as vehicles for thinking aloud, and which qualify for free speech coverage on that basis. On this view, other nonverbal expression would always be in some sense inferior to thinking aloud verbally. And this would make our justification for privileging nonverbal expression more tenuous than what we want it to be. (We will return to this issue in Section IV.)

But is it right that verbalization is a more format-preserving means for expressing thought than any other expressive format? And if so, in what sense is this true, exactly?

It is not clear how common sense or introspection can answer this question. People’s introspective sense of whether their thoughts are linguistically formatted is more variable than Carruthers appears to allow.Footnote ²³ Partly for this reason, most philosophers of mind have relied on a family of abductive arguments (most associated with Jerry Fodor) to establish the thesis we are entertaining.Footnote ²⁴ Roughly, the argument is that (1) thought has an identifiable set of functional properties, (2) these same functional properties are characteristic of language, and (3) the best explanation of this likeness is that thought has a linguistic format. It would lead us too far afield to fully unpack and address this argument, but we want to join ranks with others who find it to be subtly question-begging. We cannot assert a likeness that links the functional properties of language to the properties of thinking per se unless we have some independent reason to suppose that thought innately possesses the specific complex of capacities that a system of language contains. And we don’t see why there is any good reason to suppose this unless one has already assumed the thesis whose truth is being debated and taken the relevant supposition to be implied by that thesis.Footnote ²⁵

Rather than delving into that tangled web, we will offer another more straightforward argument against the idea that verbalization is the default, format-preserving way of expressing thought. As per our claims in Section III, it seems highly plausible that thinking aloud in words—for thoughts which are elusive or complex, and hard to introspect—does offer cognitive assistance to the speaker in realizing or grasping the content of their mind. But if thought is linguistically formatted—if what one does, in verbalizing a thought, is just to parlay mental states into an external vehicle, without any fundamental change to the structure or content of what gets externalized—then it is hard to understand why these benefits should obtain. The point we are making here is the flipside of our point above, about a thinker easily moving between the private realm of the mind and the public realm of linguistic articulation. If it really is as easy as this, then why should external verbalization be an effective way of developing an incipient thought, or coming to apprehend its content?

By contrast, the conception of thought and expression that we have been proposing—which denies that thought is essentially linguistically formatted—better accommodates the notion that thinking aloud helps people realize and apprehend their (initially) difficult-to-grasp thoughts. Verbal thinking aloud, whether alone or in conversation, in fact isn’t just a cosmetic process that publicizes some already verbally formatted mental stuff. Rather, thinking aloud is a way—often a highly effective way—of actualizing the inchoate content of the thought one is expressing. It is, itself, an active, constitutive process of thinking—that is, of imparting tangible form to an inchoate mental state. The point of thinking aloud is to realize this state by formatting it into an expressive vehicle that reveals one’s anticipatory sense of the content that was waiting, latent, in the state. And this process of realization is essentially the same, we are claiming, in expressive image-making and music-making. The difference is just that these nonverbal media provide different expressive affordances, which are apt to express inchoate content that words would be at risk of unduly precisifying. Nevertheless, thinking aloud in any of these formats is the same expressive process of turning incipient mental content into externalized representations that make it tangible.

What about a simpler version of the worry? Maybe there is something special about the verbal expression of thought, not because speech and thought have essentially congruent formats but because language as a representational system has expressive features that make it more flexible for conveying thought than other expressive media.

There is something in this claim that it would be absurd to deny. If you are trying to offer a theoretical argument, or a detailed account of your family history, or some comments on your friend’s draft of a novel, there are normally good reasons to express your thoughts in words instead of turning them into an image or melody. Taken to a caricatural extreme, it might seem as if we are saying that the choice to verbalize a thought, instead of expressing it in an image or melody, is about as consequential—which is to say, not very—as a bilingual person’s choice about whether to utter a sentence in German rather than English.

However, the account of thinking aloud that we are defending can evade these reductiones ad absurdum. We see it as uncontroversial that in many cases where a thinker has inchoate mental content they want to realize or apprehend through thinking aloud, a verbal expression will strike them—and normally will correctly strike them—as the germane option. Abstract imagery usually isn’t a useful way to realize one’s thoughts in offering an apology, or making a philosophical argument, or formalizing an institutional code of conduct. Even thoughts about subtle emotions, whose ineffability we may casually pay lip service to, can end up better expressed if we forge ahead and try to verbalize them than if we turn to music or imagery. At any rate, that seems true for most people, most of the time.

Language is a powerful medium for the expression of thought. We can use language to say anything we like, near-enough, and to say it with a distinctive kind of precision, especially in how we qualify or caveat our ideas. Language is genuinely distinctive among expressive formats, and it naturally recommends itself as an expressive medium for thinking aloud, for most of the thoughts that most of us have. But this is consistent with the main point we are pressing. When what you want to realize or grasp is a bit of mental content that is too nebulous to be faithfully rendered in words, language’s special features are precisely what makes it an unfit tool for the job. Such cases may be the exception, not the norm. But in these cases it makes good sense to think aloud in a nonverbal medium.

IV. Functional Similarity Arguments

Our overall aim in the previous section has been to explain why we are justified in including musical and pictorial expression in the scope of free speech coverage, even though they don’t share language’s full distinctive suite of representational features. But there is arguably a more straightforward justification available than the one we have offered. We could include nonverbal expressive acts in the scope of free speech coverage if and when these acts are functionally similar to verbal expression, in respect of how they convey viewpoints.

Paradigmatic forms of protected speech—political dissent, journalism, scholarship, protest marches, religious expression, literature, street preaching—are all activities that publicize viewpoints, and thus help to realize the epistemic, democratic, and autonomy-based ideals that underpin free speech norms. Music and images should fall within the scope of speech, according to this account—if and when they too publicize viewpoints. Much like a crossed-out swastika expresses a fairly clear viewpoint, an artistic image can be a way of publicizing an idea or attitude. Picasso’s Guernica and Andres Serrano’s Piss Christ express viewpoints on war and faith. Granted, they admit of multiple interpretations, as befits their status as artworks rather than unambiguous statements. But they aren’t merely ornamental entities. They present a take on things. And so do some pieces of instrumental music. Sometimes this is due to conventional associations—for example, when a piece acquires a religious or national meaning. In other cases the significations are less overt—for example, the anti-establishment or cosmopolitan ethos attributed to some instances of avant-garde composition, and the resultant suppression of that music in authoritarian regimes.

In short, nonverbal creative expression can be functionally similar to paradigmatic speech by virtue of its capacity to express viewpoints. Different authors in the free speech literature provide various points of emphasis in advancing versions of this functional similarity thesis. For instance, Randall Bezanson says nonverbal arts play an important role in sensitizing us to the elusive, borderline-ineffable viewpoints that ordinary descriptive language cannot easily capture. And this is of great value, Bezanson says, because human individuality and creativity are nourished by engaging with these kinds of viewpoints:

Extrapolating a little, we might say that such images are functionally similar to unusually mind-expanding poetic or philosophical texts. They invite us to wrestle with ideas that stretch the limits of our cognitive powers, thereby spurring us to think for ourselves.

Another way of explicating the functional similarity thesis is to highlight the emotion-capturing power of nonverbal media. We can usually verbalize our emotions well enough, but sometimes we find that words don’t do them justice, and that imagistic or musical expression does better. In a reconstruction of Mill’s view about the place of the arts in free speech, Rafael Cejudo argues that this—art’s ability to “provide irreplaceable knowledge about the emotional dimension of human life”—is the key to the arts’ importance.Footnote ²⁷

Mark Tushnet also places the emphasis here in his account of free speech and non-representational art. He cites (approvingly) Justice Harlan’s remarks in the U.S. Supreme Court case Cohen v. California (1971), about how a jacket emblazoned with the words “Fuck the Draft: says more than one which uses the word “Abolish” in place of “Fuck.” “Much linguistic expression serves a dual communicative function,” Harlan says; “it conveys not only ideas capable of relatively precise, detached explication, but otherwise inexpressible emotions as well.” Tushnet says that non-representational art can have a similar function to words that bear this duality. Like a well-placed swearword, a Jackson Pollock canvas captures emotions that would be blunted if one conveyed them in ordinary language.Footnote ²⁸

Another way of explicating a functional similarity thesis, due to Alan Chen, focuses on instrumental music, and how it can express a pro-attitude towards culturally mediated values. In some instances, this is due to formal properties—for example, in cultures that produce music with distinctive rhythms, timbres, or instrumentation, which can become synecdoches for those cultures or their values.Footnote ²⁹ In other cases, conventional associations emerge, linking cultural values and particular melodies. Whether it is rhythm, instrumentation, or melody at work, suppression of music that bears culturally resonant features can be tantamount to an attack on the relevant culture or its values. In these cases, with respect to a functional similarity thesis, compositions may be likened to verbal slogans, insofar as they encapsulate shared viewpoints that groups can rally around, and insofar as their suppression is tantamount to a suppression of those viewpoints. Indeed, as Chen observes, music’s ability to stir emotion in a way that cuts across language barriers potentially makes it more powerful than slogans in “connecting people within and between different communities.”Footnote ³⁰ Music can elicit a powerful feeling of affinity and shared values, even among people whose lack of a shared language would tend to thwart a slogan’s power.

Why not settle for these kinds of justifications? Our chief concern is that they don’t encompass enough of the expression we want to include in the protected sphere. They only explain the significance of a few instances of music and imagery—those that are expressive in a way that is functionally akin to certain kinds of verbal expression, such as avant-garde poetry, culturally resonant slogans, or well-timed swearing. But there is a lot of artistic imagery that doesn’t tick this functional similarity box, and the majority of instrumental music doesn’t seem to do so either. Bezanson, Tushnet, and Chen are highlighting relatively unusual cases of musical or pictorial expression—cases where these things are unusually similar to verbal expression in their expressive functionality. It is good to reflect on these cases because they reveal something about the communicative potential of the nonverbal. But they aren’t an ideal guide to what is ordinarily occurring (and what is ordinarily valuable) in nonverbal creative expression. The fact that a Pollock canvas can do something functionally akin to a deft bit of swearing (if it is a fact) doesn’t mean the same thing is true for most non-representational art. Likewise, the fact that some musical compositions stand in a slogan-like relationship to cultural values (if it is a fact) doesn’t mean that other compositions do something similar.

One might try to quash this concern by noting that for most types of expressive activity that we include in free speech’s scope of coverage, many token instances of the activity fail to fulfil the valuable functions in relation to which that type of activity is deemed important. For example, while journalism generates important epistemic and democratic goods, most token instances of journalism make a nugatory contribution to these goods. Similarly, religious proselytizing is a morally significant exercise of autonomy, and it has a role to play in facilitating other people’s autonomy as well. But most token instances of religious proselytization are of little value in these respects. Nevertheless, we include journalism and religious proselytization in the protected realm of free speech because (roughly) the valuable token instances of them are very valuable indeed, and because if we tried to implement our protections at the level of token rather than type, the relevant authorities would not be trustworthy or competent enough to identify and protect the protection-worthy tokens.

Thus, one might argue, what goes for journalism or religious proselytization should go for art and music. Although most art and music doesn’t function in a way that is akin to verbal expression, it is normal for broad classes of expressive acts to be covered by free speech for the sake of good-making features that only obtain in some token instances of the class.Footnote ³¹ So a functional similarity justification can—indirectly—justify covering all of the pictorial and musical expression whose coverage we were setting out to justify.

The problem is that this leaves us with an unsatisfactory story about why we are including the non-viewpoint-expressive images and music that get scooped-up into our strategically expanded net of coverage. We are invited to see non-viewpoint-expressive instances of nonverbal expression as being valuable in the same way that bad journalism and proselytization are valuable: as inferior token instances of a type of expressive act whose superior instances are valuable. And this doesn’t seem right. What we want to say is that music and artworks have a power to capture something expressively significant, even in cases that bear no functional resemblance to ordinary speech’s viewpoint-conveying capabilities.

Imagine two artists making non-representational paintings. One of them, Krasner, makes images that can be functionally likened to a well-timed piece of swearing, as in the Pollock/“Fuck the Draft” example above. Her images have the sort of expressive duality that Tushnet was adverting to. They express a kind of viewpoint—some idea of individual defiance, say—while also conveying an ineffable mood, which complements the viewpoint-like content but resists propositionalization in its own right. Krasner’s work thus seems to tick the functional similarity box. By contrast, the other artist, Thomas, produces works that don’t seem to carry even a pseudo-message, and thus don’t seem to tick the functional similarity box. Some viewers may try to read viewpoints into the work, but the consensus take, over time, for critics and audiences (and for Thomas herself, let’s say) is that her work just isn’t viewpoint-expressive. It is interested in color, or optical effects, or what have you.

Now, someone might want to rank the expressive value of these two artists’ oeuvres, prizing Krasner’s work (by analogy, as if it were a meticulously researched bit of investigative journalism by George Monbiot) and devaluing Thomas’s work (by analogy, as if it were a feeble, soapbox-y David Brooks think-piece). This person might still want Thomas’s oeuvre to be included in the scope of free speech, alongside Krasner’s, in the same way as they want second-rate journalism to be covered alongside high-quality journalism. This is just how category-based principles of free speech work, they might say. But even if this does end up covering Thomas’s work, it does so on grounds that should strike us as deficient. Thomas’s artistic work is not expressively valuable because of its typological affinity with other work whose social function is similar to the function of particular types of verbal expression. Thomas’s work is expressively valuable because of its own expressive qualities. It is a window of sorts into Thomas’s mind—the very contents of her mind that words cannot capture.

In sum, nonverbal expressive media, like visual art and instrumental music, have significant, non-derivative value in expressing the content of people’s minds, even in cases where they don’t functionally emulate linguistic expression.Footnote ³² Theories of free speech should recognise this non-derivative value when questions arise about why these forms of expression fall within the ambit of free speech. This is what makes our account of thinking aloud in Sections II and III a better way of justifying free speech coverage for musical or pictorial expression, compared with the functional similarity justifications discussed in this section. The most faithful expression of some bit of protean mental content doesn’t always involve its linguistic articulation. For some inherently nebulous bits of mental content, images or music enable more accurate expression precisely because they offer expressive affordances that are more nebulousness-friendly than the precisifying affordances of language.

V. Implications and Caveats

What are the implications of all this for law and policy? Why would our proposed shift in justificatory strategy—from an approach that says “include music and images in free speech because they do things that are functionally akin to literal speech” to an approach that says “cover these things because they’re windows into the contents of people’s minds”—make a difference in how free speech principles translate into policy?

The first upshot is with respect to whether music and images are even under consideration for free speech protections in the first place. While many modern liberal societies do accord free speech protection to these forms of expression, as Tushnet, Chen, and Blocher rightly observe, “calls for control and regulation of instrumental music”—appealing, for example, to music’s alleged degeneracy, carnality, or depravity—“have spanned millennia and have emerged from all parts of the world, from both government entities and other powerful institutions.”Footnote ³³ The same is true of calls for the restriction of provocative, obscene, or otherwise morally controversial visual art. The fears of distinctive dangers in nonverbal creative expression reflect a perennial Platonic anxiety about the power of such expression to manipulate or subvert people’s rational capacities.Footnote ³⁴ It’s a familiar theme in free speech theory that free speech honors and nurtures people’s capacity to reason, insofar as it secures access to other people’s ideas and opinions, and hence to an important source of rational stimuli.Footnote ³⁵ If Platonic anxieties are taken seriously, then protections for nonverbal arts may be perceived as a threat to these desiderata. It’s nice for the painter or composer to get to project their ineffable mental stuff into public spaces. But if the effect of this, for audiences, is somehow rationally subversive, then the artist’s interest in thinking aloud may well be outweighed by the audience’s interest in avoiding that subversive impact.

Functional similarity justifications for including nonverbal expression in the scope of free speech are meant to help us push back against these Platonic worries. But they also go some way towards (inadvertently) validating them. If we say, “protect music and artistic images because they do things that are functionally akin to literal speech,” we reinforce the notion that literal speech’s ordinary functionality is the yardstick for assessing the significance and value of all expressions. We invite a reductively rationalistic construal of the communicative interests that free speech serves, thus undermining the case for extending free speech coverage to nonverbal artistic media that, rather than verbally explicating viewpoints, operate in a communicative register of impressionistic gesture, or expressive effusion.

Our proposal avoids this. It offers an alternative perspective, both on the conception of the person that underpins our free speech principles and on how nonverbal expressive media should be seen as supporting that person’s cognitive and communicative interests. In covering nonverbal music and art in the scope of free speech because they are media that capture the nebulous contents of people’s minds—irrespective of whether they emulate literal speech’s affordances and functions—we are construing the communicative interests that free speech serves in a more pluralistic fashion. What matters most for beings like us is that we have windows into each other’s minds. It doesn’t matter if some of what we find in each other’s minds is hard to put into words, or to subject to verbally mediated debate and inquiry. The goal is to vividly encounter other people’s thinking, in all its complexity, including its hazy or impressionistic dimensions.Footnote ³⁶ Under this strategy, we avoid the trap of arguing for the privileging of nonverbal expression using appeals that tacitly disparage some of the qualities of that same expression, and thus undermine its privileged status.

The second practical implication of this shift in our justificatory strategy is with respect to which token instances of music and imagery qualify as beneficiaries of these privileges. In jurisdictions where music and artworks enjoy free speech coverage, in principle there will still be a process of determining which music and art merit protection in practice, and with what degree of stringency. In certain jurisdictions, some of the protections for art and music operate via the special treatment of creative expression under the application of other policy instruments. Such protections operate, first, by certain creative works being identified as worthy of special treatment, and second, via judgments about the relative weight of artistic ideals relative to countervailing ideals. In addition to the aesthetic heuristics that may inform those judgments—for example, “Is this work beautiful?”—another natural heuristic is to ask whether creative works are valuable qua acts of communication. If our rationale for including music and images in free speech is that they do things that are functionally similar to literal speech, we are again encouraging a narrow understanding of what communicatively valuable nonverbal expression involves, and of what entitles it to a privileged legal status. We are inadvertently licensing a familiar species of philistinism: “this picture (or musical composition) shouldn’t benefit from special artistic privileges, because apart from it not being beautiful, it doesn’t even have anything to say.”

Our alternative justificatory strategy—“protect nonverbal expression as a window into people’s minds”—reframes the question for the better. It isn’t necessarily wrong to assess aesthetic or communicative qualities when deciding which creative works enjoy a privileged legal status, but these judgements must avoid conflating communicative utility per se, with the transmission of propositionally structured information. Our story about the privilege-worthiness of music and images makes a difference here, since it suggests another heuristic for us to adopt to guide these judgments while discouraging the use of heuristics that lead arts-protective policies to withhold protection from instances of nonverbal expression that lack language’s aura of rational specialness.Footnote ³⁷ Granted, before this leads to concrete policy guidance, more needs to be said about how stringent free speech protections should be. But the contribution we are making to this issue is supposed to compatible with a range of views on that question. However stringent you believe free speech protections ought to be, you should want to ensure that those protections are not withheld from images and music that are expressively powerful, but which don’t straightforwardly say things, the way that verbal expression typically does. Our account counsels against the use of evaluative heuristics that would lead to nonverbal expressive acts being misjudged in that fashion.

Notice that, with regard to both policy-related points, above, our attention has broadened from mainly considering music and images from the speaker’s perspective to considering them from the audience perspective too. Our account in Sections II and III argued that nonverbal media can be vehicles for thinking aloud, but our reasons for caring about this relationship between thought and expression aren’t exclusively linked to the speaker’s interests in realizing or grasping their thoughts. They are also grounded in everyone’s interests, qua audience/listener, in being able to engage with the realized products of other people’s thinking aloud. These other interests have been off-stage in previous sections, but that isn’t because we want to downplay their importance. The inseparability of freedom of thought and expression matters, morally, in relation to speaker and audience interests alike. Our focus has been on the former, only because our central puzzle—how thought is related to nonverbal expression—naturally focuses the spotlight of inquiry on those interests.

Another potential misinterpretation for us to note and clarify concerns the distinction between saying and doing—or, in J. L. Austin’s terminology, between locutionary action on one hand and illocutionary/perlocutionary action on the other. Some free speech theory conceives of speech, and its value, primarily in terms of its being how we say things, while suggesting that the rationale for excluding other expression from free speech coverage is precisely that it does more than mere locutionary saying—that it constitutes a harmful illocutionary action, such as a threat or an incitement to violence.Footnote ³⁸

To clarify, our claim at the outset was that the case for free speech partly rests on the idea of speech as a privileged medium for conveying thought. We don’t think of speech as being limited to locutionary action or deny that speech does more than capturing and conveying thought. Nor are we siding with theories that tie free speech protection to sayings, while withholding protection from illocutionary or perlocutionary doings.Footnote ³⁹ For our purposes, the issue of what disqualifies speech from free speech protection should be regarded as independent of anything linked to a ‘saying v. doing’ dichotomy. The normative significance of speech is not exclusively rooted in the fact that it is a vehicle for thought. We see this as one part—an important part, but not the only part—of a multifaceted explanation of why speech is a morally distinctive region of human activity, and of why we should recognize a separate cordon of principles constraining the governance of speech.

VI. Does Our Account Prove Too Much?

Returning to policy implications, we want to conclude by acknowledging and addressing a worry about the potentially over-broad scope of our account. The worry is this: if types of expression get included in the scope of free speech coverage when or insofar as they facilitate thinking aloud, then what principled limits can we impose on the range of activities that are covered by free speech? Our account might appear to imply a scope so large as to be ultimately vacuous. After all, couldn’t any action that thinking beings like us do serve as the externalisation or expression of someone’s mental life—dropping litter, say, or trying a new recipe, or whistling a tune? Surely we don’t want to say that all these count as instances of self-expression, of the putatively special kind that we have been characterizing as meriting free speech coverage. So where, then, are the boundaries?Footnote ⁴⁰

Our account can provide an answer to this challenge. There are many ways in which people’s acts can be said to express their states of mind: your act of dropping litter; your act of donating money to one charity rather than another; your act of buying a four-wheel drive as an urban family car; your attempt to file your tax return with integrity; your prioritisation of your children’s bedtime routine over a night out with friends; and on the list goes. In one way or another, each of these acts or patterns of behaviour expresses your mental life, reflecting your sense of identity or the values by which you choose to live your life.

This is a perfectly respectable—and, potentially, theoretically rich—notion of expression, as far as it goes. But it is importantly different from our notion of thinking aloud. “In such a general sense of expression,” as Jonathan Gilmore rightly observes, “there is no distinction between expressing one’s thoughts and manifesting or revealing them.”Footnote ⁴¹

To count as an episode of thinking aloud, we have said, an expressive act must satisfy two conditions: (1) it has to realize some mental content; and (2) it must do so in a way that facilitates the thinker’s apprehension of that content. The kinds of everyday acts that we listed above might result from entertaining mental content. They might in turn cause all sorts of contents to be entertained in the actor and others. They might even be performatively executed as a way of signalling certain aspects of someone’s mental life to others. But, per Gilmore’s point above, these are expressions of mental content only in that term’s widely extended sense—not in strictu senso. Buying a car or filing taxes don’t seem to be the kinds of expressive acts that allow for the realization of mental content in an epistemically illuminating way, in the same way that the creation of a visual artwork, the composition of a piece of music, or the utterance of a natural language sentence do. For an act to facilitate the realization and apprehension of mental content, it needs to not just be caused by that content, but also to represent the content back to the actor, in a way that makes it more legible than it had been in its incipient, pre-expressed state. It cannot just be an actor exteriorizing their mental stuff; it must be an exteriorization whose representational richness, depth, and subtlety enables the actor to see inside their own mind. And it would seem grossly reductive to supposed that any and every kind of action provides the actor with this sort of representational power. So it is these two conditions in our account—realization and apprehension—that set the boundaries on the scope of expressive acts that fall within the scope of free speech coverage.

Still, a related worry may be pressed. Our account tells us which expressive acts are in principle deserving of free speech coverage. But as stakeholders in these issues—artists, musicians, consumers of the arts, policy-makers, law enforcers, and so on—we may still find it hard to know which types of expression meet this condition. Perhaps it is clear that littering or buying a car don’t meet the realization and apprehension conditions on thinking aloud. But there are borderline cases that seem harder to adjudicate. What about architecture, or fashion, or gardening? Could there be token acts of self-expression in these domains that realize mental content in a way that allows the thinker to better grasp it?

This is a practical policy question, by our lights, rather than an underlying conceptual question for our account. Our answer to it is appropriately pragmatic—viz., that we do not need to make these adjudications a priori, in a cultural vacuum. Many cultures find music and images to be powerful vehicles for the expression of thought. Few cultures find the same to be true of vehicle purchasing or dropping litter. Of course, there are cultural contingencies in play in determining which nonverbal media come to be thought of as fruitful expressive media for particular cultures and individuals. But there is no need to try to stand outside this contingency and nominate a timeless, transcultural repertoire of nonverbal expressive media to which to extend free speech coverage. This is a question for societies to continually wrestle with, via processes of cultural negotiation and expressive experimentation.

Musical and pictorial expression are already esteemed in many cultures as important media for the expression of people’s mental lives. The problem in free speech theory is that we casually endorse certain ideas that make it seem, prima facie, as if these media lack the crucial feature—namely, linguistic articulacy—that bestows special expressive utility upon a given medium. We have tried to dispel this impression by offering an expanded account of the process of thinking aloud that works for verbal and nonverbal media alike. Sometimes we come to know what we are thinking by expressing ourselves in language, sometimes through the creation of an art object. We have argued that it is this similarity, rather than any socio-functional similarity of the kind surveyed in Section IV, that should determine the scope of expressive acts falling within the coverage of free speech principles.