High Culture and Hyperstimulus

What’s the difference between a Renaissance nude and a Playboy centerfold?

Mark Twain sardonically called the painting above “too strong for any place but a public Art Gallery”.

In every gallery in Europe there are hideous pictures of blood, carnage, oozing brains, putrefaction—pictures portraying intolerable suffering—pictures alive with every conceivable horror, wrought out in dreadful detail—and similar pictures are being put on the canvas every day and publicly exhibited—without a growl from anybody—for they are innocent, they are inoffensive, being works of art.

And with a self-aware nod, we note that it would be pretty indecent to hang such a picture if it were a photograph, wouldn’t it. The knowing relativist in us suspects that maybe these were scandalous and indecent at the time, and it’s only the blind march of time that denudes them of eroticism and lends them respectability.

On the other hand, the centerfold – which itself must seem artistic in comparison to a lot of readily available internet fare today – seems more distilled. The sweetness of refined sugar rather than a fruit; the intense high of crack cocaine rather than the pleasant buzz of a coca leaf.

There is something to this intuition. And yet, what is sublimity in art, or music, or fine dining, or any other medium of human excellence, if not also the refinement of a stimulus to flood the brain with unnatural pleasure? Clearly being highbrow is not to avoid evolutionarily fruitless self-stimulation, but to be discriminating in one’s evolutionarily fruitless self-stimulations.

Here, I offer a theory of the grounds of such discrimination, and what distinguishes a human pursuit from an animal one.

Hyperstimulus#

Our brains take in input from the world via our five-ish senses and respond with actions. Certain types of inputs are experienced as pleasant because over the course of human evolutionary development, they reliably indicate beneficial things worth pursuing. Sugar tastes sweet because its usual sources in nature are energy-dense and full of nutrients. An image of a beautiful nude woman in the visual field is exciting because it suggests an opportunity to reproduce. Organisms that reliably pursue these things succeed and reproduce; those who react with indifference or aversion don’t.

This is true not just of humans, but of any organism – indeed any system at all – that responds to its environment in goal-oriented ways. Baby gulls are attuned to the red spot on the bottom of the mother’s bill and baby monkeys to soft fur because these usually indicate the presence of nurture and food. Parent birds are attuned to the volume of their children’s calls so as to preferentially feed the strongest ones. From a survival and reproduction perspective, heuristics like these are useful and cheap ways to motivate productive behavior.

In all of these cases, however, useful signals can become Goodharted. That is, any behavior leading to a desirable outcome (say, reproduction, nutrition, or feeding one’s young) whose motivation is based on a signal of that outcome – an imperfect indicator, rather than the outcome itself – is vulnerable to exploitation. An organism will be tempted to juice the signal without the actual substance that made it beneficial. Baby gulls, it turns out, will abandon their actual mothers for a stick with a bigger red spot, and baby monkeys prefer a fuzzy doll to a wire mesh with actual milk. Parent birds will sit on the biggest egg and feed the loudest chick, despite the fact that it’s implausibly large and loud for its own species, at the expense of their own actual children. For our part, we humans gorge on sugar without nutrition; on pornography without sex; on sex without conception; on Instagram likes without real friendship. In each of these cases, the organism is manipulated into wasteful or self-destructive behaviors by the separation of a signal from the benefit it usually indicates. Sometimes by a curious researcher, sometimes by its own folly, and sometimes deliberately by a parasite that can benefit itself at the organism’s expense.

For humans, restraining more destructive hyperstimuli is an important function of a normative order. So it’s tempting to think of the highbrow/lowbrow distinction as a sort of nudge away from hyperstimuli. And high society does tend to shun a variety of obvious hyperstimuli like soda pop, pop music, and big breasts.

And yet… what is a sublime experience if not a hyperstimulus itself? The Venus is not quite so viscerally erotic as a Playboy centerfold, but it is more gratifying in other ways. Surely the sublimity of a Beethoven sonata is just as pleasurable to the discriminating listener as the latest K-pop hit is to a stan. From a strictly evolutionary perspective, it’s hard to imagine the feeling of awe at the grandeur of a Bierstadt as anything other than the unnatural amplification of certain stimuli designed to trigger an emotional response without the substance that usually goes along with it.

Even within obviously destructive hyperstimuli, we respect those who discriminate: the sommelier has a respectable hobby; the wino is a loser. (There are, of course, limits: I’m not aware of any fentanyl connoisseurs.)

So why prefer the museum patron to the pornography addict, the sommelier to the wino, or the symphony enjoyer to the K-pop stan? Should art, high culture, aesthetics, sublimity – what we think of as the highest of human experiences – be classed along with drugs, pornography, and other addictive vices? At the end of the evolutionary day, they don’t necessarily cash out in higher evolutionary fitness. Indeed, the cultivation of unnatural pleasure responses – extreme fetishes, for example, but also, say, an appreciation for conceptual art – might plausibly be considered worse than the overindulgence of natural ones. Given what they share in common, on what grounds should we prefer the former to the latter?

Low Culture and Solved Stimuli#

If it’s a mistake to define low culture simply on the grounds of it being hyperstimulative, it would be even more monstrous to toss out high culture on the same grounds. We rightly celebrate high culture as, at least potentially, an avenue of human excellence; something that allows us to transcend – if only in part – the animal exigencies of survival and reproduction. Any difference between high and low culture, then, must lead us to discriminate not by the category of hyperstimulus, but within it.

To see the difference, imagine everything the human brain is attuned to and finds pleasurable. Some pleasures, like the centerfold, latch onto one specific desire, in this case for sex. Others go even lower-level: cocaine and other drugs bypass the senses entirely and go straight for the internal chemical reward signal. Some are innate, some are learned. Some aim for a small and definite set: chibi eyes stimulate the facial recognition modules of the brain, supplemented with narrative or erotic stimuli. K-pop, likewise, is a highly refined mix of martial, erotic, and parasocial stimuli.

The K-pop example is instructive:

The founder of the first of these [K-pop] companies famously declared that “S.M. Entertainment and I see culture as a type of technology.” In the ‘90s he reversed engineered this technology with methods that mirror Korea’s famous chaebol: he began by consciously breaking down the constituent parts of successful American and Japanese pop hits, simplified these parts into scripts that could be easily replicated, hired foreign expertise to shepherd the design process, and then secured government funding to jump start his new export industry.

Tanner Greer, “Why Chinese Culture Has Not Conquered Us All”

No matter the particular stimulus or stimuli being targeted, low culture is defined by there being a formula for maximum effect. Comic book movies – another enormously popular lowbrow staple – are similarly criticized for being formulaic (and yet: the formula works!).

The existence of such a formula – the “solving” of the stimulus – means the pleasure response it elicits no longer corresponds to the substantive benefit that attuned the brain to that sort of stimulus in the first place. The signal has been Goodharted. When sugar refining becomes sufficiently cheap, when we solve sweetness production, sweetness is no longer a meaningful signal of nutrition. Straightforward pleasure-maximizing behavior becomes folly: the hunter-gatherer can profitably gorge on the occasional honey cache, but the same strategy quickly leads the sedentary moderner to type 2 diabetes, as many in the developing world are discovering.

Even worse though, continuing to respond to a “solved” stimulus makes you vulnerable to parasites. It’s not just that we now have more hedonic rope with which to hang ourselves, but that the rope can also be used as a lasso: like the bird tricked into caring for unrelated chicks by an implausibly large egg, solved stimuli – Goodharted signals – can be deployed strategically to elicit behavior from others.

To take a relatively benign example, the entire field of marketing is dedicated to deploying audiovisual hyperstimuli to capture attention. Fortunately the human (really, the mammalian – but not the avian) audiovisual system is complex enough to resist the straightforward elicitation of a definite response from a given stimulus – at least so far as we know – and this complexity arguably exists precisely to prevent the sort of mechanical leveraging that actual endoparasites use to manipulate simpler organisms like insects and mollusks. Subliminal messaging isn’t real, and marketing mostly doesn’t have the magical preference-shaping effects we fear it must be having on everyone but ourselves, even if it is effective at briefly capturing attention.

The really important parasitic hyperstimuli are not the advertisements we see in passing, but the products themselves that we willingly input. The cultivation of fandoms and stan communities. Entire business models – alcohol, mobile games, and camming are notable examples – that rely on parasitizing a small number of addicts who provide a vast majority of revenue. This is only possible to the extent that, for the unfortunate addict, some pleasure-eliciting stimulus has been effectively solved: it can be profitably deployed at low cost to the issuer, and without benefit (beyond the pleasure response) to the consumer.

Naturally, these are all low-status vulnerabilities.

To look down on low-status hyperstimuli is an important meta-preference, a defense against being exploited by solved stimuli. Indeed there seems to be something “less than human” about falling for them: certainly compulsive addictive behavior, but also – though we rebel against verbalizing it for fear of pretension – the unironic enjoyment of transparently formulaic media. And not because of any intrinsic qualities of the media itself, but because its formulaicism suggests parasitic exploitation. We observe somebody else’s will legibly manifested in the person’s actions.

A Functional View of High Culture#

So much for low culture. But is high culture really any different? The typical sociological-left view of high culture is one of distinction. Upper classes try to stay ahead of the tastes of the lower classes – the mobile games and Marvel movies – in order to maintain an insider’s club where power of some sort can be exercised. The substantive difference is basically arbitrary, and without social inequality, there would be no need for high culture.

No doubt pretension and distinction play some part in some cases. But distinction is neither necessary, nor sufficient, for high culture. Instead, we can explain the relevant features – a hierarchy of taste (there is widely agreed upon better and worse, and not simply different), and the conveyor belt by which high culture becomes low status after a time – with a functional view, where the substantive difference between high and low culture is not arbitrary.

If high culture is, in principle, just as hyperstimulative as low culture, the important substantive difference is that high culture consists of unsolved stimuli. Hyperstimulus all the same, but not (yet) exploitable.

It’s easy enough to say that a truly sublime experience is ineffable. One cannot enumerate the necessary and sufficient conditions for evoking it. Close imitations might totally fail to do so. But thinking of the ineffable as an unsolved stimulus points to why the sublime has to be ineffable. Without a formula, there’s no way to “juice” the stimulus; one cannot get addicted to a singular experience.

Ineffability thus renders hyperstimulus safe, at least with respect to the obvious pitfalls. The sommelier, whether from pretension or genuine enjoyment does not matter, does not allow himself to get drunk, since that would spoil the sublimity of a good wine. High culture is controlled hyperstimulus; an inarticulable ecology of inhibitions maintained by an aesthetic community’s desire for distinction, within which hyperstimulus can be cultivated and safely enjoyed.

In this sense, the high/low culture distinction can cut both across and between subcultures. For example, the authenticity ethic of the early punk scene would count as high culture within its domain: despite the simplicity of the music itself, which might lend itself to even more straightforward formulaicism, the cultural aspect of punk – an aversion to “selling out”, which might be understood as formulaicism as such – introduced a novel set of unsolved signals independent of the music itself. Indeed, the simplicity of the music itself is a reaction to prog rock’s perceived attempt to solve the rock music stimulus through instrumental virtuosity. There are aesthetic and ideological signals involved – and of course these can come to be exploitable over time – but there can be no set of necessary and sufficient conditions for being judged as having sold out. An audio stimulus solved qua audio is now linked to more abstract signals as part of an aesthetic community – and a large part of the enjoyment of punk music comes not from the music itself, but from signals of identification with that community.

This dynamic also explains the conveyor belt. The ineffable gets effed, given enough time. First imitators try their hand, and then – as with the K-pop example – the formula is discovered, made explicit, and mass-produced (and the very aversion to this is what makes the authenticity ethic a species of high culture). For all that people complain about modern art, the decline in status of representational art was inevitable once that stimulus was “solved” by photography. Realistic representations are cheap – cheaper even than stylized ones. Even in painting, the ability to create a superficially compelling painting is widely distributed, and mostly not considered fine art at all.

The conceptual turn in art is simply a way to maintain ineffability by linking up ideological, intellectual, and other nonvisual signals in an aesthetic community independent of the visual stimuli. Art – necessarily, if lamentably – no longer stands on its own as art; it must be understood as part of a conversation. Pollock, perhaps, is easily replicated, which is exactly why there can only ever be one Pollock: a Pollock double, identical as a visual stimulus, would not give intellectual signals that stand in the same relation to the broader artistic memeplex.

Pretension, perhaps, but also a defense against compulsive consumption.

High Culture in the AI Era#

Ordinarily, the solving of a stimulus takes time. Sometimes broader interest is lost before a medium can even be solved: classical music still retains a great deal of cultural cachet because there’s very little commercial impetus to “solve” the fugue, for example. But we may stand on the brink of enormous and unprecedented changes in cultural expression if we think of AI as a stimulus solver.

Yes, the artifacts are still as yet imperfect. But style transfer works shockingly well. I can generate passable Rembrandt paintings of Shrek with nothing more than a prompt. Cover letters, novels, and research papers, that give off any signal you like.

I have no doubt ChatGPT could pass a turing test convincing anyone of its punk bona fides. It’s already been winning art competitions for a couple years now, and it’s not hard to imagine ChatGPT breaking into fine art with a suitable artist’s statement. It can even make convincing short films.

Many are celebrating these capabilities as heralding the democratization of cultural production. And yet… to democratize a medium is necessarily to render it formulaic. The democratization of cultural production cannot be the end of the story.

My prediction is that we witness a total loss of interest in many existing media, especially text media, as serious vehicles of high culture. Perhaps the novel will be killed as a serious artistic medium, though shoppers of airport bookstores will continue to read AI-generated genre fiction. Those media that do survive will be less democratic, less meritocratic, and more connection-based (and to a great extent, the fine art community is already like this). The aesthetic community rises in significance compared to the art itself. Pieces will be evaluated, not on the basis of visual, sensory, or even intellectual or ideological signals, but signals of personal identity – ones that can’t be replicated by an AI. In the limit, there need be no artifact at all: a pure and exclusive patronage network called the “art community” for historical reasons, but unencumbered by the increasingly cursory production of artifacts.

Existing great works – like Pollock today – will largely continue to stand despite being easily replicable, on the basis of their standing within the broader aesthetic community rather than their substantive aesthetic qualities. Perhaps it’s not so implausible that classical music will be the music of the 24th century.

Hermetic upper-class tastes, antiquarian middle-class tastes, and relentlessly modern low-class tastes. And a chasm will open between those willing to take pleasure in AI generated stimuli – that is, to appreciate audio, visual, or textual stimuli as such, and those unwilling to do so. The latter will seem ascetic and pretentious; the former, pitiable and compulsive.

Unfortunately, I don’t see a way out of this dilemma provided generative AI continues to improve and expand.

Conclusion#

One of the most remarkable human qualities, the true testament to the social intelligence hypothesis, is the ability to sublimate predictable pleasure responses when we perceive them to be exploitable. This sublimation makes culture possible: rather than directly sating our basic animal desires, we can develop genuine aversions to otherwise pleasurable stimuli – a sweet tooth is mutable – and take genuine pleasure in activities very far removed from any fitness benefit. We can, in other words, shape our own preferences over the timescale of acculturation, and – to a great extent – choose what sorts of hyperstimuli we will respond to, and which we will not. (Of course, the ability to do so is also highly unequal, and correlated with other important but imperfectly observable qualities, hence the use of high culture as social distinction in addition to its being genuinely enjoyed.)

Thus, on the one hand, high culture and low culture stand together as unnatural pleasure responses. We do not have to be ascetic and forswear our own pleasure responses just because we no longer live in the EEA. And yet, it would be a mistake to relativize such judgments as separate high culture from low. To be a hedonist, seeking pleasure without consideration of social status, is to offer one’s own will up to others as clay – and note that this is precisely the opposite of the authenticity ethic that seeks to liberate the “authentic” self and its natural pleasure responses from social conditioning (paradoxically, a highly socially conditioned impulse). Social judgment on this point – the willingness to look down on the enjoyment of solved stimuli – is precisely what protects our individual wills from being coopted by deft signal-wielders. The real danger is seeking pleasure without an aesthetic community.