Friday, March 4, 2022

On Juxtaposition

One of the ideas beloved of Media Theorists is that juxtaposition is a technique of great power. This is part and parcel of the whole program of arguing that Media is Extremely Powerful and Has A Great Influence On Minds. Which isn't a wrong thing to say, but Media Theorists really want it to be reducible to little digestible bites rather than a large horrible chaotic gestalt that is for all practical purposes impossible to analyze bit by bit.

This is indeed greatly to be desired, but suffers from the problem that it's not true. Media's influence on society is a large horrible chaotic gestalt, and cannot meaningfully be reduced to analysis bit by bit.

Let's examine the Kuleshov Effect. This phrase covers, in practice, quite a bit of territory. In the original form, a Russian dude named Kuleshov edited up a short film which intercut various mostly still shots with footage of an actor's face wearing an essentially unchanging neutral expression. The actor footage might be literally repeated, I don't know and it doesn't matter. A steaming bowl of soup. Actor's face. A dead child. Same actor. Etc.

The conceit is that the viewer will incorrectly perceive changes in the actor's expression, in their vibe, based on whatever was cut in adjacent.

You can find the original on youtube, and if you're a skeptic like me you watch it and think "you know, maybe not so much." It turns out that this specific effect is one of those things that essentially vanishes as soon as you properly blind the study. It's an effect that feels like it ought to be real, so it's very easy to prime your subjects, and lo, the effect appears. But it's not real, you primed them.

If you google around, you will also find a clip of Hitchcock supposedly explaining this effect, but in the first place he never mentions Kuleshov, and in the second place he's talking about a different thing: specifically, that a reaction-shot is read differently depending on what the reaction is (notionally) to.

At this point Kuleshov probably means various things to various people.

Regardless, the original was some effort to establish that juxtaposition leads us to imagine a relationship. The footage in this case is essentially stills, so it closely resembles the spatial proximity of two photographs, and so at least one "researcher" loves to drag it out when talking about still photos. The aim here is to establish some very basic, very universal, human behavior. To wit, that humans will tend to create associations between things that are spatially proximate.

The Hitchcock clip is notably different because Hitch is talking about a temporal proximity, a temporal sequence: the man sees the girl in the bikini, and then the man smiles. We perceive an actual change as "caused by" the girl in the bikini.

Kuleshov proposes that we see change where none exists. We see a notional cause, and infer an effect. This, while not strictly a spatial proximity, feels like one at least to me. Certainly media theorists believe strongly that mere spatial proximity of media fragments leads humans, at a very basic and universal, perhaps even biological level, to infer relationships.

This, as stated, is nonsense. We do not. There isn't any reason for us, for example, evolve such a behavior. This is an invention of media theorists.

What we do have is a strong tendency to believe in associations which are proximate in time. If A is followed by B, we will with alarming rapidity conclude that A causes B. If we desire B we should induce A to happen. This is the basis of magic, and seems indeed to be a basic human reaction of some sort.

The fact that the lion stands next to a boulder is of no salience. That someone winds up being eaten shortly after the lion shows up is, however, pretty interesting.

So how do we get around to the fact that, at least sometimes, spatial proximity does in fact produce meaning, does in fact propose a relationship?

I probably don't have the whole of it, but here's one path that's pretty obviously true.

A story, once written down, or committed to a cave wall, substitutes spatial proximity for temporal. Little Red Riding Hood enters the forest. The Big Bad Wolf confronts her. I have placed two sentences adjacent, and many of us will know the story. I could just as well place a photograph of a girl in a hooded red cloak on the left-hand page, and a photograph of a wolf on the right; most everyone would know the relationship between them.

This is not, I think, merely a cultural reference. One can tell new stories using the same methods.

The point is to propose a relationship in a way that appears deliberate. If I place two portraits adjacent, ostentatiously adjusting sizes and positions to make it appear that the two subjects are looking at one another, one might reasonably infer that I mean to suggest just that.

Contrariwise, if by happenstance the portrait of the celebrity appears to more or less be gazing at the model in the advertisement on the opposite page, we might justly assume that this is an accident of layout. We might thus read nothing into it.

My half-baked theory, at this point, is that the idea that a spatial proximity implies a relationship derives specifically from our human habit of writing things down (which I mean broadly, so as to include narrative drawing.) By substituting spatial adjacency for temporal adjacency, we bring that magic, that essentially A-causes-B theory, into this new regime.

At that point, the world opens.

If you stick a picture of a black man next to a picture of a watermelon, on purpose, you evoke the story of how black folk jes' loves watermelon, which in turn opens into a maze of racist tropes.

If you stick a portrait of President Obama next to an advertisement for watermelon flavored candy, well, that's probably an accident (of course there will be head-shaking and "they knew! THEY KNEW!!!" from the usuals, but nobody serious will buy it.) Nevertheless, you don't do that, because it's impolite. In the same way you check the movie marquee to make sure that the first letters of each line, read vertically, do not spell some saucy word, you check to make sure you haven't got any spreads placing black people next to fried chicken ads. It's just a bad look.

It is a second-order effect, an obviously unintentional resemblance to something unsavory, undesirable. You try to remove those things too, but not because they actually tell a story.

What this suggests to me, though, is that you can't just stick any two things together and imagine some effect. The original Kuleshov Effect is 99% bunk. You need to work harder, you need to make evident a story (broadly construed.) It needs to feel deliberate, it needs to be story-like in some sense. Or at least, these things aren't exactly going to hurt.

Worse, the ease with which clumsy workers can "reproduce" the Kuleshov effect suggests strongly that you can easily fool yourself, and your test subjects. You won't fool the readers you haven't been able to prime. They will not see the clever allusion of the hammer and the wall to neoliberalism that is so very very evident to you and to the people you explained it to.

You gotta work in a kid in a red hood, and some kind of a wolf.


  1. It should be called "The Editors Effect," as explained in this excellent video on the This Guy Edits:

    The early film editors were mostly women, and they invented the "Effect" well before Kuleshov got around to demonstrating it.

    1. What's not clear is what exactly is meant by the Kuleshov Effect here. Obviously the sequences mentioned in that video, like the Hitchcock clip, work just fine, they're basic vocabulary.

      These are, essentially, the viewer imputing a nuance to an already-expressed emotion, based on the prior shot. We see the "reaction" and assume it's a "reaction-to" because, essentially, we believe in magic.

      The sample associated direction with Kuleshov is much more static. We are, allegedly, imputing emotion to a completely neutral face. There is no "he sees the soup" because the eyelines are wrong, we're just supposed to somehow associate soup/face and impute emotion where none is expressed.

      It's maybe a spectrum or something, but there is something different between the two. The original simply doesn't work particularly well, and the modern "reaction shot" works fine, but is surprisingly technical to execute in a way that actually works.

      The whole thing feels a bit like the "rule of thirds" where we're supposed to believe some stupidly simplistic thing where in fact what's going on is a complex gestalt that depends on a lot of fiddly bits, on relationships between a lot of parts, and so on. When it works, it looks like it's very simple, but in reality it's not.

  2. Replies
    1. That whole deal is an absurdity that just won't end. It was always just a stupid twitter fight. Paul Halliday drafted his daughter to try to damage Martin Parr, for reasons we can only speculate about. Chesterton got into it for a bit, and that was it. It was just a dumb twitter fight, with essentially no consequences.

      They successfully sold a lot of books, which probably wasn't what they were looking for. The publisher got out from under the remainders, so a huge win for the publisher and some modest exposure for a pretty good, albeit minor, book.

      Neumüller wanted, I think, to ingratiate himself with Halliday and Chesterton, for reasons that are again obscure, but failed because they figured him for doing a hatchet job. He's dressed the thing up as a Serious Academic Dispute over Serious Things instead of a shitty twitter fight. Neumüller appears to be an amiable moron, which is a step up from the vicious morons he was trying to make friends with.

      Low is basically taking a swing at Neumüller I guess? Honestly, Neumüller should definitely be punched in the nose for his dogshit analysis, but I don't know if Low is going to do that, particularly.

      AD Coleman is a real boy, though, and he seems to think Low did his homework.

      I assume Coleman is getting a pile of shitty venal emails about "so problematic such racist" but honestly the dude lives in Staten Island so I can't imagine blithering pricks like Halliday and Chesterton are *that* likely to get much traction here.

    2. "vitriolic, resentful, and uneasily grandiose"