Friday Fotos: Irises, yellow & white, with cars

Globally, songs and instrumental melodies are slower and higher and use more stable pitches than speech

Yuto Ozaki, Adam Tierney, Peter Q. Pfordresher, et al. Globally, songs and instrumental melodies are slower and higher and use more stable pitches than speech: A Registered Report, Science Advances, 15 May 2024m Vol 10, Issue 20, DOI: 10.1126/sciadv.adm9797

Abstract: Both music and language are found in all known human societies, yet no studies have compared similarities and differences between song, speech, and instrumental music on a global scale. In this Registered Report, we analyzed two global datasets: (i) 300 annotated audio recordings representing matched sets of traditional songs, recited lyrics, conversational speech, and instrumental melodies from our 75 coauthors speaking 55 languages; and (ii) 418 previously published adult-directed song and speech recordings from 209 individuals speaking 16 languages. Of our six preregistered predictions, five were strongly supported: Relative to speech, songs use (i) higher pitch, (ii) slower temporal rate, and (iii) more stable pitches, while both songs and speech used similar (iv) pitch interval size and (v) timbral brightness. Exploratory analyses suggest that features vary along a “musi-linguistic” continuum when including instrumental melodies and recited lyrics. Our study provides strong empirical evidence of cross-cultural regularities in music and speech.

From the introduction:

Culturally relativistic hypotheses appear to be dominant among ethnomusicologists. For example, in a 13 January 2022 email to the International Council for Traditional Music email list entitled “What is song?,” International Council for Traditional Music Vice-President Don Niles requested definitions for “song” that might distinguish it from “speech” cross-culturally. Much debate ensued, but the closest to such a definition that appeared to emerge, was the following conclusion published by Savage et al. (25) based on a comparative analysis of 304 audio recordings of music from around the world:

“Although we found many statistical universals, absolute musical universals did not exist among the candidates we were able to test. The closest thing to an absolute universal was Lomax and Grauer’s (30) definition of a song as a vocalization using “discrete pitches or regular rhythmic patterns or both,” which applied to almost the entire sample, including instrumental music. However, three musical examples from Papua New Guinea containing combinations of friction blocks, swung slats, ribbon reeds, and moaning voices contained neither discrete pitches nor an isochronous beat. It should be noted that the editors of the Encyclopedia did not adopt a formal definition of music in choosing their selections. We thus assume that they followed the common practice in ethnomusicology of defining music as “humanly organized sound” (31) other than speech, with the distinction between speech and music being left to each culture’s emic (insider and subjective) conceptions, rather than being defined objectively by outsiders. Thus, our analyses suggest that there is no absolutely universal and objective definition of music but that Lomax and Grauer’s definition may offer a useful working definition to distinguish music from speech.”

However, the conclusion of Savage et al. (25) was based only on an analysis of music; thus, the contrast with speech is speculative and not based on comparative data. Some studies have identified differences between speech and song in specific languages, such as song being slower and higher-pitched (32–35). However, a lack of annotated cross-cultural recordings of matched speaking and singing has hampered attempts to establish cross-cultural relationships between speech and song (36). [...]

Our study overcomes these issues by creating a unique dataset of matched singing and speaking of diverse languages, with each recording manually segmented into acoustic units (e.g., syllables, notes, and phrases) by the coauthor who recorded it in their own first/heritage language.

Tulip, pale yellow with a wash of red

Unfrosted: Caught in a network of reference and allusion [Media Notes 119 F]

As counterintuitive as it might seem at first glance, Unfrosted is fundamentally an intellectual movie. It starts out that way and never strays far from it. What do I mean by “intellectual”? Rather than give you an answer right away, let’s take a look.

The movie opens with a little mystery. First a slide that says “NETFLIX Presents,” and then our mystery unfolds. These objects appear on the screen, one after the other, accompanied by quiet nostalgic music:

• a red bandana, opened and spread flat,
• a Woody Woodpecker comic book placed there by a child’s hands,
• three packets of Bazooka bubble gum,
• a rubber ball,
• a pack of baseball cards held together by a rubber band,
• GI Joe,
• a wallet of some kind,
• a penknife,
• a Slinky, and
• a red cloth that I don’t recognize, but it’s shaped a bit like a ping-pong paddle and has lettering on it.

I don’t know about you, but as I saw those little hands place those (precious) objects on the bandana I was identifying the objects and asking myself: What is going on?

We discover soon enough. A boy is running away from home and those are the things he’s taking with him. He goes into a diner, takes a seat next to some guy – why’d he do that, all the other seats were empty? – and orders: “Two Pop-Tarts, please. Leave the box.” [Did diners serve Pop-Tarts back then?]

As it so happens, the kid sat next to none other than Bob Cabana (Jerry Seinfeld), the executive who brought Pop-Tarts into the world. He starts telling the story: “Well, in the early ‘60s, the American morning was defined by milk and cereal.” And we’re off.

We’re back in the early 1960s. If you’re old enough – Seinfeld was born in 1954, I was born in 1947 – you might feel a warm glow of nostalgia wash over you as see those styles and objects from your childhood. If you’re somewhat younger, perhaps you’ve seen Mad Men or various movies set in that period; they’re all over the place. As these things unfold before you, may even nod to yourself, “I recognize that, and that, ah, so nice...” And there are more specific things. The car Cabana/Seinfeld was driving was styled like a Chevy “unsafe at any speed” Corvair, but I didn’t remember a Corvair station wagon. I eventually got around to checking on that and, yes, they made one, from 1961 to 1963, just in time for Pop-Tarts! I’m guessing that many/most in the audience may not have recognized the car (they only lasted until 1969), but I’m probably not the only one who was curious about the wagon.

Here's the point: The whole (freakin’) movie is like that. Those complicated plots within plots (Kellogg’s vs. Post, cereal vs. milk, Russia vs. America, mascots vs. management) seem to be mostly a device on which to hang an elaborate network of references and allusions, to things, events, and people, but also to movies and TV shows, not to mention the zillion actors who traipse through the movie. Whatever else you’re doing, you’re also marking those those things one by one, one after another, all the way through the film. Many of them are wry or funny, a few are hilarious, but most of them are just there. If the film were showing in theaters, they’d give you a small sheaf of papers listing all the various allusions so you could check them off, one by one, as they whizzed by. I don’t know what you’d do at the end, when all those product mascots, not just cereal mascots, stormed the Kellogg’s headquarters building. Were all of them real mascots for real products, or were some of them made up? I don’t know. Do I get a prize if I correctly identify every one of them? If so, I’d rather have diner Danish than Pop-Tarts.

And THAT’s what I mean when I say the movie is fundamentally intellectual. Whatever else is going on, you’re always thinking about this or that thing that’s just popped up on the screen.

Seinfeld has a bit about the “weight problem in this country.” It has these two lines:

The donut hole. The donut hole. Let’s stop right there. What a horrible little snack. If you want a donut, have a donut. Why are you eating the hole?

It’s such a freaky metaphysical concept to begin with. You can’t sell people holes. They… A hole, a hole does not exist. Words have meaning.

That’s intellectual. It’s also funny. A great deal of humor is about exploiting and exploding cracks in concepts. That’s fine for stand-up. But you can’t make a compelling feature-length movie out all those little cracks. That’s why Unfrosted feels unfocused. All those jokes and allusions make it impossible to build and sustain narrative momentum. This comes down to a case of not being able to see the forest for the trees.

More later.

The Pop-Tarts Portfolio

Industrial Strength Pop-Tarts

Pop-Tarts, the Monolith

Pop-Tarts, the Myth

Pop-Tarts in the Wild

Pop-Tarts, Sports Edition

Pop-Tarts: Hot! Hot! Hot!

86 Billion Neurons [& who knows how many synapses]

Bumping this to the top 1) on general principle, and 2) as a reminder.

 

Steven Mitten reviews The Human Advantage: A New Understanding of How Our Brain Became Remarkable by Suzana Herculano-Houzel, MIT Press, 256 pp., $29.95.

Energy requirements and cooking our food:

Our 86 billion neurons need so much energy that if we shared a way of life with other primates we couldn’t possibly survive: there would be insufficient hours in the day to feed our hungry brain. It needs 500 calories a day to function, which is 25 percent of what our entire body requires. That sounds like a lot, but a single cupful of glucose can fuel the brain for an entire day, with just over a teaspoon being required per hour. Nevertheless, the brains of almost all other vertebrates are responsible for a mere 10 percent of their overall metabolic needs. We evolved and learned a clever trick in our evolutionary past in order to find the time to feed our neuron-packed brains: we began to cook our food. By so doing, more energy could be extracted from the same quantity of plant stuffs or meat than from eating them raw.

Brain soup:

Brain soup was the method Herculano-Houzel devised to deal with the problem of a brain’s heterogeneity. Her procedure was to dissolve a brain of whatever species, with its millions or billions of cell membranes, in detergent to create a homogeneous distribution of free-floating cell nuclei. She could then sample the suspension, use a blue dye to stain the nuclei, count them up, and confidently extrapolate to the number of cells in the entirety of the brain, or whatever part of the brain she had begun with.

Those cells would be of three types—neurons, glial cells, and endothelial cells. Glial cells are crucial to the synaptic transmission of information across neurons, while endothelial cells form the walls of the capillaries that take oxygen and nutrients to the brain via the blood. Fortunately, the neurons could be distinguished by tagging them with a red-colored neuron-specific antibody, one that attaches to the NeuN protein within the cell nuclei. By counting the number that turned from blue to red once the antibody was added to the suspension, she could establish the proportion of the total cell count that was neurons

Human uniqueness:

Here are the numbers she found: the average human brain has 16 billion neurons in the cerebral cortex, 69 billion in the cerebellum, and slightly fewer than one billion in the rest of the brain. This fitted almost perfectly with the neuronal scaling rules derived for nonhuman primates: we have a perfectly normal primate brain, just the right number of neurons for the mass of our brain and also our body size.

That finding flew in the face of conventional wisdom, which argued that when correlations are drawn between body size and brain size for living primates (including the great apes), humans appear to have a brain size three times larger than expected. But Herculano-Houzel argues that it is the great apes, not humans, that are the exception. While the great apes also conform to the neuronal scaling rules—i.e., the average size of their neurons doesn’t increase exponentially as they gain more neurons—their brains are much smaller than should be expected for their body size.

The evolutionary story she tells by way of explanation is one of choosing between brain and brawn. Being restricted to eight hours of foraging a day, the ancestral great apes chose brawn (which, of course, means they underwent natural/sexual selection for a larger body size): the amount of energy that could be acquired was invested in building a bigger body rather than a bigger brain. At seventy-five kilograms a 30 billion–neuron brain was the maximum size that could be fueled. Ancestral Homo went a different way: it increased the energetic uptake from foraging by increased scavenging and hunting while maintaining a relatively small body size, enabling its brain to expand to an estimated 40 to 50 billion neurons for Homo habilis two million years ago. But that was the limit: there was no time left in the day and no other sources of food to exploit. Further expansion of the brain required securing more energy from the same type and quantity of foodstuffs. As from 1.5 million years ago that is just what our ancestors achieved by cooking their food.

Scaling rules:

But even though the human cerebral cortex constitutes 82 percent of the total brain mass, the largest when compared to all mammals, it was found to contain only 19 percent of the total number of neurons in the brain, the same percentage as in the guinea pig and capybara, and midway in the 15 to 25 percent range found in most mammals.

How can the human cerebral cortex have expanded so greatly in comparison to the rest of the brain while maintaining a proportion of neurons equivalent to that found in the cerebral cortex of other small-brained primates? Herculano-Houzel’s answer lies partly in the absolute number of neurons in the human cerebral cortex and partly in the fact that different scaling rules apply to the cerebral cortex and the cerebellum.

These rules are constant across all primates: when additional neurons are added to the brain, the cerebral cortex increases in mass at a much faster rate than does the cerebellum. This is because the cerebral cortex requires larger neurons than the cerebellum—neurons that have long-range connections of several centimeters to link different cortical areas; neurons in the cerebellum need to span no more than a few millimeters. As a result, the cerebral cortex becomes proportionally larger even though the ratio of cortical to cerebellar neurons remains the same. So with humans, the 16 billion neurons in the cerebral cortex result in its forming 82 percent of the total brain mass, despite the human brain’s remaining entirely typical for a primate with regard to the proportions of neurons in the cerebral cortex and in the cerebellum.

Some 1st class jazz: David Sanborn, Dizzie Gillespie, Diane Reeves, David Peaston, Onaje Allan Gumbs

At about 31:44 we have an absolutely INSANE version of "Stormy Monday" with Diane Reeves and David Peaston sharing vocal honors. I thought Sanborn and Gillespie were smokin' on "Tin Tin Deo," by Chano Pozo and Gil Fuller, but "Stormy Monday" lasted the whole damn week.

Assorted Flowers in a variety of colors and shapes

From Pop-Tarts to Jackie O’s and beyond [Media Notes 119 E]

One of the problems I’ve been having with Unfrosted is that it’s unfocused. As I mentioned in my original review, Seinfeld’s a miniaturist. His core artistic discipline is stand-up comedy, a discipline based on two-, three-, and five-, maybe six-minute bits. The Seinfeld Show was a half-hour sitcom, which means it had roughly twenty-two-and-a-half-minutes of show interrupted by commercial announcements. Those didn’t involve much of a plot. Similarly, Comedians in Cars Getting Coffee consisted of a conversation between Seinfeld and some comedian, which each show being ten to twenty minutes long – edited, I believe, from four or five hours of footage.

Unfrosted is a feature-length movie that runs an hour and twenty-five minutes, twenty-eight if you count the final song. While movies are constructed of shorter scenes, they have to have a set of events the builds and resolves over the course of the whole movie. Unfrosted doesn’t do that. Oh, we’ve got lots going on: Kellogg’s vs. Post, Big Milk vs. Big Cereal, Cuban sugar vs. Puerto-Rican sugar, Russia vs. America, mascots vs. Kellogg’s, and all the while NASA’s shooting for the moon. And somehow Pop-Tarts is supposed to hold all that together. It does that nominally, but not emotionally. There were two or three places in the movie where I felt like it should end. But it didn’t, too soon.

There’s a gag about Jackie O that illustrates one aspects of this diffuseness. The gag plays on the fact that there’s a very popular cereal called “Cheerios.” That presents the opportunity for a pun. But in order to get that pun into the movie we’ve got to get Jackie O into the movie. How do we do that? Jackie O is Jacqueline Onassis, formerly Jacqueline Kennedy, wife of President John F. Kennedy, who was in office between Jan, 1961 and Nov. 1963, when he was assassinated. Since Pop-Tarts were introduced about a year later, Kennedy died too early to play a role in this story. But that’s OK, since Unfrosted doesn’t pretend to be historically accurate about anything beyond costumes, cars, furniture, and the general physical setting.

Still, we need a pretext. What’s available. Ah, the Cuban Missile Crisis in October 1962, two years before. OK. The time frame’s close enough. But how do we make a connection between Pop-Tarts and the Cuban Missile Crisis? I can tell you easily enough. But I’m making some kind of point here, so I’m not going to do that, but if you really want to know, you can find out in the plot summary in the Wikipedia entry for the movie.

As a result of this turn of events, the top brass of Kellogg’s are summoned to the oval office at about 52 minutes into the film. President Kennedy enters the oval office: “Have you fellas ever considered calling a cereal Jackie O’s? Just a pitch but I think she’d get a big kick out of it.” There’s another second or two of chitchat and they get down to business. It seems that Post is now working with the Russians to secure a supply of sugar from Cuba. Kennedy: “The idea of our American children waking up in the morning to a commie breakfast pastry really burns my britches.” There’s more chit chat, this time a bit more serious, and Kennedy agrees to help with the situation: “I will instruct my brother Bobby to tighten the screws on those cow huggers.” If you’re wondering how THAT got in there, consult the Wikipedia entry. Out Kellogg’s honchos leave the meeting as the Doublemint Twins enter the oval office for some “executive privilege.” The scene ends at about 54 minutes.

The movie then moves back to other business. We’ve got a two-minute scene in a darkened bar between the Thurl Ravenscroft, the disgruntled actor who plays Tony the Tiger, and a representative of the milk mob. Then we have a minute with Walter Cronkite that’s split between on-air time and off. And now we have two major scenes on the main plot-line. First Kellogg’s conducts a test of their new pastry that is a parody of a rocket launch. The pastry is good, but the “taste pilot” died in a freak accident. Given the general NASA motif I assume this is a reference to the death of three astronauts when the crew cabin of Apollo 1 broke out in flames on the launchpad in 1967. That takes a two and a half minutes or so. Then we segue the funeral for the deceased. That runs for roughly three minutes.

We’re now eight or nine minutes beyond the end to the oval office scene where Kennedy make the Jackie O’s suggestion. That’s not much time at all, but it seems longer than it is because our minds have had to move through four distinctly different settings: a darked bar, a TV program, a pastry launch, and a funeral. That’s a lot of information packed into a short time.

And now we’re on the street in Battle Creek (at roughly 1:03:00). Our three Kellogg’s honchos, Edsel Kellogg (President), Bob Cabana (Head of Development?), and Donna “Stan” Stankowski (Chief Scientist?), are driving along when then see a crowd gathered in front of the local TV store where everyone is watching the news. The pull over and join the crowd President Kennedy is talking to the nation about a Russian ship near Cuba: “We believe it’s illegal Cuban sugar meant to destabilize a balanced breakfast.”

Kellogg: “I tell you the wife’s a knockout.”
Cabana: “Yeah”
Stankowski: “I got the boys working on the Jackie O thing.”
Kellogg: “Her last name’s ‘Kennedy.’ What does that even mean?”
Stankowski: “No, the shape of the cereal’s an O. Cheerios, Oreos, Jackie-O’s.”
Woman in the crowd: “Shussh! He’s talking about nuclear war, you idiots.”

And there you have it. Ten minutes after the seed was planted in the oval office, the Jackie O gag is brought to fruition. And they had to have lines in the script that told us what the joke was about.

But to what point? What does that gag have to do with Pop-Tarts? Well, Cheerios is a cereal, Jackie-0’s would be a cereal. And Pop-Tarts is being positioned as a cereal substitute. That’s the connection.

I wonder just how, in the process of developing the script, that gag got created. Perhaps the Cuban Missile Crisis bit came first, as a way of amplifying the rivalry between Kellogg’s and Post, and the Jackie O bit was tacked on because, why not? It’s such a little thing. But I could almost imagine that it went the other way. One of the writers, Seinfeld or one of his partners (Spike Feresten, Barry Marder, and Andy Robin) came up with the idea of a cereal connection between “Jackie O”, Cheerios, and Pop-Tarts, and the whole Cuban sugar connection was then worked into the plot so the Jackie O gag can be used. It really doesn’t matter.

What does matter is that Unfrosted is full of gags and references that are locally funny and clever, but that don’t build a compelling overall story-arc. Perhaps the largest such intrusion is the climactic scene where the mascots go on a rampage against the Kellogg’s building while the brass are inside overseeing the certification of Pop-Tarts. Tony the Tiger leads the charge and he’s costumed with Viking horns like the so-called QAnon shaman. Many of the shots in the scene parody shots from video footage of the attack on the United States Capitol on January 6, 2021.

Why? That event is six decades after the launching of Pop-Tarts and has absolutely nothing to do with it. Let the mascots attack the Kellogg’s building, but why distract the audience by forcing us to recall the various video shots they saw? “I saw that, and that one, I recognize that shot...” And so forth. If it’s intended to be a parody of January 6, well, it doesn’t play that way. Nor does it somehow inform or amplify the underlying story of corporate competition. Rather, it’s just management vs. labor that’s been tacked onto that story. To what end?

The Marx Brothers didn’t have me asking such questions of Duck Soup. If you want movies from the 1960s, what about Dr. Strangelove (1964) or Putney Swope (1969)? True, they’ve very different from Unfrosted. But both are comedies, though Strangelove is rather grim. And they are tightly focused. And gags? What about Slim Pickens waving his cowboy hat and riding the bomb? And if you want a comedy that makes effective use of children, how about Galaxy Quest from 1999? If Unfrosted has just turned the Butchie&Cathy line up to eleven, bringing them more closely into the action like the (somewhat older) kids in Galaxy Quest, that might have helped bring more coherence to Unfrosted. 

More later.

Just Coffee

A look at A.I. Hype

Julia Angwin, Will A.I. Ever Live Up to Its Hype? NYTimes, May 15, 2024. 

...some of A.I.’s greatest accomplishments seem inflated. Some of you may remember that the A.I. model ChatGPT-4 aced the uniform bar exam a year ago. Turns out that it scored in the 48th percentile, not the 90th, as claimed by OpenAI, according to a re-examination by the M.I.T. researcher Eric Martínez. Or what about Google’s claim that it used A.I. to discover more than two million new chemical compounds? A re-examination by experimental materials chemists at the University of California, Santa Barbara, found “scant evidence for compounds that fulfill the trifecta of novelty, credibility and utility.”

Meanwhile, researchers in many fields have found that A.I. often struggles to answer even simple questions, whether about the law, medicine or voter information. Researchers have even found that A.I. does not always improve the quality of computer programming, the task it is supposed to excel at. [...]

And consider for a moment the possibility that perhaps A.I. isn’t going to get that much better anytime soon. After all, the A.I. companies are running out of new data on which to train their models, and they are running out of energy to fuel their power-hungry A.I. machines. Meanwhile, authors and news organizations (including The New York Times) are contesting the legality of having their data ingested into the A.I. models without their consent, which could end up forcing quality data to be withdrawn from the models.

There's more at the link.

Pop-Tarts, they aren’t that good [Media Notes 119 D]

I’ve been working away on Unfrosted and have been having trouble bringing a long article to a close. I came across an interesting review by Owen Gleiberman in Variety. Here’s the third and fourth paragraphs:

As a kid growing up in the late ’60s and ’70s, I confess that I never understood Pop-Tarts. My family would buy them, and every so often I would put one in the toaster, wanting it to be a tasty treat. Such is the power of advertising that I always thought it was my fault that I found Pop-Tarts to be…just okay. Twinkies, by contrast, were junky but succulent. And even good old dry cereal, when you were in the mood for it, was pretty great — the delicate crunch of Rice Krispies, the sweet-milk-bath rapture of Sugar Frosted Flakes. To me, though, Pop-Tarts never lived up to their billing. They were bland when untoasted (though a lot of folks ate them that way). Once you toasted them, the hot fruit filling had a soothing tasty tang, but the rectangular pastry was still cardboard pie crust. It wasn’t awful, but it’s not like biting into it gave you a rush of joy. Prefab and a little dull, the Pop-Tart was a “product of the future” that seemed stuck in the past, like astronaut food.

I bring all this up because “Unfrosted” treats the origin story of the Pop-Tart with such a derisive, backhand flippancy that it’s not at all clear what Jerry and his team of screenwriter-producers [...] actually think of the Pop-Tart. Is the movie a goof because they’re making fun of what a mediocre product it was? Perhaps. Yet if the memory of Pop-Tarts actually strikes a chord of Proustian reverence in Jerry — if it’s his madeleine stuffed with fake-fruit chemicals — then why make such a misanthropic satire of it?

YES!

I’m a bit older than Seinfeld. I remember the Cuban missile crisis. And I remember Pop-Tarts.

I don’t think I ever ate one. We certainly didn’t keep them around the house. My mother prided herself on her pastry skills too much to put up with Pop-Tarts. Of course, making good pastry takes time, so we didn’t have it often, but her Danish, from a recipe she got from my grandmother...to die for! Anyhow, I actually went out and bought a package of Pop-Tarts, mostly do I could get a photo or two. They’re not that good. Putting some jelly on toast isn’t hard – ordinary jelly and standard supermarket white bread, nothing artisanal – and it tastes better.

So I’m with Gleiberman on Pop-Tarts in and of themselves. And I agree with him on that second paragraph as well. It’s not clear what Seinfeld’s attitude toward them actually is. The whole thing is a bit scattered and unfocused. I think Gleiberman’s reference to Proust is spot-on. As my friend David Porush just remarked (on the phone), this is very much an act of memory. Seinfeld is remembering the old days, but framing them in a series of ironic comic miniatures to keep them from being too seductive.

Brain encoding models that can transfer across language and vision

Jerry Tang, Meng Du, Vy Vo, VASUDEV LAL, Alexander Huth, Brain encoding models based on multimodal transformers can transfer across language and vision, Advances in Neural Information Processing Systems 36 (NeurIPS 2023) Main Conference Track

Abstract: Encoding models have been used to assess how the human brain represents concepts in language and vision. While language and vision rely on similar concept representations, current encoding models are typically trained and tested on brain responses to each modality in isolation. Recent advances in multimodal pretraining have produced transformers that can extract aligned representations of concepts in language and vision. In this work, we used representations from multimodal transformers to train encoding models that can transfer across fMRI responses to stories and movies. We found that encoding models trained on brain responses to one modality can successfully predict brain responses to the other modality, particularly in cortical regions that represent conceptual meaning. Further analysis of these encoding models revealed shared semantic dimensions that underlie concept representations in language and vision. Comparing encoding models trained using representations from multimodal and unimodal transformers, we found that multimodal transformers learn more aligned representations of concepts in language and vision. Our results demonstrate how multimodal transformers can provide insights into the brain’s capacity for multimodal processing.

Street lamp and lots of other stuff around it

The Matrix [Media Notes 120]

I saw it when it came out, and perhaps I saw in again on some streaming service, of maybe I got it on Netflix back in the DVD days. Don’t know what I thought of it back then. But now (on Netflix)...YIKES!!

I’m thinking that The Matrix is mostly a pretext for a Sci-Fi urban martial arts computational tentacle-porn pseudo-mystical mashup. Really? I’m sure I didn’t take it seriously, for whatever meaning of “seriously” makes sense in this context. It’s all about the action and the tidbits of mystical-mumbo jumbo about belief and self. And the vibe, it’s about the vibe.

And the mystical power of true love? It all comes down to a woman’s kiss? Trinity kisses Neo, All Better. The Oracle said so. He is the one! And soaring brass! Gimme a break!

* * * * *

On action, see my post, John Wick 1, 2, 3 [Media Notes 116], for some discussion.

* * * * *

This one film gave birth to three more features, one of which had a bit role for Cornel West, and a whole lot more. I’ve watched the first two sequels, but not the fourth. Apparently, there’s a fifth in the works.

Here’s a paragraph from the Wikipedia article on the franchise:

The series features a cyberpunk story of the technological fall of humanity, in which the creation of artificial intelligence led the way to a race of powerful and self-aware machines that imprisoned humans in a neural interactive simulation — the Matrix — to be farmed as a power source. Occasionally, some of the prisoners manage to break free from the system and considered a threat, become pursued by the artificial intelligence both inside and outside of it. The films focus on the plight of Neo (Keanu Reeves), Trinity (Carrie-Anne Moss), and Morpheus (Laurence Fishburne and Yahya Abdul-Mateen II) trying to free humanity from the system while pursued by its guardians, such as Agent Smith (Hugo Weaving, Abdul-Mateen II, and Jonathan Groff). The story incorporates references to numerous norms, particularly philosophical, religious, and spiritual ideas, among others the dilemma of choice vs. control, the brain in a vat thought experiment, messianism, and the concepts of interdependency and love. Influences include the principles of mythology, anime, and Hong Kong action films (particularly "heroic bloodshed" and martial arts movies). The film series is notable for its use of heavily choreographed action sequences and "bullet time" slow motion effects, which revolutionized action films to come.

Will this multimedia franchise go down in history as the end of civilization as we know it? The hinge of history heralding the Singularity (properly understood?). Who knows? (Who cares?)

Body part names across cultures

Variability in human body vocabularies. A team of linguists from @MPI_EVA_Leipzig and @UniPassau has conducted the first large-scale comparison of body part vocabularies across 1,028 #languages, using the #Lexibank database. See: https://t.co/YJfIGvtIUH & https://t.co/X81jSDweAY pic.twitter.com/WH4qH8je9z

— MPI-EVA Leipzig (@MPI_EVA_Leipzig) May 13, 2024

Abstract for the linked article:

Every human has a body. Yet, languages differ in how they divide the body into parts to name them. While universal naming strategies exist, there is also variation in the vocabularies of body parts across languages. In this study, we investigate the similarities and differences in naming two separate body parts with one word, i.e., colexifications. We use a computational approach to create networks of body part vocabularies across languages. The analyses focus on body part networks in large language families, on perceptual features that lead to colexifications of body parts, and on a comparison of network structures in different semantic domains. Our results show that adjacent body parts are colexified frequently. However, preferences for perceptual features such as shape and function lead to variations in body part vocabularies. In addition, body part colexification networks are less varied across language families than networks in the semantic domains of emotion and colour. The study presents the first large‐scale comparison of body part vocabularies in 1,028 language varieties and provides important insights into the variability of a universal human domain.

Pier 14 in Hoboken

Ethan Mollick on AI Superintelligence

Ethan Mollick has an interesting post at One Useful Thing: Superhuman? His third paragraph:

No matter what happens next, today, as anyone who uses AI knows, we do not have an AI that does every task better than a human, or even most tasks. But that doesn’t mean that AI hasn’t achieved superhuman levels of performance in some surprisingly complex jobs, at least if we define superhuman as better than most humans, or even most experts.

He goes on to review various tests and benchmarks, concluding in aggregate:

So there does seem to be some underlying ability of AI captured in many different measures, and when you combine those measures over time, you see a similar pattern - everything is moving up and to the right, approaching, often exceeding human level performance.

Zoom out, and the pattern is clear. Across a wide range of benchmarks, as flawed as they are, AI ability gains have been rapid, quickly exceeding human-level performance.

The following remarks are from the section, Alien vs. Human:

The increasing ability of AI to beat humans across a range of benchmarks is a sign of superhuman ability, but also requires some cautious interpretation. AIs are very good at some tasks, and very bad at others. When they can do something well - including very complex tasks like diagnosing disease, persuading a human in a debate, or parsing a legal contract - they are likely to increase rapidly in ability to reach superhuman levels. But related tasks that human lawyers and doctors perform may be completely outside of the abilities of LLMs. The right analogy for AI is not humans, but an alien intelligence with a distinct set of capabilities and limitations. Just because it exceeds human ability at one task doesn’t mean it can do all related work at human level. Although AIs and humans can perform some similar tasks, the underlying “cognitive” processes are fundamentally different.

What this suggests is that the AGI standard of “a machine that can do any task better than a human” may both blind us to areas where AI is already better than a human, and also make humans seem more replaceable than we are. Until LLMs get much better, having a human working as a co-intelligence with AI is going to be necessary in many cases. We might want to think of the development of AGI in tiers:

Tier 1: AGI: “a machine that can do any task better than a human.”

Tier 2: Weak AGI: at this level, a machine beats an average human expert at all the tasks in their job, but only for some jobs. There is no current Weak AGI system in the wild but keep your eyes on some aspects of legal work, some types of coaching, and customer service.

Tier 3: Artificial Focused Intelligence: AIs beat an average human expert at a clearly defined, important, and intellectually challenging task. Once AI reaches this level, you would rather consult an AI to get help with this matter than a random expert, though the best performing humans would still exceed an AI. We are likely already here for aspects of medicine, writing, law, consulting, and a variety of other fields. The problem is that a lack of clear specialized benchmarks and studies means that we don’t have good comparisons with humans to base our assessments of AI on.

Tier 4: Co-Intelligence: Humans working with AI often exceed the best performance of either alone. When used properly, AI is a tool, our first general-purpose way of improving intellectual performance. It can directly help us come up with new strategies and approaches, or just provide a sounding board for our thoughts. I suspect that there are very few cognitively demanding jobs where AI cannot be of some use, even if it just to bounce ideas off of.

I think Tier 3 and Tier 4 is where we’re headed. He goes on to remark:

Even though tests and benchmarks are flawed, they still show us the rapid improvement in AI abilities. I do not know how long co-intelligence will dominate over AI agents working independently, because in some areas, like diagnosing complex diseases, it appears that adding human judgement actually lowers decision-making ability relative to AI alone. We need expert-established benchmarks across fields (not just coding) to get a better understanding of how these AI abilities are evolving. I would love to see large-scale efforts to measure AI abilities across academic and professional disciplines, because that may be the only way to get a sense of when we are approaching AGI.

Here's a comment I made in response to the post:

For what it's worth, I strongly suspect AI experts are, shall we say, a bit naive in how they think about human ability and put far too much stock in all those benchmarks originally designed to gauge human ability. Those tests were designed to differentiate between humans in a way that's easy to measure. And that's not necessarily a way to probe human ability deeply. Rodney Brooks on The Seven Deadly Sins of Predicting the Future of AI has some interesting remarks on performance and competence that are germane.

I've written an article in which I express skepticism about that ability of AI "expert" to gauge human ability: Aye Aye, Cap’n! Investing in AI is like buying shares in a whaling voyage captained by a man who knows all about ships and little about whales.

More recently, I've taken a look at analogical reasoning, which Geoffrey Hinton seems to think will confer some advantage on AIs because they know so much more than we do. And, yes, there's an obvious and important way in which they DO know so much more than individual humans. But identifying and explicating intellectually fruitful analogies is something else. That's what I explore here: Intelligence, A.I. and analogy: Jaws & Girard, kumquats & MiGs, double-entry bookkeeping & supply and demand.

And here's a comment I'm about to post:

On competence "across academic and professional disciplines," I've been interesting in ChatGPT's ability at interpreting films. Here's a piece where I manage to prompt it to a high-school level interpretation of a film: Conversing with ChatGPT about Jaws, Mimetic Desire, and Sacrifice. I hazard to guess at what will be required for an AI to reach professional level, and if we're talking about film interpretation rather than interpreting novels and poems, well, that will require an AI that can actually watch and understand what's happening in a film. I have no idea what that will require.

More recently I've considered the question of using an AI to determine the formal structure of literary texts. It's not rocket science, but it's tricky because it requires a kind of "free-floating" analytic awareness. I'm not sure how well an AI can approximate that with lots of compute.

The popular cover song died two decades ago

The cover song is dead pic.twitter.com/VQ2KNuB6Lv

— Chris Dalla Riva (@cdallarivamusic) May 11, 2024

My mother loved irises

Aesthetic preferences, political preferences, and cultural rank

From TheMoneyIllusion:

The arts are often viewed as being in some sense “liberal”. This could mean many different things. Art might make people more liberal. Liberals might be more likely to make art. Liberals might be more likely to appreciate art.

I don’t know enough about music to comment, but I have noticed that liberals are more likely to appreciate the visual arts. Here’s Psychology Today:

We already know from prior studies that conservatives prefer simple representational art over abstract art, traditional poetry over the avant-garde, and music that is simple, familiar, and ‘safe’.

I am not going to argue that abstract art is better than representational art—indeed most of the very best paintings are representational. Instead I’ll argue that the appreciation of abstract art is usually associated with a stronger attraction to art in general.

Consider a random sample of people that go to a museum show of abstract art, say a Klee or Kandinsky exhibit. Those people are also much more interested in representational art than the average person. They’d be far more likely to attend a representational art museum show (say Monet or Caravaggio), as compared to a random person that did not like abstract art. Abstract art is difficult, and a strong interest in abstract art is usually associated with an intense interest in the visual arts in general.

Again, I’m not arguing that abstract art is better (I like it a bit less, on average). Rather my claim is that liberals tend to have a stronger preference for the visual arts in general. I have no idea why.

I think it's a matter of cultural rank. People of the left are more likely to hold Rank 3 and even Rank 4 cultural values than those on the right. See the chapter, "Politics, Cognition, and Personality" in David Hays, The Evolution of Technology through Four Cognitive Ranks. An explicit argument is needed to get from the material in that chapter to the point I am suggesting in this post, but I don't have time to make that argument. You can find ideas suitable for such argument in the paper Hays and I wrote together, The Evolution of Cognition, and papers we wrote separately, Hays on The Evolution of Expressive Culture and me on The Evolution of Narrative and the Self. There may be material as well in some of my many posts on cultural rank.

A source of blue dye in rural China

Around the corner at Language Log Victor Mair quotes from a recent paper, E.J.W. Barber, "Of Salt Men and Cloth: The Remarkable Textile History Preserved in Eurasian Salt-beds," Sino-Platonic Papers, No. 345, My 2024.

Sinologists have long wondered why the words for “blue” and “cabbage” in Chinese are homonyms: both 藍 lán in Mandarin. But just recently, perusing dye information about woad [VHM: a common plant dye for blue] from Richard Laursen, Victor Mair noticed that woad is actually in the cabbage family, Brassicaceae (earlier called Cruciferae), and that rural people have long found ways to get blue coloring out of a number of types of cabbage (Mair 11/22/2023), especially the purple kind. Hence the unexpected homonyms. Thus, from all these textiles preserved in salt, we even have the solution of an interesting etymological conundrum.

The cabbage in the photo above is from the Lafayette Community Learning Garden in Jersey City, started by June Jones of the Morris Canal Redevelopment Area CDC. It ran for two years, 2011-2012. The post at the following link has more shots of purple cabbage.

If language is for communication, what does that imply about LLMs?

Noam Chomsky famously believes that language originated to facilitate thought, but then came to be a medium of communication. Others believe the reverse, that it originated as a facility for communication which turned out to facilitate thinking. That is certainly my view.

If that is so, then one would think that language is structured to facilitate communication. Communication is serial, one token at a time, token after token. That would imply that language is structured to facilitate next-token prediction. That would in-turn imply that the relational structure of semantics would evolve to facilitate mapping between the linear structure of the language string and the multidimensional structure of meaning. You want to be able to efficiently project multidimensional semantic structure onto a string and to reconstruction multidimensional semantic structure from a string.

How are LLMs trained? By next token prediction. That is to say, the training regime mirrors the primary communication constraint governing the structure of language. So it is with text generation as well. Language is spoken one token at a time, and so LLMs generate texts, one token at a time.

The tasks that LLMs have trouble with, such as planning and arithmetic, ARE NOT primarily communicative in nature. They are tasks for thought, for reasoning.

Only the fork is left, and some syrup

It looks like China is eating America's EV lunch. And Elon is MIA.

Kevin Williams, I Went To China And Drove A Dozen Electric Cars. Western Automakers Are Cooked, InsideEVs, May 9, 2024.

Late in the article, after looking at a number of very well-made Chinese EVs at auto shows in Shanghai and Beijing:

Here I was in China, trying to empathize with Western brands, thinking they were being pushed out of China due to politics and things that were no fault of their own.

In reality, it felt like it was the late 1980s again, when American manufacturers felt like they could sell whatever underdeveloped models its accounting department had cooked up to the public, and we’d just have to deal with it. Now that I’ve seen a glimpse of what’s going on in China, the Western manufacturers, particularly the American ones, don’t seem like they’re trying at all.

Writer and podcaster Ed Zitron said something interesting during an episode of his podcast, Better Offline. Americans are almost made to apologize for their preferences when it comes to Big Tech. Some bigwig or boisterous startup guy had a big idea for a widget that no one wanted, and decided to half-ass a product that doesn’t work all that well.

When the public rightfully ignores a bad or unwanted product, there’s a new trend in tech to blame the clientele for not being smart enough, rather than facing the music that what was created just wasn’t all that good. I mean, just look at all the terrible AI-based pins that don’t do anything.

And then there's TikTok:

America’s looming TikTok ban feels like a direct allegory for China’s relationship with its electric car exports. I use TikTok; I understand how it works, and I agree that there are plenty of valid critiques to be leveled at the platform’s ability to spread misinformation, or how its endless scroll probably isn’t great for anyone’s mental health, especially that of the teens and tweens who love the platform so much.

Yet, so much of the coverage of TikTok’s ban refuses to acknowledge one fact: The platform is really, really well executed. TikTok’s algorithm is fantastic; it can compile a near-endless scroll of content that feels fresh, positive, fun and eerily, directly targeted to you. I’ve watched the viral power of TikTok straight up create music artists like PinkPantheress, or revive the career and launch classic artists like Sophie Ellis-Bextor or Kate Bush back onto the charts.

TikTok’s culture isn’t perfect, but it’s a hell of a lot healthier than whatever Meta, Google, and Twitter have created, where death by a thousand cuts of “enshittification” have made their services hostile and less useful to the end user. On Instagram Reels, the content moderation is so poor, that it’s not uncommon to see someone literally die on screen.

So, when automakers, tech companies and regulators push back on China, the sentiments that they’re just protecting our market from unsafe or security-challenged products feel hollow. Instead, it feels like grandstanding, and a tacit admission that they have no intention of trying to do better.

Instead of competing, they’d rather just shut out competition entirely. The concerns about cybersecurity don’t address the elephant in the room here: Your product sucks, compared to what China is putting out now. It doesn’t go as far. It’s not as well-made. It’s not as nice. It’s not as connected.

Western automakers aren’t entangled deeply with tech companies in ways that would serve the end user, Chinese or otherwise. They didn’t get way ahead of the curve to establish a battery supply chain in the ways China did. And they don’t seem to want to cater to the Chinese market (or any market, rather) through continuous updates and agility with their product line.

Even Tesla in China can’t be bothered to update one of its most important products, the Model Y, in this hyper-competitive market. Instead, it relies on margin-hurting gimmicks to move units, like constant price cuts, subsidized trade-in incentives, and 0% financing to get customers to buy a car that is aged and now uncompetitive.

Tesla didn’t even have a presence at the Beijing Auto Show. Elon Musk came and went to Beijing during the show, only to make a case for his robotaxi pivot with government officials. It’s like he’s already given up on cars here.

There's more at the link.

MUSE - "Hysteria" - KIDS Collaboration Cover

Produced by Kids Rock For Kids
www.kidsrockforkids.com

KRFK brought together extraordinary young performers for a mind-blowing collaboration!
FOLLOW and SUBSCRIBE to @kidsrockforkids for more incredible content
This first of a series of music videos including these future legends!!

Cover: Hysteria by Muse @muse

Vocals: Vinya Chhabra • age 13 from New Jersey
YOUTUBE @vinya.chhabra
INSTAGRAM @vinyachhabra

Bass: Ellen Alaverdyan - age 11 from Las Vegas
YOUTUBE @ellenplaysbass
INSTAGRAM @ellenplaysbass
FACEBOOK @ellenplaysbass

Drums: Yoyoka Soma - age 14 from Japan
YOUTUBE @yoyoka_soma
INSTAGRAM @yoyoka_drums
FACEBOOK @yoyoka_soma
Spotify @yoyoka

Guitar: Bay Melnick Virgolino (aka The Only Bay) - age 9 from NYC
YOUTUBE @TheOnlyBayMusic
INSTAGRAM @the_only_bay

Guitar & Backing Vocals: Charlotte Milstein - age 14 from San Diego
YOUTUBE @charlottemilsteinguitar
INSTAGRAM @charlottemguitar
TIKTOK @charlottemilsteinguitar

Backing Vocals: Ella Milstein - age 16 from San Diego
INSTAGRAM @ellairismusic

Shoutout to the music video tech team!!
Video: Brian Tweedt @briantweedt
Video: Rob Cammidge @cammidge
Audio Mix: Hovak Alaverdyan @hovakalaverdyan
Music Director: Josh Margolis @gowanusmusicclub
Video Edit: @henry_xie0902
Recorded and Filmed at Tropico Union Los Angeles, CA @tropicounion
Cover Graphics: Cara Radom @csradom

KIDS ROCK FOR KIDS is a NYC-based nonprofit that brings together exceptionally talented and passionate young musicians from around the world – giving them the opportunity to do what they love and connect with their kindred spirits in music, while helping kids in need.

KRFK - Helping the next generation help the world!

Those scamps at Hydraulic Press Channel are at it again: Extruding gummy bears!

From the YouTube page:

In this exciting video, we use a hydraulic press to conduct a series of experiments that explore the incredible power of hydraulic pressure. We start by pushing an orange through a 2.5mm diameter hole, then work our way down to smaller holes, including 1.5mm and 1mm, to see what objects like gummy bears, soap and candles can be pushed through. Our slow-motion footage captures the explosive results of each experiment, revealing just how small of a hole a hydraulic press can push objects through. Don't miss this fascinating look at the power of hydraulic pressure and the incredible things it can do.

A look back at winter

The Mexican drug cartels are a bit "like junior partners to corrupt government officials"

Samo Burja, Mexico’s Drug Cartels Are Not Competitors to the State, Bismarck Brief, May 8. 2024.

As of early 2024, despite the incarceration of leading cartel figures such as Joaquín "El Chapo" Guzmán Loera, the organization he headed, the Sinaloa Cartel, remains the dominant cartel in Mexico and is also an increasingly powerful force in drug networks across the world. Its main competitor is the Jalisco New Generation Cartel (CJNG) and the two often engage in violent competition, alongside smaller cartels like the Gulf Cartel, the Juarez Cartel, the La Familia cartel, and many more local criminal organizations. In 2017, Americans consumed $153 billion worth of banned narcotics.4 The cartels satisfy a large fraction of this demand. There are no precise estimates of cartel revenues and profits, but it is likely that annual revenues are in the low tens of billions of dollars and profits total several billion after the costs of business, including bribes. The cartels also generate revenue from other criminal activities like human trafficking, extortion, and even illegal logging.

Around the world, such criminal activities have shown to be lucrative enough and resilient enough to state persecution to fund rebellions that could topple governments. For example, the Marxist FARC guerillas in Colombia, as well as multiple generations of Taliban rebels in Afghanistan—first fighting the Soviets, then the U.S.—were funded in this way. Because of the drug war, ongoing violence, and continued influence of cartels in Mexican society, Mexico has sometimes been described as a failed state and some U.S. politicians, such as former President Donald Trump and Republican Senator Tom Cotton, have even called for taking unilateral military action against the cartels, as was done against ISIS, the short-lived Islamist statelet in Iraq and Syria.5

But Mexico’s cartels are not ideologically or politically-motivated groups making the jump to crime to fund their activities. They are rather amorphous criminal groups motivated by profit-seeking, usually relying on familial and regional ties. From a business perspective, it is preferable to collaborate with the government when possible, rather than invite anarchy. Since, through bribery, the cartels represent an important source of revenue for Mexico’s elites, this interest is mutual.

As a result, the cartels are far more like junior partners to corrupt government officials rather than an independent and competing force of their own, though their allegiances have ebbed and flowed from the state level to the federal level—Mexico is a federation of united states—and seemingly back over the last sixty years. This makes Mexico’s cartels clients of the Mexican state, not its competitors, and, in turn, Mexico’s status as a client of the U.S. explains why the cartels continue to flourish and why there is unlikely to be any U.S. intervention in the near future.

And: "The drug trade is powerful in Mexico not because of its size, but because of its liquidity, anonymity, and informality, which makes it easy to enrich particular individuals."

There's much more at the link.

H/t Tyler Cowen.

Friday Fotos: Flowers & flowers and more flowers