Artist trained rats to take selfies to make a point about social media | Hacker News Reader

Artist trained rats to take selfies to make a point about social media

cbc.ca

126 pointspseudolus90 comments

It was a serious oversight to not have the rats take selfies against a green background.

Agreed! Nothing that can't be easily remedied though! https://www.remove.bg/

Is it still called a "selfie" if the rats don't get to see the resulting pictures? The dictionaries I consulted define "selfie" as the photographic output, and not the act of taking pictures.[1][2]

I don't see any display device in that setup, so it seems the rats are just pressing buttons for sugar, independent of any camera activity.

[1] https://www.oed.com/dictionary/selfie_n

[2] https://www.merriam-webster.com/dictionary/selfie

Edit: Looks like I just didn't see the screen but it's there.

> In Lignier's box, when Arthur and Augustin pushed the lever, a camera would snap their picture and display it on a screen in front of them.

The rats do see the pictures.

When I was.a kid, I "learned" from some authority (TV? book?) that cats and dogs could not recognize process two dimensional images, and that a TV was therefore just visual noise to them.

Same here and while that's demonstrably not true (the explanation I heard was something about the colors and refresh rate so maybe this was true at some point in the past) the article does state the rats don't recognize themselves - that's different from recognizing the picture or being able to see it.

That said, it's really more of an art project than anything to do with science as the article makes it clear the rats were trained to push buttons to get sugar, the selfies were just artistic flourish to make a (fairly overt) point.

> That said, it's really more of an art project than anything to do with science as the article makes it clear the rats were trained to push buttons to get sugar, the selfies were just artistic flourish to make a (fairly overt) point.

according to the article, this was just to get them hooked at first, but they kept doing it later:

> After the rats were trained to push the level for sugar, Lignier changed the experiment's parameters. Sometimes taking a picture would yield a piece of sugar, and sometimes would not.

> Nevertheless, the rats kept on jamming the button, taking dozens of selfies. After awhile, he says they largely stopped bothering to eat the sugar, even when it came out.

Same here. Obviously this is not true.

A ton of "science" taught to kids through seemingly legitimate channels is in fact just the off the cuff intuitions of some thoroughly average adult, and adults on average tend to be staggeringly uninformed on most topics.

EDIT: And even more so for history.

It may have been true during the days of CRTs. I have had a lot of pets and don't remember any of them showing interest in the televisions whatsoever, other than a source of warmth. And that is definitely no longer true.

The observable non-reaction was true, what was not true was the cause, which is the low frame rate.

This is the problem with this kind of thing so often, people observe something and an explanation that sounds completely logical gets spread around.

I remember a friend's cat following the mouse pointer on an Amiga with a CRT monitor. When moved off the bottom of the screen, the cat would go right up to the screen and try to peer down to see where it had gone.

Not sure about dogs, but cats definitely could see a plain 60Hz CRT display... atleast some of them anyway.

Sure but a cat would probably have the same reaction if an arrow behaved like that irl, it's not the same as recognizing something as a living being due to how it moves.

I had a dog growing up that insisted on watching The Price Is Right every weekday morning and she'd bark at you until you changed the channel. Then she'd push a stool in front of the TV, sit on it, and watch Bob Barker for an hour and when the show was over she'd get up and leave.

Something about a dog liking to watch someone whose last name is Barker makes this a great story!

I worked for a TV production company, and we made a show called Kitten TV, in which each episode was 45 minutes of kittens playing in different custom built themed playsets.

On Instagram we received hundreds of pics submitted by viewers of their cats watching Kitten TV, it became a meme.

It may be that at least part of the story was that some animals had issues purely with CRT technologies. A shift from shooting electrons through a window, to source pixels of very specific frequencies, may have been better.

For example, this spectral comparison implies there is a rather large difference in the color experience of watching CRT vs LCD. Very pronounced and spikey red component vs smooth gaussians. Canines are notorious for issues with red colors.

https://www.researchgate.net/profile/Alireza-Shahsafi/public...

My border collie dog was definitely interested by any sheep appearing on my old CRT.

Assuming you looked up the current info on that, according to a quick Google current research all seems to show they can recognize familiar faces in photos.

When I was a kid I was taught by every adult that "humans cannot discern more than 24 fps".

Except that we could all immediately tell if a Commodore Amiga demo was running at 25fps or 50fps and groups that'd have the frame rate drop from 50 to 25fps on some frames were "lame".

So I knew it was obviously false that humans couldn't discern when something was running at more than 24 fps.

Same when little Windows 95 utilities allowed to bump the framerate from 60 to 72 or 75 Hz. I remember those: they'd enhance the experience. Yet everybody was still telling me: "Why do you even bother, 60 Hz is enough, humans cannot discern more than 24 images per second".

Same for cats and dogs and the TV of course. Which I knew was wrong because on my Commodore Amiga I'd take a big circle, filled with a solid color, and make it bounce across the screen and watch my cats' reactions.

People can be really extremely dumb, even in the face of evidence.

The claim that nobody could discern more than 24 fps is an odd one (and not one I've ever heard), since it's been generally known that's not true even back before 24 fps was chosen as a standard. Originally some (including Edison) wanted to standardise on higher frame rates for better motion. 24fps was basically compromise to balance against the amount of film used, being fairly close to the minimum speed that was considered satisfactory for motion.

Even then, a two bladed shutter was used to flash the image on the screen twice for each frame, and more modern projectors used a three-bladed shutter to reduce flicker.

> being fairly close to the minimum speed that was considered satisfactory for motion.

Basically that: It's a repeated and extremely common misunderstanding of the actual claim, that 24fps is right about the threshold between interpreting a series of images as motion instead of a series of images.

A lot of people assume that once you see it as motion, there's nothing more that can be gained.

Fake knowledge was the most common thing circulating among people back then.

People mostly couldn't verify anything.

> People mostly couldn't verify anything.

May I inquire: How do we verify things today that we did not back then?

The literature is all there on the internet. If you want to repeat the experiments, the instructions are all there on the internet. If you want to look at the theory from what the fact was derived, it's there on the internet.

Approximately nobody had access to anything before it.

Indeed, I think that you can definitely see the difference between 25 and 50 fps... even 24/25 fps and 30 fps don't exactly look the same.

The 60/72/75/... Hz (some screens could display 85 or even 90 Hz) of the Windows 95 epoch was the refresh rate of the CRT tube: the higher, the less flicker. Once you tried 75 Hz, you clearly didn't want to go back to 60 Hz (or at least, that was my experience).

i can feel 144-200hz. above that i cant tell anymore, but probably pixel response times arent good enough yet on consumer hardware to provide a real test to me. 60hz feels awful to me. side note, i can discern stutters at 144hz + when i take a deep breathe.

Do you actively practice being able to discern high frame rates, or is it a "born with skill?" I often wonder if FPS gamers self select for those who can discern high frame rates, which then drives screen sales. Do you attempt to increase your frame rate perception?

Do you find normal movies in the theater difficult to watch because of your "high frame rate?" If I could see gaps in 144 Hz, then it seems (totally imagining with math), like it would be the equivalent of a 4 Hz frame rate for "normal" eyesight. 24 Hz / 144 Hz ( 24 Hz normal movie) = 4 Hz. 4 Hz looks super choppy bad. Seems like it would turn into the Flash trying to watch a human movie.

Course, I asked colorblind people about colorblindness, and it was totally crazy all the different colorblindness types they talked about. So could be bad mental picture.

In the theatre no, but scrolling shots on a fast panel can be particularly bad. There is no smoothing because the pixel response times are fast. I perceive a reverse motion of scrolling shots. If there is a slow scanning shot, say from left to right, after each frame stops, my eyes trick me that the whole image moves back to the left about an 8th of its rightbound motion. Its very mildly annoying, but I still watch though. I swear i can hear the frames stopping.

Many people have noted my particularly small head. But my reaction time is average. So time perception and reaction time may not be related.

Thanks, that gives me a much better idea about the experience from somebody else's perspective. A blog post or something would be interesting, if you have one on the subject. However, I may be a rather limited audience. Apparently up above somebody's saying 500 Hz tends to be the cutoff, yet I suspect there are large portions of the population with significant variations.

Like with colorblindness/synesthesia, I'm constantly surprised by all these different people I talk to who experience everything differently. Would not have guessed it would cause reverse motion scrolling. The eye exams and charts always made it sound like 20/20 at "normal speed" was correct, and the only change from correct was bad. Sounds like y'all just look at something different.

This is basically just a misunderstanding. There are multiple overlapping effects. Most of the 'hard' numbers for claims along those lines come from the flicker fusion frequency [0] around 50 Hz or so, which means that a video recorded at 25 fps and replayed by flashing every frame twice (like in old movie theaters) will not appear to flicker.

However, there's another important factor, which is that if you move an edge accross your visual field and sample at a certain frequency, depending on the speed of the edge, you might get a stroboscope-like effect and see multiple parallel edges. This is something a lot of people notice in 60 Hz computer rendered video, or when moving quickly under a lamp flickering at the mains frequency, because every frame is sharp without any motion blur. By contrast, you can run video filmed at an exposure time similar to the frame time at lower frame rates and it'll look fine, because the motion blur removes that stroboscope-like appearance.

[0] https://en.wikipedia.org/wiki/Flicker_fusion_threshold

There is research on humans showing that we can perceive at 500 Hz. There are devices that try to simulate a color by modulating a single LED (no color filter) and they don’t work on humans until you go past around 1 kHz.

https://www.nature.com/articles/srep07861

Yes, that's what I was referring to. If you have an edge flickering at a high frequency and perform a saccade (fast eye movement) over the edge, your retina will be exposed to the edge at regularly spaced intervals. So the motion will not appear smooth. That's what they call a 'flicker artifact' and I called a 'stroboscope-like effect'.

Importantly though, this is not us humans detecting 500 Hz flicker itself, quite the opposite -- the reason the artifact is visible is that our retina is not sensitive to fast motion, it integrates over a period of time in which the edge appears to be in multiple locations.

What's really crazy to me is that when you watch any film in >24(30?) fps, it looks like trash. I suppose it's because we're used to higher fps being home video? I wonder if films will ever be accepted or made in higher fps. Maybe if VR films actually become a real thing.

Part of the issue with higher fps is that it should be balanced with a shorter frame exposure time, so motion blur matches the frame rate. The rule of thumb is shutter speed should be 2x the fps, so shoot at 30fps, you want the shutter around 1/60s, if you shoot at 60fps, you should use 1/120.

This means anything that adds or remove frames is going to quickly cause the motion to look unnatural or at least different from the original.

Cats and dogs need higher refresh rates to see TV as a movie rather than a slideshow of images: https://biology.stackexchange.com/questions/8921/why-can-cat...

Very interesting. Its like smaller brains have a higher clock speed. global oscillations?

Less processing to be done? Something like a fly has a much lower reaction time than a chicken.

Smaller systems tend to tick over faster. Compare a fly heart with the heart of a mammal and the heart of the child with the heart of an adult. Or an elephant... Neural pathways will be longer and so slower, they may have more synapses in them. Overall it would be highly surprising for a larger system to have a higher clockrate than a smaller one.

It is very easy (and amusing) to prove this wrong by playing a video of birds / fish / other wild animals and let your cat watch :)

Sure, now it is very easy. But back then animals rarely reacted to CRT tvs

depending on your age that may have been true, I think old crt tvs refresh rate made it so that dogs would not be able to perceive the actual image.

Cats and dogs rely on other senses such as smell and hearing a lot more than humans, who are very visually focussed. They just "see" the world differently and might not be that excited about something that mainly revolves around visual stimuli. I suppose a theoretical cat or dog TV would included a lot of smells, which might not be something you'd actually want in your living room...

My sisters dog will bark incessantly at the TV if there are specific animals on the screen, such as horses and elephants. Surprisingly, cats and dogs are allowed.

My aunt’s dog does the same! Horses are not allowed nor are people swimming in any way on TV. The dog also seems to prefer when the actors have British accents.

this brings up complicated questions about perception, and the way that thinking links to senses.

In other words, it is quite possible that dogs see (poorly, without red) very much what we see, but the cognitive ability and then the "thinking" that happens after that, is very different.

What's the copyright situation of rat selfies?

I think it's established that photos taken by animals are not copyrightable works, so you can do anything you want with them, at least in the USA.

That currently applies to photos with animals deliberately triggering the shutter. I would hesistate to say "it's established" without more than one precedent in US case law.

However, candid nature photographs taken with camera traps are copyrighted, and Wikimedia Commons has tolerated this status since 2011.

https://en.wikipedia.org/wiki/Camera_trap

https://commons.wikimedia.org/wiki/Category:Photos_taken_wit...

Surprisingly (or unsurprisingly?) it's not the first time that question has come up.

https://en.wikipedia.org/wiki/Monkey_selfie_copyright_disput...

Note that this also may have implications for AI-generated work.

Here's how, ignoring the many legal theories, it's handled in practice in the US right now:

> On 21 August 2014 the United States Copyright Office published an opinion [..] to clarify that "only works created by a human can be copyrighted under United States law, which excludes photographs and artwork created by animals or by machines without human intervention".

However note that whether it's "copyright free" will depend on the jurisdiction and also that the article is talking about a french artist. While copyright law is very same-ish across the globe (and countries recognize each other's copyright), it often varies in the fine details.

> Here's how, [...] it's handled in practice in the US right now:

Forgive my ignorance of the nuances of US copyright enforcement, but in what form is the opinion of the US Copyright Office important? The US joined the Berne Convention in 1989, which includes automatic copyright protection without the need to register anything. And if you believe somebody violated your copyright you take them to court, which also doesn't involve the US Copyright Office. Unless they release actual regulation I don't believe their opinion on how to interpret existing regulation really matters, or am I missing something?

In the very ultimate limit, because it falls under the same jurisdiction as the USS Nimitz.

> Unless they release actual regulation I don't believe their opinion on how to interpret existing regulation really matters, or am I missing something?

To us their opinion only matters insofar that they are experts on the matter, but I don't think you are really missing anything. Courts are free to decide otherwise.

I believe the US copyright office's opinions are based on precedent set in court and on their own expert understanding of the relevant laws, so you should not read that as them setting policy, but rather as them explaining and summarizing.

However note that the copyright office runs a small claims tribunal for disputes under $30k. If one chooses to make use of that, their opinion do start to matter a bit more.

Tangential, but come on.. how can you not love the lady who "taught her rats to paint"?

https://www.youtube.com/watch?v=7M9us5DQyrY

https://www.youtube.com/watch?v=5NFTMuNxQhY

The connection to selfies is really weak. Its just classical conditioning. The stimulus is the food not the selfie taking, and the continued pressing of the selfie button without food is conditioned response.

Even that is attributing too much. When you give animals food, they tend to do things to get that food, humans being no exception. If you plant a soft-serve machine in the middle of a city square, you can bet people will press the lever to receive free ice cream. Hell, just give out coupons for free anything and people will show up. In either case, if you take away the free thing, people will keep returning in hopes the thing will come back, not just because they've been conditioned but because they want more of the free thing.

Neat little art project, but it's not that insightful when you think outside the 34th percentile.

> not just because they've been conditioned but because they want more of the free thing

That is what literally what "conditioned" means. You're belabouring a point that OP is already making, and then following that up by sneering at the "34th percentile". Maybe the real art is always in the comments section.

> You're belabouring a point that OP is already making

I don't believe I am. I'll try to clarify.

> That is what literally what "conditioned" means.

Yes and no. I'm suggesting that attributing the effect merely to "conditioning" is reductionist. There's some level of conditioning to all biological behavior, and repeating a behavior in response to a reward is not entirely irrational. In principle, it's very rational for rats to keep returning to perform an action they know through experience provided food at some point. The rats in this art project aren't necessarily as thoughtless as conveniently suggested by the conditioning explanation. There's plenty of evidence that rats have the ability to remember details about the past and plan for the future.

> and then following that up by sneering at the "34th percentile".

I didn't mean it that way, and in retrospect I admit shouldn't have said that. It was rude but that's truly not the attitude I had in mind. Apologies to whom I was replying to.

When you give human likes...

The selfie taking draws a parallel between rats in a Skinner box and humans in social media. The point is to make people wonder whether social media companies view their users the same way Skinner viewed his rats in a box: dumb, and able to be manipulated with meager and cheap rewards.

It is not a functional piece of the "experiment", but I don't think it was ever meant to be functional or a real experiment. The artist says the rats don't recognize themselves, so I would assume they predicted the selfies would do nothing.

The selfie taking is the only interesting part of the piece, imo. Without that, it's just replicating an 80 year old experiment that has already been replicated to death.

It's an artistic remix of the classic Skinner Box experiment. The creator is an artist, not a scientist, and his point is being artistically rather than scientifically.

As a sidenote: Social media has been described as a skinner box a million times before, but expressing it literally like this is, IMO, clever, interesting, and also cute. The idea using scientific experiments as metaphor to make art is cool, and I hope people start to do more things like this.

I don't know, it seems a bit dishonest to imply that the rats wanted the selfies(I got that impression), when all they wanted was pressing the button, because they learned, it would produce treats reliable, and later at least sometimes. If at some point, there never would be treats coming anymore, that would uncondition themself and at some further point, they would not care about that stupid button anymore. Unless of course, they would have nothing else to play with. (Is that also a metapher about FB?)

Did it make you feel something to see this article? Frustration, interest, excitement, anything?

It's art. How you interpret it is sort of the point. It's not science, it's someone trying to make you think, and it seems like you thought in response to it. Successful art then.

I don't see how it's dishonest. The artist is transparent in his methodology, and the metaphor (engagement = treats) is clear. Either way, this is an art piece so it isn't subject to the same kinds of rigor or scrutiny of a scientific experiment.

"For me, they're really like performers, you know? They perform for the camera," he said. "They look cute."

Yeah, it is art. It wants to be interesting nested in many layers. I still don't buy this.

you're allowed to disagree with the message of the piece

Here I disagree and criticize, how it is promoted.

The constant reward is a way to quickly condition a behavior. Random reward has a higher permanence… however, it’s clear that they weren’t doing it for the treats when they started ignoring the treats they received.

Yeah, after they were full, they were just after the conditioned kick. But where they ever looking at their picture, or even had the chance to do so? In my understanding no. So they never cared about the camera, or the picture.

Chase truth. Not coolness. Don't fool yourself.

How am I fooling myself by appreciating a well executed metaphor? I see a lot of comments interrogating an art piece as if it were a scientific experiment, and to me that misses the point. You wouldn't look at Van Gogh's Starry Night and say "Well actually..."

I don’t think it matters. While it would have been interesting if the rats got pleasure looking at themselves, ultimately this is an art project about human society that is meant for human consumption. The fact that we read about it means it was somewhat successful. The issue of whether the rats got a dopamine kick from just pressing the button is irrelevant.

Isn't basic conditioning the point the artist is attempting to make? I interpreted the project as pointing out that social media uses fundamentally similar reward mechanisms as we use in training animals

Yes, and I'm saying I'm finding his point weak. It is a old and well know experiment and the social-media-is-a-skinnerbox talking point is as old as social media. I find the selfie angle tacked on without adding or illuminating the subject further. The work doesn't work for me. But I appreciate the cute rat photos.

This is an odd comment to me.

It's like saying "people have been painting realistic portraits of women for centuries, Leonardo, and this one doesn't add anything or illuminate the subject of portrait-taking further. Still, I appreciate her ambiguous smile."

> Yes, and I'm saying I'm finding his point weak.

This would be a stronger point itself if you didn't succinctly validate it with the second sentence of your initial post!

>> After the rats were trained to push the level for sugar, Lignier changed the experiment's parameters. Sometimes taking a picture would yield a piece of sugar, and sometimes would not.

And this part is using an extinction schedule. When you provide a reward for a behavior consistently and then take away the reward (extinction), the behavior tends to stop immediately. By randomizing the reward and maybe gradually decreasing the frequency you have a much better chance that the behavior will persist without the reward. Because he did this we can't really be sure the selfies are their own reward, since the rats may have been simply trained to push a button.

Rats simply like playing games, I'm not surprised that they thought pressing the button is fun.

But really this is a variant of the mirror test. These rats were freshly adopted and have probably never seen a mirror, and have no idea what they look like. If they were given a mirror for a while before running this experiment, we might see different behavior.

I found a paper [0] suggesting that rats do recognize other rats in mirrors, still photos, and video, but the results aren't very strong. I think the theory is that their eyesight is pretty poor and they rely mostly on scent to identify other rats.

[0] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5849344/

> The connection to selfies is really weak. It's just classical conditioning.

Isn't the idea that the compulsion to take a selfie being classical conditioning the very point of the piece?

The rats were conditioned to push the button. Whether the rats were conditioned to take a selfie is unclear. The sugar is an unconditioned stimulus, the rat's pleasure of receiving sugar is an unconditioned response, and pushing the button is a conditioned response. The button is a conditioned stimulus. Is the selfie a conditioned stimulus as well? Melodramatically put, is the selfie more than a pearl cast in front of a swine?

My killjoy answer is: give the rat two tunnels. The rat can see partway into the tunnel from the entrance. One tunnel has a selfie inside. The other tunnel has a picture of the wall in the background of the selfie. Add more tunnels and check which tunnels the rat prefers.

My artistic answer: give yourself the pleasure of assuming that the rats like the selfies.

Well sure—I'd also like to point out that rats aren't human, if that helps you understand the piece better.

Maybe they continue with pressing the button because if it brings them food maybe it will also free them from Skinner box.

Very true, although this type of conditioning is operant conditioning since the rats are performing a voluntary behaviour as opposed to classical conditioning like Pavlov's dogs.

I guess I don’t really see this as hacker news material. Popular media aggregators tend to promote these kinds of statement art projects. I suppose I’m a little weary and wary of it all. I do not need a sanctioned individual—that we must know an Artist or Philosopher did this—to command some reaction from me on a rather tired topic.

One might argue that the fact we still argue over something shows the need for such art. Perhaps some people could use that slightest nudge. But I find these works often subtly disingenuous, and maybe serve better to reinforce a crowd’s own thinking. Like I heard taking shallow versions of opposing views tends to just strengthen already held beliefs, might some art be the same?

What if the quality and quantity of rewards were linked to the engagement metrics on the selfie posts? Wouldn't that be a better approximation of our current addiction?

* More likes -> more food

* Retweets -> better food

* More followers -> larger cage

The idea that the rats are taking photos of themselves for the (self-aware) pleasure is pretty silly. The rats could have been trained to do anything with this method. Moving from rewarding every time for the desired behavior to occasionally rewarding causes the animal to keep doing the behavior for the reward. And if you stop rewarding entirely, the behavior will continue for a while, and it'll eventually stop. But by then the artist has made their point and doesn't care.

> (self-aware) pleasure

Made me think: when humans take photos on {plataform} to receive likes, doesn't it means that humans know to apreciate what comes after those likes? Money, sex, lots of "friends", relationships, oportunities... The pleasure doesn't come from the post itself, and every like isn't a simple like, it is a whole bunch of things that comes with "I like that from you"

Reply on news.ycombinator.com