Join 3,514 readers in helping fund MetaFilter (Hide)


Simon Singh will mess with your head
July 11, 2011 3:11 AM   Subscribe

A dramatic and shocking demonstration of how your brain gets fooled to see something that is not there because of your biases, prejudices and expectations.
posted by pharm (65 comments total) 37 users marked this as a favorite

 
More whimsical and goofy than dramatic and shocking, I'd say.
posted by robself at 3:26 AM on July 11, 2011 [4 favorites]


Must. Sacrifice. Goats.
posted by item at 3:26 AM on July 11, 2011 [3 favorites]


Perhaps the youtube description was a little hyperbolic robself!
posted by pharm at 3:32 AM on July 11, 2011


Back in the days of my wild and crazy youth, when I finally had a PC powerful enough to do something so amazingly complicated as audio editing (click "reverse", go get lunch, come back to hear the result), I played around with Stairway (and other songs) just for fun.

I never heard the "whole" version Singh played, but I never had any trouble at all hearing the lead-in of "here's to my sweet satan". For that reason, it surprises me more people didn't raise their hand to his first question (if not the second).

Great demo of the power of our minds to superimpose (nonexistent) meaning on chaos, though.
posted by pla at 3:38 AM on July 11, 2011


This has been going around for a while now. More examples.
posted by shoebox at 3:41 AM on July 11, 2011


pla, Why would you be reversing Stairway to Heaven if you weren't already primed to hear satanical messages?
posted by Saddo at 3:42 AM on July 11, 2011 [3 favorites]


The second time I closed my eyes, and even having been told what the lyrics were going to be, I still didn't hear it.

You have to visually see the lyrics light up in order to hear it.
posted by Civil_Disobedient at 3:51 AM on July 11, 2011 [1 favorite]


Party tonight, midnight, in Satan's toolshed. Be there.
posted by caddis at 3:54 AM on July 11, 2011 [10 favorites]


Civil: Ditto. I caught the "It's my sweet Satan" part because he'd said it so many times, but the rest remained gibberish.
posted by Holy Zarquon's Singing Fish at 3:56 AM on July 11, 2011 [1 favorite]


Is toolshed an euphemism for underwear? Because if it isn't, it should be.
posted by maxwelton at 4:01 AM on July 11, 2011 [11 favorites]


Apparently Simon Singh did not experience the "Hell's Bells" video series in junior high.
posted by dubold at 4:05 AM on July 11, 2011


Heard "My Sweet Satan" on the first go-round, watching this, but only because I was primed to hear it by reading "The Dark Knight Returns".

And because I was in middle and/or high school during the height of the Satanic Panic.

My Dad was a game warden at the time, and had a police training seminar about Satanic ritual abuse (I know, I know), and I copped a look at his notebook shortly thereafter (it was open on his desk, in case you think I was snooping through his notebooks or something. The house computer was on the same desk).

Dad was in his late twenties when "Stairway to Heaven" came out, and his Zep records are still in good condition, so it was no surprise to me when I saw (funny, but no surprise) that in Dad's notebook from his Satanism training, I saw the "offending" lines from "Stairway to Heaven" written in his notebook, and then written backwards, followed on the next line by the word "bullshit".
posted by Mister Moofoo at 4:13 AM on July 11, 2011 [19 favorites]


The entire subgenre of "misheard lyrics" videos on Youtube rely on this phenomenon. As long as the subtitles match the sound of the lyrics at least somewhat, they can be very convincing.

It happens with movies as well. As soon as someone told me about the "Good teenagers, take of your clothes" line from Disney's Aladdin, I couldn't un-hear it. But until it was suggested to me, I had never interpreted the line that way.
posted by ShutterBun at 4:17 AM on July 11, 2011


So, in other words, we can't trust anything we see and hear? That's the basic premise of the demonstration. If the Big Bang Theory is incorrect, does that mean global warming, evolution, or any other controversial thoughts are potentially wrong, too? Not a particularly encouraging thesis.
posted by crunchland at 4:36 AM on July 11, 2011


This is an old parlor trick. I heard Howard Stern pull it on Billy Joel about 100 years ago. Stern accused him of racism for the lines (at about 1.33) on Movin' Out:

"If he can't drive with a broken back / At least he can polish the fenders"

Which Stern insisted were sung as "Nigga can't drive..."

Once that thought was in your head, it's impossible to hear it as anything but.
posted by three blind mice at 4:40 AM on July 11, 2011


Crunch, perhaps it's a reminder that doing good science (ie, letting the data talk to you rather than the other way round) is much harder than you might expect & it's very easy for your own biases to change what you see. Hence if your honest about wanting real answers to the questions you ask, you should aim to set up experiments that eliminate experimenter bias as much as possible.
posted by pharm at 4:41 AM on July 11, 2011 [1 favorite]


I vaguely remember at the trials of alleged mob members in New York in the John Gotti era, jurors didn't just listen to the surveillance tapes from wiretaps--they iistened along while reading a government-provided transcript of the conversations.

Later, the same tapes were used in another trial with different transcripts, because the government needed to demonstrate the guilt of a different alleged gangster.

I remember thinking at the time it wasn't fair to provide the jury with transcripts. The tapes, after all, should speak for themselves. The transcripts, to me, were hearsay.
posted by layceepee at 4:41 AM on July 11, 2011 [6 favorites]


Obligatory: Bill Hicks on playing records backwards to find Satanic messages (relevant bit at around 2:30).

Crunchland: Agreed. Singh doesn't leave scientists much of an out. Double-blind studies of astronomical data?
posted by HeroZero at 4:42 AM on July 11, 2011


I couldn't even follow the lyrics in the forward version.

In the backwards and with the lyrics on screen, I wouldn't say I heard those words. I'd say my brain agreed that those sounds could be the words my eyes were reading. They were consistent with, but not "the same as".
posted by DU at 4:48 AM on July 11, 2011


More songs to watch out for.
posted by pinothefrog at 4:53 AM on July 11, 2011


The effect is near-homophony, also known as mondegreen, and reading unambiguous text while listening to ambiguous sounds that could be spoken versions of the text, will cause you to hear those definite sounds. Compare with Buffalax's Benny Lava, and also with Prisencolinensinainciusol, without and with subtitles.
posted by aeschenkarnos at 4:53 AM on July 11, 2011 [3 favorites]


This is a good demonstration that what we perceive is not a moment-by-moment reconstruction of what our senses observe, but rather a pre-existing internal state that is updated via a combination of some sensory input, the current scene, and how we expect that current scene to change. But it's also important to remember that this is not a bad thing. We (and pretty much all other animals) are well tuned to do this precisely because it allows us to take a great deal of information out of a complex, noisy environment without undergoing a tremendous amounts of effort.

Cruchland and HeroZero: Trying to remove this sort of expectation bias is a large part of experimental design and statistical methods in big experiments. Precision measurement experiments (like measuring the mass of a new particle) actually do black box their data. For example it can come in the form of injecting some unknown, but fixed constant shift into the data, for instance shifting all measurements by a random fixed value known only to a small subset of people who aren't doing the analysis themselves. Then after the analysis is done, the shift is unboxed and the actual measurement appears, without a bias as to what the real value "should" have been. You can do something similar by doing a statistical analysis on a small subset of data, and then when satisfied that you understand what you're doing (i.e. are happy that you're taking into account systematic uncertainties correctly and modeling what you want to be), applying _exactly_ the same method to the other, larger data set and treating only those second values as measurements. This is also why exploratory data analysis is hypothesis generating, not explanation/results-generating.
posted by Schismatic at 4:56 AM on July 11, 2011 [5 favorites]


A good presentation. But he's using it to illustrate scientists not seeing a pattern that is there (the big bang) - surely a different phenomenon?
posted by iotic at 4:58 AM on July 11, 2011


So, in other words, we can't trust anything we see and hear? That's the basic premise of the demonstration. If the Big Bang Theory is incorrect, does that mean global warming, evolution, or any other controversial thoughts are potentially wrong, too? Not a particularly encouraging thesis.

We can't necessarily trust things we see and hear, but we can trust actual scientific results. Global warming isn't something you see or hear, it's something you measure with instruments.

If you measure something just by looking at it and guessing how long you think it is, you'll probably guess right. If you use a ruler, you'll probably be correct, provided your perception isn't so off that you misread the ruler. Like if you see an 8 and remember a 9. Or you're really tired and dream you measured it when in reality you didn't. Stuff like that.

But when we really know we know is when it's independently verified. You're not going to have two people have the same dream where they measure something, and the length comes out the same.
posted by delmoi at 4:59 AM on July 11, 2011


"If the Big Bang Theory is incorrect" [...]

The really interesting demonstration of unconscious misdirection in this video is how Singh seems to be arguing that the Big Bang theory is incorrect, i.e., that it's a false positive. However, he's really arguing that *not* believing in the Big Bang is false - that not believing it is a false negative.

The implantation of a make-believe pattern in the lyrics sets you up with the expectation that he's going to be making a point about the similar non-existence of a pattern in nature, but he goes the other direction.

The whole video is like a self-referential meta-example of his thesis - interesting.
posted by facetious at 4:59 AM on July 11, 2011 [7 favorites]


HeroZero - Agreed. Singh doesn't leave scientists much of an out. Double-blind studies of astronomical data?

Most bench scientists I've spoken to don't think that they need blinding for their work because the results tend to be so unambiguous: the machine spits out numbers that are higher or lower, the test tube goes BOOM! or it doesn't. However, there's a definite pattern of experiments which produce the "wrong" result being critiqued more harshly and repeated more often than those that produced the "right" one before the result is finally accepted. One of the worst that I've seen was a researcher who failed to demonstrate the expected change in a reading *five times*, then saw it on the sixth; the first five were declared to have not worked and the sixth taken as confirmation of their hypothesis. It's a tricky assay that you expect to fail occasionally, but the only reason they believed the one positive result over the five negatives is that it's the one that they were expecting/hoping to see.

I think that the principle holds fairly well. Unless interpreting the outcome of the experiment is largely subjective (or subjective opinions can influence the outcome, as in clinical trials), blinding within a given experiment isn't usually such a big deal. Much more important is trying to design a set of experiments and deciding before you start exactly what criteria will lead to results being believed or rejected, to avoid the bias introduced by being more likely to dismiss results that you don't like.

It is sometimes done better though, especially where the outcome of the experiment is more obviously a judgement call. In a couple of studies that involved looking at changes to the structure of tumour tissues (extremely hard to spot for a non-expert, and even experts will disagree sometimes), they took blinding pretty seriously and gave all of the tumour images random filenames before passing them to the expert.
posted by metaBugs at 5:11 AM on July 11, 2011 [2 favorites]


I'm pretty sure this is why I can still understand someone talking with a thick Scottish accent, while my native German speaking (who is fluent in English) girlfriend cannot. My brain kind knows what to expect while hers does not.

I don't think showing an example of how our brain inserts things is the same as our reluctance to accept new ideas, though.
posted by kisch mokusch at 5:14 AM on July 11, 2011


Next you'll be telling me that in Numa, Numa, they're not saying "you be the man to mash the feta cheese."
posted by MtDewd at 5:29 AM on July 11, 2011


I hate when someone calls the brain a computer without, at least, saying it's just a metaphor.
posted by Obscure Reference at 5:32 AM on July 11, 2011


When it comes to agency detection, a false negative might be more disastrous than a false positive (e.g. was that noise caused by a predator, or was it just the wind?). This might be why our pattern recognition algorithms are prone to apophenia.
posted by Human Flesh at 5:33 AM on July 11, 2011 [2 favorites]


Saddo : pla, Why would you be reversing Stairway to Heaven if you weren't already primed to hear satanical messages?

Fair point - Most of the tracks I screwed around with, I did so because I had heard something backward-sounding in it (ie, Pink Floyd's Empty Spaces). Stairway, though, I chose because I had heard vague rumors that it had some sort of satanic message when reversed. Still, not quite the same as reading along with the lyrics.

And I still don't hear the part about the toolshed. :)
posted by pla at 5:49 AM on July 11, 2011


This seems like how most *-ism (both "good" and "bad") are justified. There is a lot of that same pattern going on. "You didn't hear anything wrong with that statement, right? But how about if I tell you that A, B and C words really mean Z, Y and X, then what do you hear?"
posted by gjc at 6:07 AM on July 11, 2011 [1 favorite]


I heard it all the first time, but then, I heard Brother Billy Mayo's satanic backwards masking sermons back in the 80s. He didn't mention the toolshed, so I didn't hear that bit, and also in Brother Billy's version "He is my prince Satan, the one who lit up the night Lord you make me shout in glory to Satan" so that's what I heard the first time round.

Brother Billy also called the brain a computer, which is how, he explained, it can "hear" the backwards messages even when played forward.

I have certain biases towards Simon Singh, incidentally. I am prejudiced towards him for his fantastic book, 'Fermat's Enigma' but that is at least countered, if not overpowered, by his hair.
posted by Legomancer at 6:33 AM on July 11, 2011 [1 favorite]


In 1982, pastor Gary Greenwald hilariously claimed Queen's Another One Bites the Dust backmasked revealed "It's fun to smoke marijuana." (DJ Lobsterdust remixes).

Of course, this mean's Weird Al's "Another One Rides the Bus" backmasks to "Someone's on marijuana".
posted by Apropos of Something at 6:38 AM on July 11, 2011


"Go to church. Say your prayers. Tithe! Tithe!"

----Deathtöngue
posted by Runes at 6:52 AM on July 11, 2011 [11 favorites]


Reminds me of this excellent show:
NPR: Sometimes words behave so strangely.
Once you've heard it, you'll remember it (Next time you hear "Words behave so strangely")
posted by Dub at 6:56 AM on July 11, 2011 [2 favorites]


I have an idea for an experiment (which someone has probably done) to test the amount of data transmitted by speech. We clearly have two thresholds for comprehension - a lower one where we match what we hear to speech we've been primed to expect (like in the reversed lyrics above) and a higher one where we can match what we hear to something novel and unprimed. Knowing this, it should be possible to take some clear spoken text and artificially degrade it until it's (a) not comprehensible on its own, without priming; and (b) not recognisable even when the listener is primed. The difference between the two is the amount needed to actually convey meaning.
posted by Joe in Australia at 6:58 AM on July 11, 2011


Backmasking is really neat. Jeff Milner's Backmasking Collection has some great examples. (FD: Milner is a good friend of mine, but also quite internet-famous for this stuff)
posted by arcticwoman at 7:08 AM on July 11, 2011


I like this phenomenon a lot more when it's applied to Bollywood.
posted by JHarris at 7:20 AM on July 11, 2011


. . . .'scuse me while I kiss this guy satan.
posted by Herodios at 7:23 AM on July 11, 2011 [3 favorites]


Singh's book "Big Bang: The Origin of the Universe" is a great read. I't does an excellent job of teaching a complex topic to the layman. He is a brilliant guy.
posted by MotorNeuron at 7:23 AM on July 11, 2011 [1 favorite]


Radiohead - Like Spinning Plates

This one always takes a couple minutes to wrap one's head around. Mr. Yorke sang the song backwards, and the "backmask" is the forward-playing song on the album. Eminently intelligible!
posted by obscurator at 7:37 AM on July 11, 2011 [1 favorite]



My favourite satanic backmasking story indicted the Mister Ed show theme music.

In 1986 a Rev. Jim Brown of Ironton, Ohio claimed to find satanic backward messages in the theme music for the then twenty-years-cancelled Mister Ed, such as "sing a song of satan" and "the source is satan".

I say:
Go straight to the source and ask the horse
He'll give you the answer that you'll endorse.
posted by Herodios at 7:43 AM on July 11, 2011


It must have been an exciting time for certain people, when they were busy discovering all these messages. Imagine uncovering such a vast conspiracy like that, bit by bit, phrase by phrase, brain ticking along like mad working out all the little tricks they used to disguise them.
posted by Wolfdog at 7:52 AM on July 11, 2011


This reminds me of a great talk I saw by Ian Shipsey, a (formerly) deaf physicist from Purdue who had cochlea implant surgery in 2002 to restore hearing lost during cancer treatment years earlier.

His website links to a set of listening demos. Though I'm not sure if they are the same as the ones he used in his talk, they provide a good demonstration of how important context is to interpreting sound. The second link is to a page with a number of examples of how speech can be garbled when all you have is a few implants in place of countless hair cells in your ear. As you go through the table of examples, first listen to the processed speech, then the original speech, then the processed speech again. The 4-channel simulation works best for me. By the numbers, the amount of information loss that the human brain can accurately replace using contextual information is just astounding.

Professor Shipsey also played some music that had been processed in this way--I think it was a Sousa tune. At first, it sounded completely garbled, as organized as coins rattling around in a tin can. But as he started to hum (and bounce in rhythm), the melody came through, and it seemed that barely anything had been done to the recording.

The speaker in this video has taken an example of the human brain's incredible audio processing capabilities and claimed that it supports his theory, which is about how scientists tend to see support for their theories in data that doesn't in fact support them. I think this is a good example of what I like to call comic anti-irony, which is when something is amusing because what occurs is exactly what you should have expected.
posted by dsword at 7:56 AM on July 11, 2011 [2 favorites]


There was a great example of this at the last Dragon*Con. I forget who gave the talk, but it was a series of such things given to a packed room. Very awesome.
posted by odinsdream at 7:58 AM on July 11, 2011


This guy's got huge balls. He's giving away one of their best their tricks. Did anyone else watch this while wearing the sunglasses? They're half the audience, and they don't look too happy.
posted by Hoopo at 8:47 AM on July 11, 2011 [2 favorites]


Eminently intelligible! --- I think that's a bit of an exaggeration. But really, how different is this from Abba or Aqua singing songs in english though they have no idea how to speak the language? It's all phonetic.
posted by crunchland at 9:07 AM on July 11, 2011


If you hear something that "isn't there", doesn't that make it "there"? Regardless of whether it was intended to be?
posted by Liquidwolf at 9:16 AM on July 11, 2011


I am just supposed to pretend that I've never hear of this before? I didn't watch the whole video. Does he go on to eat a whole carton of Pop Rocks and drink a soda? Probably wraps the presetation up with "I'm a wiiiiiiiiild and craaaaaaazzy guy!!"

I mean he's even stolen Angelo Moore's haircut.


/meh
posted by humboldt32 at 9:34 AM on July 11, 2011


How did he find a room full of people that had never heard Stairway to Heaven backwards, or is that just something I think everyone’s heard 100 times because it was part of my growing up?
posted by bongo_x at 9:47 AM on July 11, 2011


I was rather hoping he'd play the backwards clip a third time, with different text showing.
posted by galadriel at 10:01 AM on July 11, 2011 [1 favorite]


I've certainly heard of it, but never tried it, and always assumed that people were hearing stuff after 100 listens and that no one else could hear the same thing.
posted by smackfu at 10:03 AM on July 11, 2011


How did he find a room full of people that had never heard Stairway to Heaven backwards, or is that just something I think everyone’s heard 100 times because it was part of my growing up?

I hadn't, but I'd heard of messages when playing records backwards before (although it was Ozzy Osbourne when I heard about it). I think the remarkable thing about this presentation though was that even if you'd heard "Satan" or "here's to my sweet Satan" before, with karaoke-style text playing at the same time he was able to construct an entire verse that's fairly convincing where nothing really exists. It's a pretty convincing demonstration of the power of suggestion, although I'm not sure that was his whole point.
posted by Hoopo at 10:04 AM on July 11, 2011


Pony request: could we add "toolshed" to the tag list?
posted by treepour at 10:29 AM on July 11, 2011 [1 favorite]


Must. Sacrifice. Goats.

To the toolshed!
posted by qxntpqbbbqxl at 11:40 AM on July 11, 2011 [1 favorite]


And that's totally Jesus on my toast.
posted by Decani at 11:44 AM on July 11, 2011


Also, I think The Beatles did this best. (NSFW)
posted by Decani at 11:47 AM on July 11, 2011 [2 favorites]


Allow me to introduce you to Benny Lava - a Bollywoodesque Tamil song-and-dance subtitled with "english lyrics".
posted by vidur at 12:30 PM on July 11, 2011 [2 favorites]


And that's totally Jesus on my toast.

Sorry, that was just me screwing with you.
posted by FatherDagon at 12:32 PM on July 11, 2011


Sorry aeschenkarnos, I see you already linked this
posted by vidur at 12:32 PM on July 11, 2011


But you really have to start worrying when your toast looks like this.
posted by FatherDagon at 12:33 PM on July 11, 2011


Those guys on Ghost Adventures use this (rather cheap) trick quite often, with unintended hilarity. (Gotta say I'm a little jealous that I didn't think of their tough-guy-to-ghosts routine first...oooh! They're so edgy!)
posted by malocchio at 12:47 PM on July 11, 2011


Hey, whaddaya know - Robert Plant's singing is just as intelligible backward as it is played normally!
posted by Greg_Ace at 8:06 PM on July 11, 2011


I could have sworn the presentation this was cribbed from was posted to metafilter. It was done much better the first time.
posted by Blasdelb at 4:48 AM on July 12, 2011


« Older Spanish police are investigating the disappearance...  |  In 1989, Eric Holmberg and The... Newer »


This thread has been archived and is closed to new comments