I am unable to find the link

April 19, 2003 9:47 AM

There once was a MeFi thread about a psychological experiment in which people were presented several pictures of women and men and they had to choose the most beautiful among them. The most beautiful chosen were computer generated pictures.
I am unable to find the link and have extensively used the search engine. Anyone care to help me out?
posted by zerofoks to MetaFilter-related at 9:47 AM (18 comments total)

I can't remember seeing this on MeFi but could it be
this or this. These lead to a number of experiments that use computer-generated blends of faces from the Perception lab at St Andrew's psychology department.
posted by jamespake at 11:27 AM on April 19, 2003

Doing this first test mentioned by jamespake, I'm intrigued by how my answers for the questions (before the picture queries which I went through and then #20 just repeated & I never got any results) asking for a description of myself are rather average (and occasionally insulting) but the choices I'm making for my "ideal partner" could also be the qualities I'd most like to see in myself, if I didn't instinctively (day in & out) act like such an asshole. There was a recent episode of Crime Scene Investigations where the victim of a murder was lured into a trap by someone who created a realistic photo-image of a (fictional) secret online admirer, whose face was actually made by computer and whose likeness was a feminized version of the victim. In other words, the killer lured the victim with a feminized image of himself - like attracts like - the ultimate in narcicism.

I don't recall the thread either, zerofoks, but it does sound familiar to me as well.
posted by ZachsMind at 12:28 PM on April 19, 2003

I don't remember that link, but it sounds very similar to some of the work being done by the psychology department of St Andrews University
posted by gravelshoes at 12:40 PM on April 19, 2003

That CSI, always on the cutting edge of hard science.
posted by Hildago at 3:51 PM on April 19, 2003

That CSI, always on the cutting edge of hard science.

Or hogwash.
posted by y2karl at 4:14 PM on April 19, 2003

I'm having the same perpetual loop problem with #20 so I can't get the results. That was a waste of 15 minutes.
posted by jaden at 1:16 AM on April 20, 2003

maybe it's one of those typical psychological tests that say they're testing for one thing, but really they're testing for something else, so really it's testing for correlations between stupidity/stubbornheadedness (how many times you try to get past #20) and facial shapes.
posted by kv at 6:06 AM on April 20, 2003

Good christ I wish I had read through these responses before clicking on that first BBC link.

Number 20...number 20...number 20...

Ack. Ten minutes I won't get back.
posted by adamgreenfield at 7:18 AM on April 20, 2003

Those pictures in teh BBC link are pretty nasty looking, too. I thought I was picking wrong as they just seemed to keep getting uglier.
posted by Space Coyote at 7:38 AM on April 20, 2003

I like CSI
posted by matteo at 10:45 AM on April 20, 2003

The pictures in the BBC test are nothing compared to those in St. Andrews Univerity's test, Space Coyote. I'd like to believe I was serious about doing the test, but how was I supposed to choose between two nearly identical pictures of this pretty lady?
posted by Hjorth at 11:13 AM on April 20, 2003

Yikes!
posted by Samsonov14 at 11:21 AM on April 20, 2003

It's funny because she's an ugly person!
posted by Mid at 12:45 PM on April 20, 2003

all the ugly people in the house, raise up sticks. you know, our ugly sticks.

*prepares to smack some of the pretty people it must be so pretty to be*
posted by folktrash at 4:24 PM on April 20, 2003

you mean all this time, when people look like they just got beat with the ugly stick, it was ugly people who were doin' the beatin', with sticks they carry with them?

Every day I learn something new from MetaFilter.
posted by soyjoy at 12:56 PM on April 21, 2003

Thanks, Wet Spot! :)
posted by zerofoks at 11:05 PM on April 22, 2003

Mitchel - you're first point could be right on the mark - as you can see from this explanation of face averaging, the technique starts with a sample of faces that has been rated on some quality by a group of people. If the faces you show are all white then your end result will be white. Similarly, if the raters are all of one racial group and they prefer people of their own race, that may be reflected in the scores. Incidentally, this site came up on Metatalk last month.
posted by jamespake at 5:08 PM on May 18, 2003

Doh! That last comment was meant to go here. I'm not trying to create a metaloop, honest.
posted by jamespake at 5:12 PM on May 18, 2003