Follow Slashdot stories on Twitter


Forgot your password?
Games Science

Predicting Color Blindness, ADD, or Learning Disorders From Game Data 65

An anonymous reader tips a story at VentureBeat about a company that helps game developers analyze data gathered from their games to detect cheaters. But now, the company says this data can also be used to determine other traits of the players, like whether they're minors, or whether they like to gamble. Their CEO, Lukasz Twardowski, expects such analysis will soon be able to reveal even more traits, like whether a player is color blind, has a developmental disorder, or has Alzheimer's disease. "'Games are the richest and the most meaningful form of human-computer interaction. ...By tracking how they play games, we can learn a lot about people,' Twardowski explained. Hesitatingly, he added: 'That will be a huge responsibility for us later on.' ... Academics have begun to take games more seriously, as a window into the human psyche. Games are addictive and immersive and are built to command hours of our time and attention. What better testbed for myriad psychological and medical conditions? A good game pushes us to our limits, challenging us to use both the analytical and intuitive sides of our brain.
This discussion has been archived. No new comments can be posted.

Predicting Color Blindness, ADD, or Learning Disorders From Game Data

Comments Filter:
  • by Black Parrot ( 19622 ) on Friday July 27, 2012 @11:23PM (#40798811)

    I'm red-green colorblind, and it has always been a *huge* problem with games.

    Also with graphs and slides. When people use color coding in graphs I often can't tell which line or bar is which. Likewise when people plot data on maps by coloring the regions - annoys the heck out of me when I'm reading something interesting and can't process the graphical data display. For slides, same problem as graphs but worse - sometimes I literally don't see one or more of the lines on a graph projected as a slide.

    As for camoflage, folklore has it that during WWII a bomber was flying over the Pacific and one of the crew, colorblind, spotted an enemy ship that none of the others could see even when he pointed it out. But he convinced them to fly down for a closer look, and then they saw it.

    Possibly colorblind people are more attuned to value (lightness/darkness), since that's information we can reliably process. Very often I can only spot food stains on shirts if they are darker or lighter than the material.

  • by houstonbofh ( 602064 ) on Saturday July 28, 2012 @12:32AM (#40799033)
    Color is differentiated by hue and intensity. "Normal" vision is more hue than intensity. The common forms of color blindness is more intensity than hue. So camo is less effective. You can see if a portion of a car is repainted when others can not. Also, since people know this, traffic lights are not just red and green. The red is a darker intensity and the green is a lighter intensity. Even in black and white you can tell the difference.

    I am color blind. Friends often take me to look at used cars. :)
  • by Sir_Sri ( 199544 ) on Saturday July 28, 2012 @11:59AM (#40801759)

    with any certainty

    Well there's the question. If you can get 75 or 80 % accuracy in an academic setting (Which is not too hard to achieve) that may not be all that useful in a commercial environment where having a 20% error rate might completely wreck the experience.

    Unfortunately all of this stuff falls under 'human testing' where I am. We wanted to do a (really short) experiment to see how different coloured icons effected a players ability to remember what an ability was. Once we realized it was something like 70 pages of paperwork and a lot of money for oversight, and this was a class project, not a research project, we basically threw our hands in the air and said fuck it, not worth it. When you are dealing with people with learning disabilities you have to worry (a lot) about anonymizing the data, making sure there's no way this could get out etc. You have to prove your experiment couldn't cause further harm to someone who has an LD etc. Honestly the rules are a bit overzealous for this kinda stuff, but that's what we have to do.

    Believe it or not, the symptoms of ADHD can be quantified pretty well. You need a large data set, and you do end up with a situation where sleep deprivation mimics ADHD symptoms, so you could easily not be able to distinguish between the two. But you can track the symptoms of problems well. For colour blindness it's really really straightforward to test. Different specific problems will be more or less easy to try and detect of course.

10.0 times 0.1 is hardly ever 1.0.