« Running with Coconuts | Main | ICAHK »

December 15, 2009



Personally I would just have assumed the blighter in the blue shirt should have been drinking from a sippy cup, and I would have avoided calling attention to it so as not to embarrass him (further).
People get paid for this?! Man, did I miss the boat...


Ok, no more sarc., I have a confession. I have a problem with facial recognition; I categorize people by their features. Unless I know someone very well, I'm just as likely to confuse them with someone else, especially if I see them in a different context then I would normally. The problem? I don't admit it, I just bluff my way through the encounter hoping that the person will say something which will confirm their identity.
I have a huge distrust of the eyewitness I.D. concept!
Remember 'Kim's game'?


No, you've got the wrong end of the stick with the change-blindness thing. None of it is obvious, which is why Prof Simons may appear rather smug about what he's discovered (Not just him, of course - there's about 10 discoverers of the phenomenon). It's a beautiful disproof of a certain naive-but-compelling theory of vision. That theory is that the visual system constructs a comprehensive "model" of what is out there - a set of brain activities that are in some sense equivalent to the objects in the world. This theory chimes with our own introspection, and seemed sensible enough that it drove high-level computer vision research for years. There are even some experiments that seem to support it (mental rotation, mental map following).

However, what the change-blindness people showed is that it isn't true. If you create a detailed mental model, you ought/need to be able to spot when it doesn't jibe with your surroundings and so update it. This doesn't happen - so the mental model isn't very complete at all. The current thinking is that we don't bother with the mental model, because the world serves as this model - why, for example, remember in detail what a vase looks like, when you can just look at it; or remember the colour of his jersey when you can just see it (which is why he can change the colour without you noticing - you never encoded the colour in memory).

You'd think that this would be a problem, except that we have an alerting mechanism that tells us when something changes. One thing the change-blindness people were able to do was to subvert this alerting mechanism, by doing film cuts (as in the movie) or flashing blobs over the parts of a still image that would change.

Anyway, I think it's a fascinating topic, and worthy of study; it isn't the usual psychology bullshit. These guys are serious scientists.

Mick H

That was a very lucid explanation - thanks.

Yes, I realise that they are disproving what you describe as a "naive-but-compelling theory of vision". I'm writing partly tongue-in-cheek. I can understand why they're pleased with themselves. It's just - well - this "naive-but-compelling theory of vision" was promulgated by other psychologists, and a lot of us outside the inner circle will be thinking - well, of course - what did you expect? People notice what they need to notice.

Plus there's a very fine line sometimes between candid camera and psychology experiments like this. It's hard not to think that these psychologists enjoy showing people how stupid they are.


William's comment reminded me of my Siberian Huskies of years past; they were consummate hunters. The theory is that dogs 'see' things in a series of stills, not unlike an old silent movie. They don't need colour vision so much as low light capability, the opposite of human evolvement. When they hunt, they are focusing on movement, ie "change"...cats would seem to act in the same manner, they move their heads incrementally as they focus their attention on a specific view.
Getting back to the Huskies, they'd hunt amongst the beach rocks, the same colour as the rodents, concentrating on the slightest movement (change). Their attention to the task (and patience)was incredible. If they had a quarry, they would sit motionless and silent for as long as it took for the target to break cover and make a run for it. Big mistake for the rat.
I think we're talking about an evolutionary path here, we're hardwired to perceive things the way we do. The question is, what's the advantage?



This sort of thing has a practical application in teaching people how they can miss things. In the case of complex systems (such as medicine) it is extremely important.

The comments to this entry are closed.