[Previous entry: "On This Day in Christory..."] [Next entry: "Great Hitch-Hikes - the York Jazz Festival."]
05/04/2004: "Facial Deprivation"
I won't go into the gory details, but I am living alone again for the first time in many years, and I guess I'd forgotten what its like. Its a little like travelling for a while, when you start to see familiar faces everywhere. Somone suggested that that might be because of a limited number of different "people" types in the world. But although I used to play with that idea, of a world where there were only a few hundred principal actors, a couple thousand "extras" for the crowd scenes ( like news pictures and so on), and the rest are only rumours, crowd-dots on a far away blue-screen, I no longer really subscribe to that philosophy.
But there is a definate effect I call "facial deprivation", where your mind, used to a certain number of pattern-matching hits a day (in the "human recognition area - which yes, is different to your, say apple recognition area), starts to see a rapid change in match-frequency, and it assumes that there is something wrong with the mechanism. It tries changing the parameters to achieve the old coverage, (although it is the input sample that has changed), so all of a sudden you start getting a lot of "false positives".
(click below to read more about facial-deprivation)
So, from that, I assume the brain has many of these built in self-calibration routines, so when a pattern-frequency suddenly changes, it throws in some positive feedback to try and re-match - a sort of simulated annealing for pattern recognition. For example, say you are wounded in one eye, so vision is impaired. For a while, you start not recognising people, and so the brain says "try and adjust your parameters to see if our eyes are out of adjustment". Sure enough, you start to recognise people again, and so the parameters stay.
Interesting, I wonder if they ever change back again. So when you return to normal, after a parameter adjustment of a certain pattern, you may still reocgnise a person that you did before the adjustment, so there would be no reason for the parameters to change back. But we might assume that in most cases, this increases the chance of a false positive - that any broadening of parameters... oh crap! I just realised it does happen. The parameters don't just widen, they _change_. What a design defect! So the parameter changes, and unless we get an occasional test trigger (a positive or negative), then the parameters will creep until we no longer recognise the person - exactly what happens when we haven't seen someone for a long time...
But why does the facial deprivation effect only become noticeable during a period of sharp change? For example, when you travel, it usually hapens only for the first couple of days - at least, you only notice it then. I guess that it is basically an internal debug that is popping up unexpectedly here, but there is a check routine that finds debug messages that were left behind (and there must be a lot), and remove them - slowly, over the course of a few days, but it does it. Not permanently of course. So, just how much debug code is there in there? I already ran into a core dump message, a leaping tiger, and a "normal service will be resumed shortly" message, but I expect there are a lot of more minor debug messages going on. If only we could figure out how to decode this debug, we might learn a whole bunch about how the brain works...
Oh, so another corolary of the above (pronounced "core-ola-ree" for you Aussie's, but "Cora - larry" for you yanks) is that by artifically altering our pattern recognition for a short time, we can actually change the parameters in which we recognise things. So, for example, wearing blurry glasses for a while may help us recognise things more easily when we take those glasses off. Anyone wanna experiment?
Replies: 8 Comments
On Tuesday, May 4th, at 10:35 PST, the princess (yank) said:
what? Are you trying to say you miss me? 
, I'll reread that some other time and try to decipher it. or maybe you can draw me a picture or something. I got the part about wearing blurry glasses and I'm up for a round of experiments. Or BEER glasses....hmmm...
On Tuesday, May 4th, at 15:05 PST, Ch(K)ris said:
Personally, I would "see" the issue as one of learned habits (or Foucaultian discursive practices). We are "trained" to exist within a number of relationships - some intimate, some close, and some with people we regularly see (co-workers, commuters who get on the same train every morning). We define ourselves as consisting of these relationships.
No need to see the brain as a separate entity of analysis... (which is could for some who have concerns for the state of theirs)
Periods of "sharp change" are times when our self-identity is being challenged - to compensate we endeavour to find other relationships, no matter how distant, wherever we look...
On Tuesday, May 4th, at 15:50 PST, Chris Tann said:
Well, I also subscribe to the "stack" theory of intelligence. While it is not so clean as, say, an IP protocol stack, I think it has many similarities. So the question is, at what layer does facial recognition occur. While agree that the relationship _behind_ the recognition works at a higher trained level in the brain, I think the actual pattern recognition functions themselves exist as a lower level function. Pattern recognition exists in even the lowliest of creatures, whereas "relationships" don't start until you get way up the protocol stack.
On Tuesday, May 4th, at 17:58 PST, Mother said:
This discourse is way above my poor mental faculties. I would need it explained in a less erudite fashion.
On Tuesday, May 4th, at 19:20 PST, AJ said:
You know, I thought I was the only one that experienced that sort of thing while I was traveling.
On Friday, May 7th, at 03:16 PST, Samos said:
An interesting point I find with this facial recognition is when you have a baby. What is the first thing that a lot of people say?? Oh he looks like his mum/dad/gran/pop/milkman.........and you will get just about all of these responses.
So the facial recognition for the same person (ie the baby) is different for different people, and is attributable to the expectations of the viewer.
Also in this example people are looking for individual parts of the face that are "from" each parent, I don't think we look at other people we have only just met looking for similarities to someone else, even though there may invariably be similarities.
well thats my 2c worth
Samos
On Friday, May 7th, at 10:43 PST, Chris Tann said:
As I mentioned above, I believe that "facial recognition" is much more highly developed than, say, apple recognition. Hmmm, well now that I think about it, peer recognition is. How many times have you heard "Those all look the same".
Anyway, being massively parallel, the brain does exactly what you said - pattern matches the nose, the eyes, the chin, etc. in parallel, and then pattern matches that reduced set of features against faces it knows. OK, thats a gross over-simplification, but again it goes back to my stack theory - there is a high level "peer recognition function", built on top of a lower "feature recognition function", which in turn is built on a primitive "pattern recognition function".
On Monday, January 31st, at 15:29 PST, Chris Tann said:
I watched a program on DIscovery the other day (Narrated by Alan Alda - apparently he is right into this whole "science of the brain" thing), and it had some interesting empirical results - although I think that there main point was fundamentally flawed (or perhaps just fundamentally over-simplified).
They did some "emotional memory" experiments, showing test subjects a bunch of pictures that provided various emotional reactions, and then at some later stage, did a memory test to see how well the images were remembered. For some of the subjects, they subjected them to a painful experience immediately after watching the images - in this case, thrusting an arm into ice water and leaving it there for several minutes. Although the "painful event" happened after the "memorising events", there was a strong correlation suggesting that those subjects remembered the images better than the "placebo" subjects.
I know that when I was studying, I would often get up and bang my head against the wall - I just didn't realise that there was a physiological reason why it made me remember things better!
Oh, and when I say the program was fundamentally flawed, it was just that they made no differentiation between short-term and long-term memory, and so I don't know if the "extra-memorisation" effect was merely a short-term one, or a more permanent one.
