Monday, September 14, 2009

As it Turns Out

Cognitive Science is pretty neat.  Another neat thing that turns out to be true: fMRI is the Mother Of All Brain Imagining Techniques.  So far, the only reasons to do anything /but/ an fMRI is that you can't, won't, or just really like scans with names like "CAT" and "PET".  As it turns out, MRIs (functional and otherwise) use GIANT MAGNETS (awesome!!!) to align those little dipoles we like to call hydrogen atoms.  And then the gizmo measures the time it takes your dipoles to discombobulate.  In turn, this tells us something about someone's brain.  What's particularly audacious about these things is how dramatic the post-processing needs to be in order to get something useful.  First you have to compensate, computationally of course, for movement along all three spatial axes.  Then you gotta antialias your voxels.  Then you're looking at collating your slices into a 3D image.  And, of course, you have to choose your imaging type, probably T2.  Lastly you'll want to take relative averages against your baseline for each time-slice.  And Voila!  You have a vague idea that something may have happened...and you know vaguely where it happened.  Now we're cooking.

In another not-so-recent not-so-scientific breakthrough, it turns out that single-speed bikes--you know, the ones people like because they're cheap and simple--are both expensive to buy and complicated enough to stump the most Irish of bike-store people.  So it goes.

But!  I have a number of nifty ideas for a cognitive psychology question.  As it turns out, coming up with smart questions really /is/ almost as hard as answering them.  We had to come up with a question concerning human cognition for this week's class.  My ideas are here in list form:

1) How do people separate words in spoken speech?
2) How do people simplify language for assumed learners (baby talk, pet talk)?
3) How do, if at all, people find the meaning of made-up adjectives?
4) How well can people make up a verb for a given noun?
5) How accurate are people at determining projectile trajectory in 2D?
    5a) How fast can someone /decide/ accuracy?  And how is this speed related to accuracy?
    5b) How might visual distractions affect accuracy?
    5c) Does technical knowledge of spatial geometry have any effect on accuracy or speed?

What interests me most about number 5 are the potential evolutionary aspects.  For instance, there's an obvious advantage to determining the accuracy of parabolic curves (so as to throw something accurately...or dodge that dodge-ball).  There may also be advantages to accuracy in linear prediction.  But might we be significantly less accurate for logarithmic or exponential curves?

As it turns out, you'll have to check back in seven years to get answers to any of these questions.  And, as it will likely turn out, the answers won't make anyone better at basketball.

3 comments:

  1. I've been meaning to respond to this since the first day you've posted it, but since it's well thought out, it probably needs a well thought out response.

    All of your questions seem to boil down to unsupervised learning. Which makes sense, because that's what the brain DOES. Everything from learning a language, to figuring out directions, to deciding what's your favorite flavor of pop, there's no programmer out there popping in your head telling you where correct and incorrect answers lie. Although, there might be a genetic component that nudges us in the right direction. Read, Chomsky.

    >1) How do people separate words in spoken speech?

    Through the subconscious analysis of the frequency of sounds, syllables, stress patterns and pauses.

    Here we're dealing with very minute differences born not out of something you consciously construct, but out of mimicing of the people you learned it from. There is a HUGE amount of language that isn't in the words at all!

    For example, try reading this sentence out loud:

    I bought some cauliflower at the grocery store.
    I've been meaning to respond to this since the first day you've posted it, but since it's well thought out, it probably needs a well thought out response.

    All of your questions seem to boil down to unsupervised learning. Which makes sense, because that's what the brain DOES. Everything from learning a language, to figuring out directions, to deciding what's your favorite flavor of pop, there's no programmer out there popping in your head telling you where correct and incorrect answers lie. Although, there might be a genetic component that nudges us in the right direction. Read, Chomsky.

    ReplyDelete
  2. >1) How do people separate words in spoken speech?

    Through the subconscious analysis of the frequency of sounds, syllables, stress patterns and pauses.

    Here we're dealing with very minute differences born not out of something you consciously construct, but you can still figure out where those boundaries are out of previous exposure to the language and speaker’s subconscious pauses. There is a HUGE amount of language that isn't in the words at all!

    For example, try reading this sentence out loud:

    I bought some cauliflower at the grocery store.

    There is a very, very good chances your stresses were on the syllables right before the ' character:

    I' bought' some' cau'liflo'wer at' the' gro'cery store'.

    Now let's destroy the word boundaries of this sentence and recreate them. Try reading this sentence, slowly as if you were reading it to a kid:

    Iboughtsome cauliflo weratthe gro cerystore.

    Were your stresses somewhere along the lines of

    Ibought’some cau’liflo wer’at the gro’ ce’rystore.

    ? There’s also a good chance that they were.

    You are following some set of stress rules, even if you don’t know where you learned them, or what those stress rules are. You can also see that a lot of the stresses stayed the same, but some were changed. If you moved some of the stresses around, you can feel how it changes some of the context of the sentence. Saying “cery’store” feels like you empathically punching the air. Very odd.

    ReplyDelete
  3. >How do people simplify language for assumed learners (baby talk, pet talk)?

    This dissertation was interesting: http://repository.upenn.edu/dissertations/AAI9627938/ What happened was a number of Japanese families’ dinner conversations were recorded both on their own and with their home stay student and the results were compared. What was the most interesting in my mind was how families who spoke in different dialects made the same generalizations to their students.

    >3) How do, if at all, people find the meaning of made-up adjectives?

    Through context of previous/following sentences and expressions. Take for example, “Jenny got a very smorfblradic haircut! She always has a sense of style!” and “The way Bob just barges in here is a very plaspignit behavior.” This is why grade school teachers tell their kids to write sentences for their vocab words. It reinforces the context in their heads. Same for nouns.

    You may want to check out this website: http://linguistica.uchicago.edu/ Very interesting, even if they take some shortcuts with things like requiring words to be separated by a space. Very accurate with determining a grammar too.

    ReplyDelete

Keep it real