Thursday, November 25, 2010

Reading about Tetris

Tetris ... who would have thought that this simple, yet highly addictive, game would have spawned so much interesting research?

It makes sense though, the game has a number of properties that make it an interesting and useful research tool. Most importantly, in my opinion, it's fun to play (this is one of Kirsh and Maglio's observations) so actually getting the subjects to sit down and play the game for the required amounts of time is nowhere near as challenging as if, say, you were getting calculus naive subjects to sit down and learn differentiation.
That's obviously not all there is to it though - other than the pragmatic value in it being fun, some of the really interesting properties are that
  1. It's simple - it has a very low barrier to entry (unlike the learning differential calculus example above)
  2. It (potentially) engages a number of cognitive processes (like perception, planning, mental rotation, learning etc.)
  3. The processes that are engaged are put under pressure because of the high speed of game play.
For the most part this general specification of the properties of Tetris more than suffice to justify it as a research tool.
I'm going to be blogging about any interesting papers / stories I find about Tetris in the next couple of months, but to illustrate how Tetris has been put to use, I'll give a quick gloss of two recent studies involving the game.

Tetris "inoculation" against Post-traumatic stress disorder

Of course one can not be literally "inoculated" against PTSD, but a recent study suggests a way of reducing the number of trauma related flashbacks using Tetris (or, presumably, something that has the same kinds of functional properties as Tetris).
The researchers had their subjects watch a distressing film in which people were injured and killed. They then randomly selected subjects to play Tetris for 10 minutes (after a shared 30 minute break). The subjects were then asked to record the number of times during the week in which they experienced flashbacks of the film. It turns out that the group of participants who engaged in the Tetris task experienced significantly fewer flashbacks.

The basic idea behind the study is nicely summarised by the authors - this is taken from the abstract

Our theory is based on two key findings: 1) Cognitive science suggests that the brain has selective resources with limited capacity; 2) The neurobiology of memory suggests a 6-hr window to disrupt memory consolidation. The rationale for a ‘cognitive vaccine’ approach is as follows: Trauma flashbacks are sensory-perceptual, visuospatial mental images. Visuospatial cognitive tasks selectively compete for resources required to generate mental images. Thus, a visuospatial computer game (e.g. “Tetris”) will interfere with flashbacks.
You can read the paper here

(thanks to Michael Meadon for drawing my attention to this study)

Tetris increases cortical thickness and changes brain functioning

Another recent study comes from Richard Haier, Karama, Leyba, and Jung who took a group of adolescent girls and had them play Tetris a few hours a week for three months. They also obtained structural and functional data on the girl's brains
before and after the Tetris training.

What they found was doubly interesting - firstly, they found that there was increased cortical thickness (structural change) as well as functional changes in the brains of the girls who had been playing Tetris compared to the control group who hadn't. Something along these lines were expected, however, there was another interesting result, and that was that the location of the thickening of the cortext and the functional changes were in different places in the brain.

The paper, in all it's gory detail, can be found online here

I'm going to hold off discussing this paper in any depth though - it links up with a number of other interesting things that I think deserve a little more attention. So give it a read, and we can get into it next time.

Wednesday, November 17, 2010

NaNoWriMo - look Ma, I wrote a book!

Right, so every year when November rolls around it's time for National Novel Writing Month (despite the name it's very much an international affair).

NaNoWriMo is a challenge to anyone who has ever wished they could write a novel (and most of us who love books have, at one stage or another, dreamt that they could write one of their own). The challenge is to do just that, write your damn book!

You are challenged to put aside the things that usually distract you, those things that usually absorb all of your free time, and concentrate on simply putting pen to paper (or fingers to keyboard) to knock out 1667 words per day for 30 days.

This is the first year in a long time that I've actually had a little free time (no exams, no assignments - bliss) and I had desperately wanted to do it in 2009 when I'd first heard about it - so I decided that I would give it a shot.

I don't want to give a blow by blow account, but it was surprisingly easy to write a coherent story of at least 50000 words (the, disputed, minimum length of a novel).
The trick for me was to switch off my inner censor and just write, regardless of whether I thought I was pushing out absolute tripe. Having the minimum word count (and a fancy graph that let's you know if you've fallen behind) was something that I found seriously motivating.

Although my novel is really only good enough for kindling, I learnt a lot along the way. First off, I learnt that writing something of novel length is not nearly as intimidating as it seems at first (although writing something of quality at novel length is a slightly different story). Secondly, I learnt a lot about pacing something that's longer than a few pages - I learnt that I can slow things down a little, take my time to develop ideas and scenes. Pushing through this first NaNovel was an invaluable lesson in writing, worth way more to me than final product.

I can't recommend NaNoWriMo enough, it was a fantastic experience. If you've ever wanted to write a novel, but have never managed to get past the first few pages, consider tackling NaNoWriMo next year.

Sunday, November 7, 2010

Intelligent machinery

Anyone with an interest in AI will know of Alan Turing - he, along with a handful of other mathematical geniuses with a penchant for the pragmatic, lay down the foundations of modern computer science.

Turing is known primarily for the contributions that bear his name - that is, Turing machines, and the Turing test (although just what the Turing test is supposed to be testing, and whether it can actually discern what it's supposed to be testing for, is an issue that has never quite been resolved to anyone's satisfaction).

These two ideas were laid out in two supremely famous papers - the Turing test is described in Turing's, much read, paper Computing machinery and intelligence (although Turing called it the "imitation game"), and his Turing machine is described in the much less widely read (there are far clearer and accessible treatments than his original paper - but I don't think that's why people don't read it) On Computable Numbers, with an Application to the Entscheidungsproblem

However, there is a third paper that deserves to be equally well known - that is his remarkable paper "Intelligent machinery" (not to be confused with the similarly named "Intelligent Machinery, a heretical theory" - I would guess that it is generally unknown because Turing never actually published it which is a real shame. In B. Jack Copeland's words the paper

is a wide-ranging and strikingly original survey of the prospects for Artificial Intelligence. In it Turing brilliantly introduced many of the concepts that were later to become central in the field (Copeland 2004 : 401)

These concepts included brief treatments of genetic algorithms, logic based programming, and an early description of the Turing test - however
"[t]he major part ... consists of an exquisite discussion of machine learning, in which Turing anticipated the modern approach to AI known as connectionism" (Copeland 2004 : 401)

If you're at all interested in the history of AI, Turing, or the ideas that have shaped our last century - you can (and should) read this document, available in Copeland's collection The essential Turing

References :
Copeland, B. J. (ed) 2004. The essential Turing. Oxford : Oxford University Press