Derek M Jones wrote on 17/4/01 1:58 am:
> Is there a cognitive psychology researchers yellow pages? Some
> advice on where to send anybody who calls me up would be
> appreciated.
There have been a few attempts to establish such a thing. But most people
find that they have enough to do without offering themselves as a source of
free advice. Pity.
One of the big problems, about to become my bee-in-the-bonnet for a bit, is
the difficulty non-specialists have in generalising usefully from lab
studies to their own problems. Often they assume that if nobody has looked
exactly at weasel-sprogging in Fortran II (or whatever they're into) then
there's no relevant research, when in fact there might be useful results on
limited aspects of mammal-sprogging in a different language that might quite
plausibly apply to other cases.
This problem seems to me to arise partly because psychs don't get round to
making their work very accessible in a user-friendly manner, and partly
because non-psychs aren't too good at seeing the links (or even looking for
them, some say).
> What research topic should I recommend?
> Build a model of program understanding using ACT-R, SOAR or some
> other model?
>
> Research on program complexity measures (but then a lot of people
> have already looked at this without much to show for it).
Forget program complexity. Program understanding would be a better target.
Where 'understanding' might mean extracting 'programming plans' - though
ideally it should mean building a proper situation model, and so on.
SOAR has a high entry cost. I did some work on a very similar idea using
SOAR and decided it wasn't going to be cost-effective for practical
purposes. (However, Richard Young thinks otherwise.)
ACT-R might be easier. In cardiff, Andrew Howes and Stephen Payne have done
work on other, but related, aspects of user behaviour using ACT-R. But I
understand their typical approach is to knock up a quick model in Prolog,
decide what they should have really done, then recode in ACT-R. (Rapid
prototyping in the cognition lab?) Mail Andrew or Stephen for more detail:
"Andrew Howes" <[EMAIL PROTECTED]>
"Stephen Payne" <[EMAIL PROTECTED]>
I think you could go a long way towards establishing some useful results
without invoking any specific cognitive architecture in any detail. A long
while ago I did a bit on understanding Prolog versus Pascal. We used a model
of human parsing but we didn't relate it to ACT SOAR or anything else, we
just programmed the model. (BTW Prolog is harder to parse, according to that
model.) More recently Alan Blackwell simulated a model of decisions about
investment of attention, which I think is highly relevant to many questions
(e.g. when to delve into understanding a function and when not) and again
didn't relate it to any specific architecture. I don't see what would be
wrong with you doing something similar.
> These people will want
> some payback in terms of results showing how to reduce faults,
> improve productivity, etc ...
They won't get cost-effectiveness predictions but you could get, for
example, a dramatic curve showing how the cost of debugging code increases
as the code gets more dislocated and fragmented (as in a typical scripting
language, such as Hypercard). What would be the real contribution would be
demonstrating the interplay of various factors. You'd get rhetorical
evidence, not real dollars-and-cents evidence; the Hard-Nosed Ones would
complain that you didn't have any empirical data and even if you did you
couldn't turn it into an estimate of how much they'd save by doing X instead
of Y, but nevertheless the Soft-Nosed Ones (of whom there are plenty,
luckily) would use your results to help inform their ideas. Any use to you?
Thomas
----
T. R. G. Green also at:
preferred postal address: Computer-Based Learning Unit
Oriel House, 27 Allerton Park, University of Leeds
Leeds LS7 4ND, U.K. Leeds LS2 9JT, U.K.
0113-226-6687 (tel)
0113-226-2751 (fax)
http://www.ndirect.co.uk/~thomas.green
- Automatic footer for [EMAIL PROTECTED] ----------------------------------
To unsubscribe from this list, mail [EMAIL PROTECTED] unsubscribe discuss
To join the announcements list, mail [EMAIL PROTECTED] subscribe announce
To receive a help file, mail [EMAIL PROTECTED] help
This list is archived at http://www.mail-archive.com/discuss%40ppig.org/
If you have any problems or questions, please mail [EMAIL PROTECTED]
