[ACT-R-users] Postdoc needed on predicting and preventing procedural errors

[Greg Trafton]

I am currently looking for a postdoc to work on predicting and
preventing errors in procedural tasks.  A brief description of the
position is below.

In the last few years, we have been working on errors that occur in
procedural tasks from both a theoretical and applied perspective. We
attempt to understand errors by experimentation and modeling of the
perceptual, goal-based, and memory processes that occur when people
make procedural errors, and by studying how those processes differ
from correct behavior.  We use our theoretical models to build
statistical models to predict when people will make errors.  Finally,
we use our predictive statistical models running in real-time on
procedural tasks to prevent people from making errors.  We have used
this approach for post-completion errors and have recently been
working on sequence errors.

I am looking for people that

* Have a Ph.D in cognitive psychology, computer science, or cognitive science
* Have cognitive modeling experience
* Have experience with eye-tracking methods 
* Have a very strong statistical background, including logistic
  regression and signal detection theory
* Are interested in both theory and application

We develop theories and have the potential to carry them through to
practical application.  The opportunity to participate in all phases
of this process is rarely found in the research world, and offers its
own set of challenges and satisfactions.

The position would be an NRL/NRC postdoc for 2-3 years; payment is
approximately 74,000 / year + a travel stipend + moving expenses.  NRL
is located in the Washington, DC area.

Only US citizens and permanent residents are eligible.

The deadline for application is February 1, 2011.  
Details about NRC postdocs are at http://hroffice.nrl.navy.mil/jobs/postdoc.htm
If interested, please contact me at greg.trafton at nrl.navy.mil

http://www.nrl.navy.mil/aic/iss/aas/errors.php gives some recent
papers on this work.

ACT-R modelers are especially encouraged to contact me ;-)

thanks,
greg