Recent ACT-R journal submissions

[ERIK M. ALTMANN]

Available at http://hfac.gmu.edu/people/altmann/pubs.html; comments
welcome. 

Erik.

-----------------------
Erik M. Altmann
Psychology 2E5
George Mason University
Fairfax, VA  22030
703-993-1326
altmann at gmu.edu
hfac.gmu.edu/~altmann
-----------------------


Altmann, E. M. & Gray, W. D. (1999).  Preparing to forget: Memory and
functional decay in serial attention.  Manuscript submitted for
publication.

Abstract - Research on serial attention has investigated the switching
of attention from one task to another but has largely overlooked the
maintenance of attention on one task between switches.  A decay
hypothesis is advanced that describes attention maintenance and
relates it functionally to attention switching.  Attention maintenance
involves retrieving the current task from memory regularly between
task switches.  However, despite the strengthening that accompanies
retrieval, the current task must decay to prevent interference with
the next task.  The current task will decay if its initial activation
is high enough that subsequent retrievals cannot sustain that level of
activation.  The preparatory encoding responsible for this high
initial activation is the proposed explanation for switch cost.  Four
studies uphold the decay hypothesis, showing a pervasive slowing trend
of 3 to 5 msec per trial (between switches) as the current task
decays.


Altmann, E. M. & Trafton, J. G. (1999).  Memory for goals in
means-ends behavior.  Manuscript submitted for publication.

Abstract - Means-ends behavior arises when goals (the ends) need to be
suspended to achieve subgoals (the means) and then resumed again.  One
traditional explanation of such behavior is that cognition stores
goals on a stack (the task-goal stack, or TGS).  However, the TGS
lacks face validity, and the authors propose the memory-as-goal-store
(or MAGS) approach instead.  A MAGS model explains means-ends
performance better than a published TGS model implemented in the same
cognitive architecture (ACT-R).  Experimental data are presented that
contradict even weak predictions of the TGS and support the MAGS view.
For example, pending goals intrude more often than chance as a
function of active maintenance.  The computational processes of the
MAGS model are specific enough to make fine-grained predictions and
general enough to explain memory for goals in diverse contexts.