From pirolli at parc.xerox.com  Wed Nov  8 17:04:43 2000
From: pirolli at parc.xerox.com (Peter Pirolli)
Date: Wed, 8 Nov 2000 14:04:43 PST
Subject: Foraging & ACT-R
Message-ID: <mailman.46.1617128263.6729.act-r-users@act-r.psy.cmu.edu>

and ACT-R:
 
http://www.newscientist.com/features/features_226445.html
 


From Wolfgang.Schoppek at uni-bayreuth.de  Fri Nov 10 07:59:36 2000
From: Wolfgang.Schoppek at uni-bayreuth.de (Wolfgang Schoppek)
Date: Fri, 10 Nov 2000 13:59:36 +0100
Subject: Dynamic systems & ACT-R
Message-ID: <mailman.57.1617128263.6729.act-r-users@act-r.psy.cmu.edu>

the next issues of "Cognitive Science Quarterly". It contains a running
ACT-R model.
 
http://www.uni-bayreuth.de/departments/psychologie/ExRulStratScho.pdf
 
Comments are welcome,
  Wolfgang Schoppek
 
Title
Examples, Rules, and Strategies in the Control of Dynamic Systems
Abstract
Two main types of knowledge are considered relevant to successful
control of dynamic systems: input-output knowledge (I-O-knowledge),
which represents specific input values together with the corres-pon-ding
output values, and structural knowledge, defined as general know-ledge
about the variables of a system and their causal relations. While
I-O-knowledge has proven important for the control of small systems,
structural knowledge is expected to enhance performance when dealing
with more complex systems. In an experiment, structural knowledge about
a complex system was manipulated. Although the experimental group had
better structural knowledge, the control group was equally successful in
reaching new goals. That seems to contradict other studies where effects
of structural knowledge on performance have been found. To resolve these
contra-dic-tions, the consideration of a third type of knowledge -
strategic knowledge - is suggested. The postulated effects of different
levels of structural and strategic knowledge are explored with a
computational model. The three knowledge types are used to interpret the
variety of findings within a unitary conceptual framework.
 
 


From tkelley at arl.army.mil  Mon Nov 13 10:51:13 2000
From: tkelley at arl.army.mil (Troy Kelley)
Date: Mon, 13 Nov 2000 10:51:13 -0500
Subject: Difficulty of remembering a series of items
Message-ID: <mailman.68.1617128263.6729.act-r-users@act-r.psy.cmu.edu>

 
Anyone out there know of any research that pertains to the difficulty of
remembering a series of unrelated items.  For example, how much harder is
it to recall one item vs three?  Is it just a straight probability?.. you
have a 1/3 chance of remembering one item out of three, and that you have a
1/1 chance of remembering just one?   I would like to incorporate
activation levels into the formula.  So, If they were related items, my
activation levels (through spreading activation) takes that into accout, so
perhaps I can use the same formula as long as I use activation levels?  For
example, suppose I have 3 items with activation levels of .3, .4 and .1..
how much harder is it to remember that than one item with an activation
level of .2?
 
Hope this is clear,
 
Thanks
Troy Kelley
 


From tkelley at arl.army.mil  Mon Nov 13 13:13:07 2000
From: tkelley at arl.army.mil (Troy Kelley)
Date: Mon, 13 Nov 2000 13:13:07 -0500
Subject: List Memory
Message-ID: <mailman.79.1617128263.6729.act-r-users@act-r.psy.cmu.edu>

 
 
 
 
 
To:   Troy Kelley <tkelley at arl.army.mil>
cc:
Subject:  Re: List Memory
 
 
>Anyone out there know of any research that pertains to the difficulty of
>remembering a series of unrelated items.  For example, how much harder is
>it to recall one item vs three?  Is it just a straight probability?.. you
>have a 1/3 chance of remembering one item out of three, and that you have
a
>1/1 chance of remembering just one?   I would like to incorporate
>activation levels into the formula.  So, If they were related items, my
>activation levels (through spreading activation) takes that into accout,
so
>perhaps I can use the same formula as long as I use activation levels?
For
>example, suppose I have 3 items with activation levels of .3, .4 and .1..
>how much harder is it to remember that than one item with an activation
>level of .2?
 
-The short answer is that the probability of retrieving a series of
-unrelated items is a non-linear function that can depend on a number of
-experimental factors and a number of effects such as recency and primacy.
-But chapter 7 has some formulas resulting from the activation calculus
that
-are fairly simple but do a good job of approximating reality.
 
-As for your last question, i.e. the probability of remembering items as a
-function of their activation, it is given directly by the Retrieval
-Probability Equation (3.7 p. 74) and the Chunk Choice Equation (3.9 p.
77).
-Those equations will be unified in ACT-R 5.0 but they have to do for now.
 
I guess I wasn't quite clear as I should have been.  Given that ACT-R has
taken care of the effects of recency and primacy and incorporated those
effects into activation levels, then how hard would it be to do each
retrieval assuming that each activation level is above the necessary
threshold.  Are all retrievals equally as likely assuming that each chunk
is above the necessary threshold?  Or should you incorporated a probability
estimate for each group of retrievals (probability of 3 items is .33,
probability of 2 items is .5).  For example, suppose you had to recall X &
Y, and X & Y had certain activation levels associated with each.  Now
suppose you had to remember Z, how much easier is it to remember Z, where Z
has a given activation level as well, and Z is unrelated to X or Y?  Let's
say that the activation level of X is .26 and Y is .31 and Z is .25, so is
it still easier to remember Z, just because there is only one retrieval
involved? or are both groups equally as "easy" simply because each one has
an activation level above a given threshold?
 
Troy
 
 


From Wolfgang.Schoppek at uni-bayreuth.de  Tue Nov 14 04:40:06 2000
From: Wolfgang.Schoppek at uni-bayreuth.de (Wolfgang Schoppek)
Date: Tue, 14 Nov 2000 10:40:06 +0100
Subject: List Memory
Message-ID: <mailman.90.1617128263.6729.act-r-users@act-r.psy.cmu.edu>

> I guess I wasn't quite clear as I should have been.  Given that ACT-R has
> taken care of the effects of recency and primacy and incorporated those
> effects into activation levels, then how hard would it be to do each
> retrieval assuming that each activation level is above the necessary
> threshold.  Are all retrievals equally as likely assuming that each chunk
> is above the necessary threshold?  Or should you incorporated a probability
> estimate for each group of retrievals (probability of 3 items is .33,
> probability of 2 items is .5).  For example, suppose you had to recall X &
> Y, and X & Y had certain activation levels associated with each.  Now
> suppose you had to remember Z, how much easier is it to remember Z, where Z
> has a given activation level as well, and Z is unrelated to X or Y?  Let's
> say that the activation level of X is .26 and Y is .31 and Z is .25, so is
> it still easier to remember Z, just because there is only one retrieval
> involved? or are both groups equally as "easy" simply because each one has
> an activation level above a given threshold?
>
> Troy
 
I guess the key to solving that confusion is that you must not "assume"
activation levels. They are not arbitrary values but the result of the
mechanisms of ACT-R that come into play depending on your specific model (e.g.,il
,
rehearsing less items in the same amount of time will return a higher baseleve
l
for each item). Once you have the activation levels determined by the model,
the equations mentioned by Christian will give you estimates about probability
 
of retrieval.
 
Wolfgang
 


From tkelley at arl.army.mil  Wed Nov 22 12:00:25 2000
From: tkelley at arl.army.mil (Troy Kelley)
Date: Wed, 22 Nov 2000 12:00:25 -0500
Subject: Post Doctorial Position at ARL
Message-ID: <mailman.101.1617128263.6729.act-r-users@act-r.psy.cmu.edu>

---------------------------
 
 
Troy Kelley <tkelley at arl.army.mil> on 11/22/2000 11:16:35 AM
 
 
 
 
 
To:   act-r-users/arl at andrew.cmu.edu, soar-group at eecs.umich.edu,
      goms-list at usabilityfirst.com
cc:
Subject:  Post Doctorial Position at ARL
 
 
Hello,
 
We have a position available at the Army Research Laboratory in
Aberdeen, MD beginning Jan., 1, 2001 for a AI researcher skilled in LISP.
 
The position is for one year,  with a possibility of renewal after one
year.  There is also the possibility for a permanent position.  Pay would
be in the 40s.
This is a National Research Council sponsored position.  There is also a
possibility
of an American Society of Engineering Education sponsored position as well.
 
Anyone interested can contact me directly via e-mail.
tkelley at arl.army.mil
 
Troy Kelley
Army Research Laboratory
 
 
 


From ema at msu.edu  Sat Nov 25 18:04:01 2000
From: ema at msu.edu (Erik M. Altmann)
Date: Sat, 25 Nov 2000 15:04:01 -0800
Subject: Decay lives (new paper)
Message-ID: <mailman.113.1617128264.6729.act-r-users@act-r.psy.cmu.edu>

Content-Type: text/plain; charset="us-ascii" ; format="flowed"
 
 
Cognitive psychology textbooks often cite Waugh and Norman (1965) as
evidence that forgetting in short-term memory is determined by interference
alone -- i.e., that there is no decay.  This poses a problem for ACT-R,
in which decay is a primary mechanism of forgetting.
 
We show that there are indeed decay effects in the Waugh and Norman data.
The data contain a presentation rate x serial position interaction that,
despite being reliable, replicable, and rather obvious, has been ignored
since the data were published.  Our model explains the interaction in a
structural (parameter-insensitive) way, in terms of decay moderating the
effects of interference.  This moderating relationship is based on
functional logic, which says that without decay the cognitive system would
grind to a halt due to interference.  The same model provides an improved
account of Peterson and Peterson (1959) -- data classically used to argue
for decay alone.  To put the work in context, we survey a set of undergrad
psychology textbooks for their conclusions about decay and interference.
 
Decay in the model is described by ACT-R's (optimized) base-level learning
equation, and interference by the chunk choice equation.  ACT-R's retrieval
threshold is not a factor -- retrieval probability is solely a function of
target activation relative to distractor activation.
 
Manuscript and models are at http://www.msu.edu/~ema/ramodels.  This is
recently submitted, and comments are welcome.
 
Erik and Chris.
 
 
A Relative Activation Model of Memory:
Integrating Decay and Interference to Reinterpret Some Classical Data
 
Erik M. Altmann and Christian D. Schunn
 
A long-running debate in cognitive psychology concerns the processes 
of forgetting: How is information lost from human memory? The debate 
has often focussed on the question of decay (indexed by time) versus 
interference (indexed by the addition of distracting information to 
memory).  The relative activation model is a formal model grounded in 
extant memory theory that incorporates both processes.  The 
functional basis for integrating the two processes is that decay 
mitigates interference.  The empirical justification is that the 
model explains a previously- ignored interaction in a data set that 
is commonly used to argue for interference over decay (Waugh & 
Norman, 1965). The model also provides an improved account of data 
that were an original basis of decay theory (Peterson & Peterson, 
1959).
-- 
 
~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~
Erik M. Altmann
Department of Psychology
Michigan State University
East Lansing, MI  48824
517-353-4406 (voice)
517-353-1652 (fax)
ema at msu.edu
http://www.msu.edu/~ema
~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~
--============_-1236929052==_ma============
Content-Type: text/html; charset="us-ascii"
 
<!doctype html public "-//W3C//DTD W3 HTML//EN">
<html><head><style type="text/css"><!--
blockquote, dl, ul, ol, li { margin-top: 0 ; margin-bottom: 0 }
 --></style><title>Decay lives (new paper)</title></head><body>
<div><br></div>
<div>Cognitive psychology textbooks often cite Waugh and Norman
(1965) as</div>
<div>evidence that forgetting in short-term memory is determined by
interference</div>
<div>alone -- i.e., that there is no decay.&nbsp; This poses a
problem for ACT-R,</div>
<div>in which decay is a primary mechanism of forgetting.<br>
&nbsp;<br>
We show that there are indeed decay effects in the Waugh and Norman
data. </div>
<div>The data contain a presentation rate x serial position
interaction that,</div>
<div>despite being reliable, replicable, and rather obvious, has been
ignored<br>
since the data were published.&nbsp; Our model explains the
interaction in a<br>
structural (parameter-insensitive) way, in terms of decay moderating
the<br>
effects of interference.&nbsp; This moderating relationship is based
on<br>
functional logic, which says that without decay the cognitive system
would<br>
grind to a halt due to interference.&nbsp; The same model provides an
improved<br>
account of Peterson and Peterson (1959) -- data classically used to
argue<br>
for decay alone.&nbsp; To put the work in context, we survey a set of
undergrad<br>
psychology textbooks for their conclusions about decay and
interference.<br>
<br>
Decay in the model is described by ACT-R's (optimized) base-level
learning<br>
equation, and interference by the chunk choice equation.&nbsp;
ACT-R's retrieval<br>
threshold is not a factor -- retrieval probability is solely a
function of<br>
target activation relative to distractor activation.<br>
<br>
Manuscript and models are at http://www.msu.edu/~ema/ramodels.&nbsp;
This is<br>
recently submitted, and comments are welcome.<br>
</div>
<div>Erik and Chris.</div>
<div><br></div>
<div><br></div>
<div><font face="Times New Roman" size="+2" color="#000000">A
Relative Activation Model of Memory:<br>
Integrating Decay and Interference to Reinterpret Some Classical
Data</font></div>
<div><font face="Times New Roman" size="+2"
color="#000000"><br></font></div>
<div><font face="Times New Roman" size="+2" color="#000000">Erik M.
Altmann and Christian D. Schunn</font></div>
<div><font face="Times New Roman" size="+2"
color="#000000"><br></font></div>
<div><font face="Times" size="+2" color="#000000">A long-running
debate in cognitive psychology concerns the processes of forgetting:
How is information lost from human memory? The debate has often
focussed on the question of decay (indexed by time) versus
interference (indexed by the addition of distracting information to
memory).&nbsp; The relative activation model is a formal model
grounded in extant memory theory that incorporates both
processes.&nbsp; The functional basis for integrating the two
processes is that decay mitigates interference.&nbsp; The empirical
justification is that the model explains a previously- ignored
interaction in a data set that is commonly used to argue for
interference over decay</font><font face="Times New Roman" size="+2"
color="#000000"> (Waugh &amp; Norman, 1965)</font><font face="Times"
size="+2" color="#000000">. The model also provides an improved
account of data that were an original basis of decay
theory</font><font face="Times New Roman" size="+2" color="#000000">
(Peterson &amp; Peterson, 1959)</font><font face="Times" size="+2"
color="#000000">.</font></div>
 
<div>-- <br>
<br>
~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~<br>
Erik M. Altmann<br>
Department of Psychology<br>
Michigan State University<br>
East Lansing, MI&nbsp; 48824<br>
517-353-4406 (voice)&nbsp; <br>
517-353-1652 (fax)<br>
ema at msu.edu<br>
http://www.msu.edu/~ema<br>
~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~</div>
</body>
</html>
--============_-1236929052==_ma============--