From trafton at itd.nrl.navy.mil Fri Mar 1 16:03:14 2002 From: trafton at itd.nrl.navy.mil (Greg Trafton) Date: 01 Mar 2002 16:03:14 -0500 Subject: postdoc position Message-ID: ACT-R projects going on here in addition to that advertised below (putting some ACT-R models on a robot (!)). If folks are interested, please let me know. Postdoc Position Naval Research Laboratory Washington, DC We are currently looking for 1 or 2 postdocs. Brief descriptions of current projects are below. The cognition of complex visualizations: We are examining how experts (scientists and weather forecasters, primarily) comprehend and use complex graphs and visualizations. Our current emphasis is on the higher level reasoning processes that go into extracting information from meteorological visualizations and scientific visualizations as experts work on them. Background or experience with a hard science (physics or meteorology/oceanography in particular) would be an advantage, though not necessary. The study of interruptions in complex domains: We are examining how people deal with interruptions, particularly the resumption process (i.e., what people do when the resume an interrupted process or goal). For all projects, we use a combination of protocol analysis, experimental methods (running subjects in experiments), and computational cognitive modeling (primarily in the ACT-R framework). Candidates are not expected to be experts in all methods, but the more familiarity with each the better. A list of papers and background on these projects is available at: http://www.aic.nrl.navy.mil/~trafton/ If you have an interest in these projects, please contact me via email (trafton at itd.nrl.navy.mil) or phone (202-767-3479). The appointment would be for one year; an additional two years would be contingent on availability of funds and performance. This appointment would be an NRC postdoc position; only American citizens are eligible. Salary range: $40,000 - $55,000 per year. -- Greg Trafton, Ph.D (trafton at itd.nrl.navy.mil) http://www.aic.nrl.navy.mil/~trafton/ From r.m.young at herts.ac.uk Wed Mar 20 03:57:48 2002 From: r.m.young at herts.ac.uk (Richard M Young) Date: Wed, 20 Mar 2002 08:57:48 +0000 Subject: Competing "retrieval" rule vs compiled "habit" rule? Message-ID: I have what I think is a technical "how to" query about ACT-R, although it sometimes turns out that what I think are "how to" issues are really deep architectural ones. Because I'm a bit confused, the question will be a bit muddled, so I'd better explain the context. I have been supervising a student project which is to re-implement the Byrne & Bovair 3CAPS model of "post-completion errors" (PCEs) in a way natural to ACT-R. PCEs occur when the final step of some procedure is to be taken after its main goal has been achieved, in which case the final step has a tendency to be forgotten. A classic example is making 2 photocopies of a document but leaving the original on the copier. All great fun, but raising some issues that I'm finding tricky. The particular aspect I'm finding difficult is capturing the effect of WM load. The data on PCEs, although sparse, suggest that the error is more likely to occur under WM stress -- essentially you get distracted. My problem is that if one models PCE as arising from production competition (with noise), then conflict resolution between productions is NOT sensitive to WM, i.e. to declarative item activation -- only to the E=PG-C on the competing rules. (I had better say that my framework for thinking about this problem is ACT-R 4.0, not 5.0. I have no objection to 5.0, but I teach 4.0 because it links with the 1998 textbook.) Imagine this scenario, which is not a PCE but more like a prospective memory task, but I think will make the point. Suppose you drive home on a regular route, so we assume (at least, I assume) that the decision-making productions for driving that route are all compiled into productions, i.e. "habit" productions. (That would happen automatically in 5.0, yes?) Suppose on a particular day that you are meant to stop at the bread shop and buy some bread. So (a) you say to yourself "Must stop at bread shop, must stop at bread shop, ..." and with any luck, when you see the shop you stop and buy some bread. We might also believe (b) that if you were listening to the radio at that point and got interested in one of the news items, or if you were engaged in conversation with a passenger, or if some interesting outside event happened at that point on the drive, then the "must buy bread" would slip your mind (i.e. WM loss), and you are more likely to forget to stop and buy. How does this map into ACT-R? In scenario (a), we presumably have competition between, on the one hand, a compiled "habit" rule saying essentially "at this point, keep on driving", and on the other hand, an "interpretative" rule saying something like "if you have an intention in mind and the time is right, then do it". Now, I'm already uncomfortable with the idea of competition between those two rules. The *interpretative* rule is presumably a very general one, with a long history of being applied to totally different tasks in totally different circumstances, so I find it difficult to make sense of that rule competing with a very specific "habit" rule on the grounds of their respective statistical histories -- it seems like comparing chalk and cheese. Worse, that rule conflict is not, according to ACT-R, in any way sensitive to the activation level of the declarative items that the interpretative rule must retrieve. (True, if the activation is below threshold then the interpretative rule will fail, but that seems a very crude mechanism, and also *assumes* that the interpretative rule will necessarily win the conflict resolution.) So the outcome is unaffected by whether you're firmly saying to yourself "stop and buy bread" or whether that has slipped your mind. And that seems to contradict both data and also common sense. Can anyone unmuddle me about this? I'm also struck by the fact that there are no "retrieval" productions in 5.0, as I understand it. Does this make a difference to the problem? -- Richard From ja+ at cmu.edu Wed Mar 20 08:12:02 2002 From: ja+ at cmu.edu (John Anderson) Date: Wed, 20 Mar 2002 08:12:02 -0500 Subject: Competing "retrieval" rule vs compiled "habit" rule? Message-ID: non-trivial, but I think I can see the broad outline of two kinds of solution: One is to relate it to declarative activation as you wish. If so, then one could assume that there were more slots of the goal filled in the high-load condition and a lower probability of retrieving the key chunk "stop and buy bread" because of less associate activation. I do not understand why the probability of this would be unaffected by rehearsal of the chunk as you seem to imply. The other solution is to relate working memory load to competition among procedures rather than declarative resource competition. Then the model would go something like 1. Assume as you did that there is a "time is right and do it" production. 2. Assume as you do that it is competing with other productions. 3. Lets assume that it has lower priority than these other productions. 4. In the low load condition, just driving, there is only an occasional production firing and so lots of opportunity for the time-is-right production to insert itself. 5. In the high load condition, conversation and driving, there are the additional conversation productions that crowd out the time-is-right production. A variant of this would to assume that only the conversation productions have priority over the time-is-right production. I guess that the key concept here is the idea that in the low-load condition there is a lot of slack time when no production is firing. This is very much a feature of 5.0 due to the thorough infusion of R/PM into the model. -- ========================================================== John R. Anderson Carnegie Mellon University Pittsburgh, PA 15213 Phone: 412-268-2788 Fax: 412-268-2844 email: ja at cmu.edu URL: http://act.psy.cmu.edu/ From hedderik at van-rijn.org Wed Mar 20 08:05:51 2002 From: hedderik at van-rijn.org (Hedderik van Rijn) Date: Wed, 20 Mar 2002 14:05:51 +0100 Subject: Competing "retrieval" rule vs compiled "habit" rule? Message-ID: > I have what I think is a technical "how to" query about ACT-R, although > it > sometimes turns out that what I think are "how to" issues are really > deep Although I never modeled a task like you described, I've done some work both on interpretative rules vs proceduralized rules and on two interfering tasks, both in 5.0. So, I'll just report what my ideas were while reading your example of getting bread while driving home. The major change from 4 to 5 with respect to the buffers might be useful in this example. Although I'm not really into the motor and perceptual stuff, so I'll probably violate some ideas about these, the new buffer structure might be able to help. 1. proceduralized rules do the driving. They tell the motor-system to do its driving job. The motor-system does its driving, and reports back any abnormalities in the motor-buffer. 2. proceduralized rules do the perception. As soon as something enters the perceptual buffer, they take care of it. I.e., tell the motor-system to stop for that old lady that wants to cross the street. 3. interpretative rules check the retrieval buffer, if anything is available there, then process that chunk. Now, if the 1 and 2 rules are rather generic (i.e., either have a generic goal if that still exists, or all the goals in the system are of the same chunk-type, as in some of Niels' proceduralization models), they can always interfere. Because being proceduralized and often successfully used, they have a high utility, and will therefore (luckily, as I'd rather not miss that large truck that suddenly comes to a stop) interfere any process based on more costly or less successful interpretative rules. However, if nothing is available in the motor and perception buffers, the system is free to donate some time to the important task of not forgetting to buy bread. (Note that in 5 the model initiates a retrieval, and is then free to do whatever it want... it doesn't lock up until the chunk is actually retrieved.) But, if involved in a conversation (i.e., buffers are filled with other information than "getting bread") or a lot of things are happing on the road (i.e., the proceduralized rules do not allow the interpretative rule to fire), the "remember to get some bread" chunks will likely not be retrieved until parking the car at home. > rule on the grounds of their respective statistical histories -- it > seems > like comparing chalk and cheese. Which, in some countries, is not too ridiculous... Hedderik. From ngbrann at sandia.gov Wed Mar 20 09:46:10 2002 From: ngbrann at sandia.gov (Brannon, Nathan G) Date: Wed, 20 Mar 2002 07:46:10 -0700 Subject: Competing "retrieval" rule vs compiled "habit" rule? Message-ID: (G). NGB ________________________________________ Nathan G. Brannon, Ph.D. Senior Member Technical Staff Sandia National Laboratories P.O. Box 5800, MS 0830 Albuquerque, NM 87185 (505) 845-7055 > ---------- > From: Richard M Young > Sent: Wednesday, March 20, 2002 1:57 AM > To: act-r-users at andrew.cmu.edu > Subject: Competing "retrieval" rule vs compiled "habit" rule? > > ACTers: > > I have what I think is a technical "how to" query about ACT-R, although it > sometimes turns out that what I think are "how to" issues are really deep > architectural ones. > > Because I'm a bit confused, the question will be a bit muddled, so I'd > better explain the context. I have been supervising a student project > which is to re-implement the Byrne & Bovair 3CAPS model of > "post-completion > errors" (PCEs) in a way natural to ACT-R. PCEs occur when the final step > of some procedure is to be taken after its main goal has been achieved, in > which case the final step has a tendency to be forgotten. A classic > example is making 2 photocopies of a document but leaving the original on > the copier. All great fun, but raising some issues that I'm finding > tricky. > > The particular aspect I'm finding difficult is capturing the effect of WM > load. The data on PCEs, although sparse, suggest that the error is more > likely to occur under WM stress -- essentially you get distracted. My > problem is that if one models PCE as arising from production competition > (with noise), then conflict resolution between productions is NOT > sensitive > to WM, i.e. to declarative item activation -- only to the E=PG-C on the > competing rules. > > (I had better say that my framework for thinking about this problem is > ACT-R 4.0, not 5.0. I have no objection to 5.0, but I teach 4.0 because > it > links with the 1998 textbook.) > > Imagine this scenario, which is not a PCE but more like a prospective > memory task, but I think will make the point. Suppose you drive home on a > regular route, so we assume (at least, I assume) that the decision-making > productions for driving that route are all compiled into productions, i.e. > "habit" productions. (That would happen automatically in 5.0, yes?) > Suppose on a particular day that you are meant to stop at the bread shop > and buy some bread. So (a) you say to yourself "Must stop at bread shop, > must stop at bread shop, ..." and with any luck, when you see the shop you > stop and buy some bread. We might also believe (b) that if you were > listening to the radio at that point and got interested in one of the news > items, or if you were engaged in conversation with a passenger, or if some > interesting outside event happened at that point on the drive, then the > "must buy bread" would slip your mind (i.e. WM loss), and you are more > likely to forget to stop and buy. > > How does this map into ACT-R? In scenario (a), we presumably have > competition between, on the one hand, a compiled "habit" rule saying > essentially "at this point, keep on driving", and on the other hand, an > "interpretative" rule saying something like "if you have an intention in > mind and the time is right, then do it". Now, I'm already uncomfortable > with the idea of competition between those two rules. The > *interpretative* > rule is presumably a very general one, with a long history of being > applied > to totally different tasks in totally different circumstances, so I find > it > difficult to make sense of that rule competing with a very specific > "habit" > rule on the grounds of their respective statistical histories -- it seems > like comparing chalk and cheese. > > Worse, that rule conflict is not, according to ACT-R, in any way sensitive > to the activation level of the declarative items that the interpretative > rule must retrieve. (True, if the activation is below threshold then the > interpretative rule will fail, but that seems a very crude mechanism, and > also *assumes* that the interpretative rule will necessarily win the > conflict resolution.) So the outcome is unaffected by whether you're > firmly saying to yourself "stop and buy bread" or whether that has slipped > your mind. And that seems to contradict both data and also common sense. > > Can anyone unmuddle me about this? > > I'm also struck by the fact that there are no "retrieval" productions in > 5.0, as I understand it. Does this make a difference to the problem? > > -- Richard > > > > From Wolfgang.Schoppek at uni-bayreuth.de Wed Mar 20 11:02:24 2002 From: Wolfgang.Schoppek at uni-bayreuth.de (Wolfgang Schoppek) Date: Wed, 20 Mar 2002 17:02:24 +0100 Subject: Competing "retrieval" rule vs compiled "habit" rule? Message-ID: declatative structures in my ACT-Fly model ( http://www.uni-bayreuth.de/departments/psychologie/ACTR-2000/ ). The model "translates" GOMS analyses into ACT-R. The assumption is that declarative steps (that contain the operators in a slot) are linked together purely by associations (no symbolic links). This produces step-skipping especially under high WM-load (see http://www.uni-bayreuth.de/departments/psychologie/iccm-2000.html). The drawback of this conception is that it cannot model the process of proceduralization. However, I'm not sure about up to what hierarchical level proceduralization takes place. Maybe, the solution of representing methods as declarative structures is appropriate for higher level methods? A paper by Hunt and Lansman also popped to my mind: They argue for the assumption that production rules activate each other: Hunt E., & Lansman M. (1986). Unified model of attention and problem solving. Psychological Review, 93, 446-461. -- WS ----------------------------------------------------------------- Dr. Wolfgang Schoppek Universitaet Bayreuth Tel.: +49 921 554140 http://www.uni-bayreuth.de/departments/psychologie/wolfgang.html ----------------------------------------------------------------- From ema at msu.edu Wed Mar 20 10:15:11 2002 From: ema at msu.edu (Erik M. Altmann) Date: Wed, 20 Mar 2002 11:15:11 -0400 Subject: Competing "retrieval" rule vs compiled "habit" rule? Message-ID: There's some discussion of post-completion error and prospective memory from an (one) ACT-R perspective in the recent Altmann & Trafton paper on memory for goals (linked to the URL below, if your most recent issue of Cognitive Science isn't handy). The relevant gist is that to avoid a post-completion error, or to act on an intention in a prospective memory task, there has to be an association created between a cue and a declarative representation of the to-be-performed goal. We argue that the way people manage to avoid most post-completion errors is that through practice, or deliberate effort, they create associative links between the final "real" goal (getting the cash from the atm) and the post-completion goal (removing the atm card from the slot). In your bread-errand scenario, you would have to have created an associative link between some cue related to the bread shop (a mental image of the sign, or a nearby landmark), such that when you attend to that cue in the actual environment, it will prime the goal. The effects of WM load seem to fit naturally into this framework, because the greater the activation of distractor elements at that moment (elements related to the scintillating conversation you are engaged in), the less likely the relevant goal is to "intrude" on your thoughts such that you break off, apologize to your conversation partner, and park at the bread shop. In terms of implementation, as I've sketched the model it implicates declarative memory, associative learning, and very general retrieval productions that simply take whatever's most active in memory at a given moment, as a function of base-level activation combined with associative activation. Though this seems to fit in the general spirit of the rational memory model, I'm not sure it fits so well with current modeling practices. Erik. -- ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 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 ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ From rvb at Cs.Nott.AC.UK Wed Mar 20 11:48:59 2002 From: rvb at Cs.Nott.AC.UK (Roman Belavkin) Date: Wed, 20 Mar 2002 16:48:59 +0000 Subject: Competing "retrieval" rule vs compiled "habit" rule? Message-ID: > I have what I think is a technical "how to" query about ACT-R, although it > sometimes turns out that what I think are "how to" issues are really deep > architectural ones. > > Because I'm a bit confused, the question will be a bit muddled, so I'd > better explain the context. I have been supervising a student project > which is to re-implement the Byrne & Bovair 3CAPS model of "post-completion > errors" (PCEs) in a way natural to ACT-R. PCEs occur when the final step > of some procedure is to be taken after its main goal has been achieved, in > which case the final step has a tendency to be forgotten. A classic > example is making 2 photocopies of a document but leaving the original on > the copier. All great fun, but raising some issues that I'm finding tricky. I am not sure if I answer the question, but here is my thought. If I was modelling this kind of task, then I would model it using, what I call, a "chunk-activation-based goal stack". Traditional way is to use the standard stack of Act-r, which is too idealistic: you push a subgoal, when the subgoal is completed you always return to the parent goal. In the case of the photocopier you have something like this: make 2 copies, 0 copies made original in the photocopier, 0 copies made press the button "make 2 copies" original in the photocopier, 2 copies made make 2 copies, 2 copies made task complete you can realise the same thing without using the goal stack, but using activations of the goal chunks. Remember that in Act-r activation decays with time. So, more recently created goals have higher activations. Thus they have higher chance to be retrieved. You can have a simple system of two rules to handle the creation of subgoals and their retrievals. the first rule creates subgoals, but not pushing them on the stack: if the goal is not complete and there is an undefined variable ==> create a subgoal to define the variable focus on the subgoal the second rule would retrieve an unfinished goal if the goal is complete and there exists an unfinished goal ==> focus on the unfinished goal (a bit like soar, isn't it?) The rules above will have to have a syntax depending on your chunks format, but the idea is simple, and it worked for me. I used it in my Yerkes-Dodson mouse model for navigation tasks. For example, when you would like to proceed from point A to C, but there is an obstacle between A and C, so you can't go directly. In this case using the first rule I put off the goal for a while as unfinished, and then create another goal to go to some point B. When I arrive to point B, the system registers the goal as finished, so the second rule retrieves my previously unfinished goal. Being located at point B I may find that now I can go directly from B to C. My final path would be A->B->C. When I experimented with such rules what I find interesting is that the system does not necessarily follow the order of the times of subgoals creation, unlike the order strictly defined by the goal stack. If you increase the activation noise variance (or :ans parameter), then the order of retrievals does not follow recency principle as much as with no noise. The error you referred to may well be produced when the model fails to retrieve an unfinished goal. If a subject is very excited (particularly by the fact the that main goal is complete), then you can model it increasing the activation noise. hope you find it helpful Roman --------------------------------------------------------------------- Roman Belavkin School of Computer Science and Information Technology University of Nottingham Jubilee Campus Phone: (0115) 846-6532 Wollaton Road E-mail: rvb at Cs.Nott.AC.UK Nottingham NG8 1BB http://www.cs.nott.ac.uk/~rvb/ --------------------------------------------------------------------- From byrne at acm.org Wed Mar 20 11:46:08 2002 From: byrne at acm.org (Mike Byrne) Date: Wed, 20 Mar 2002 10:46:08 -0600 Subject: Competing "retrieval" rule vs compiled "habit" rule? Message-ID: the whole list: -------------------------------------------------- First, let me say I'm honored that my old master's thesis causes consideration of deep architectural issues. Please keep those citations coming! Second, I am also teaching an ACT-R class this semester which requires projects. One student will be doing his project on goal management in general (that is, without a stack) and one student will be modeling PCEs in particular. So I'm about to run into the same kinds of issues. Now, let me say that I believe that 5.0 is more appropriate than 4.0 for this kind of thing if for no other reason than because we've done away with the goal stack (at least in priciple). I've always thought of PCE's as being a good piece of evidence against goal stacks. Anyway, more recent work on PCEs in my lab suggests that doing a task containing a potential PCE over and over does reduce the error rate, but that there's a cost--subjects never get faster at the PCE step. This suggests deliberation, probably retrieval-based, about what to do at that point. WM load would obviously interfere with such deliberations. Under one 5.0 view of goal management, the completion of any goal not explicitly tied to some other goal (i.e. any goal that does not lead automatically to a new goal when it completes) will result in an attempt or attempts to retrive unsatisfied goals from declarative memory, so that the system knows what to do next. Combine this with something like display-based (or environment-based) problem-solving strategies, and you should* get an account of habit-capture errors and PCEs. My idea goes like this: [1] The system completes some goal, and attempts to retrive an unsatisfied goal (of which "get bread" might be one) because it needs to know what to do next. [2] This retrieval fails (due, perhaps, to high WM load), so the system tries to (re)construct a current goal based on the present display (or environment). [3] The environment contains all the cues for "drive home", so that's the goal that gets (re)constructed. Note that if WM load is low, however, then there's a better chance that the "get bread" goal will be retrieved, and thus no goal (re)construction attempted. So you get WM-driven error. (* I say "should" because I have no built model--yet--that does this.) Notice there's not even a mention of conflict resolution in that story--I don't think it's a necessary part of the account. On a more general note, I think the removal of the goal stack will cause us as a community to have to think a lot harder about goal management as a general problem and this will lead to interesting results on a number of fronts, especially with respect to error behavior. > I'm also struck by the fact that there are no "retrieval" > productions in 5.0, as I understand it. Does this make a > difference to the problem? I'm not sure exactly what you mean by "retrieval" production here--productions in 5.0 can both harvest and initiate retrievals, but the same production cannot both initiate and harvest the same retrieval (as in 4.0). However, I do think 5.0's asynchronous, buffer-based retrievals will change how conflict resolution effectively works for some models. One can now attempt a retrieval with very little real cost. Let's say you have two strategies for solving a problem, A, which requires a lengthy and expensive retrieval, and B, which requires a bunch of production-based computation or perceptual-motor activity (e.g. reading off a display). In ACT-R 4, productions which implement A & B will compete, with only one winner. Learning over this will be *highly* sensitive to the success rate and duration of the retrieval, because merely attempting the retrieval can cost the system all kinds of time--you get the double-whammy for retrieval failure because it hurts success rate as well as costing a lot of time. Now, enter 5.0 world. Instead of conflict, you have one production which both initiates the A's retreival and starts the ball rolling on B's processing. B gets going while the retrieval is being attempted. If the retrieval eventually comes back as a failure, no worries, just continue with B. If it's successful, great, stop doing B and just proceed with the A-based strategy. This completely changes the nature of the conflict resolution problem--essentially the only time you'll really even enter CR here is when the retrieval completes--you want the production that checks the retrieval buffer to be guaranteed to win in CR so the buffer actually gets checked. Of course, A could be your general-purpose kind "retrieve an intention" and B could be "keep driving". If A fails, no problem, just keep driving. This is an alternative account to the previous one, but it doesn't cover PCEs so I favor the other one--but this should illustrate my point about how the more parallel world of 5.0 can dramatically change things like CR and strategy selection. I'd be interested to hear other people's thoughts on this, especially Marsha, who's done a lot of work on both WM and CR. Marsha? -Mike =========================================================== Mike Byrne, Ph.D. byrne at acm.org Assistant Professor, Psychology Department Rice University, MS-25 http://chil.rice.edu/byrne/ 6100 Main Street +1 713-348-3770 voice Houston, TX 77005-1892 +1 713-348-5221 fax From N.A.Taatgen at ai.rug.nl Thu Mar 21 04:07:57 2002 From: N.A.Taatgen at ai.rug.nl (Niels Taatgen) Date: Thu, 21 Mar 2002 10:07:57 +0100 Subject: Competing "retrieval" rule vs compiled "habit" rule? Message-ID: attachment to an email you send off. I almost thought I caught one of those in Erik's email, but later on realized he only promised that the paper he mentioned could be found on his homepage. I think many people already made good points about the topic, so I would like to elaborate on the issue of working-memory capacity, as this is mentioned by a couple of people. My intuition is that there can be two reasons for forgetting to buy bread. One reason is that the chunk encoding this task has lost its activation and cannot be retrieved anymore. Although this might be a legitimate reason, I do not think it is the case that occurs most often, as these types of errands tend to pop up later in your memory, even though the context may do nothing to prompt it. This means that working-memory capacity in the usual ACT-R sense (retrieval failures tied to W) cannot be the problem. Now, what's happening in this car driving/chatting/thinking about the bread scenario is that we have a heavy case of multi-tasking. So you might consider it a small miracle that you think about the bread after all! Frank (Lee) and I have been trying to model multi-tasking, and it turns out declarative memory is the main bottle-neck. No, that's not really true, all buffers can be bottle-necks, but declarative memory is one we can do something about, as production compilation removes declarative retrievals. If declarative memory is "doing nothing", as is the case when you are driving a car, because you are waiting for perceptual things to happen and motor things to conclude, there is plenty of time to engage declarative memory in other things (daydreaming, having a conversation, thinking about the bread). You need some way to flag the busy-state of declarative memory, so Frank and I do this in the goal. In our very small demo-model, there is a main task A, and a secondary task B. The main thing for the model to do is task A, but when it has some slack time it can work on B as well. It turned out that task B only gets a chance to be processed in parallel with A once A has been proceduralized, producing some slack time for declarative memory (by the way: Frank and I but together a paper for the cognitive science conference on this http://www.ai.rug.nl/~niels/CogSci2002Final.pdf). A similar situation is going on the the bread-on-the-way scenario: if you are too much engaged in conversation, there is no opportunity to steal slack time at the right moment. But if the conversation gets boring, and your minds drifts off... well you might just remember the bread! Niels =============================================== Niels Taatgen - University of Groningen, Artificial Intelligence web: http://www.ai.rug.nl/~niels email: niels at ai.rug.nl =============================================== From byrne at acm.org Thu Mar 21 10:36:53 2002 From: byrne at acm.org (Mike Byrne) Date: Thu, 21 Mar 2002 09:36:53 -0600 Subject: Competing "retrieval" rule vs compiled "habit" rule? Message-ID: > Let me start by mentioning my favorite PCE: forgetting to add a > promised attachment to an email you send off. People ask me about this all the time, and I'm not yet convinced this is actually a PCE. While it seems pretty clear to me that the goal of using a photocopier is to get photocopies (the 21st century version of this error is leaving your original on a flatbed scanner, by the way), it is not 100% clear to me that there's a main goal that's been completed when an attachment has been neglected. I tend to be of the opinion that this is a habit-capture error. So for instance, when I type "-Mike" at the bottom of my email, habit says the next thing to do is execute the keystroke that does a send. So one could think of it more as a capture error, but it certainly has a PCE feel to it as well. Definitely a grey area. -Mike =========================================================== Mike Byrne, Ph.D. byrne at acm.org Assistant Professor, Psychology Department Rice University, MS-25 http://chil.rice.edu/byrne/ 6100 Main Street +1 713-348-3770 voice Houston, TX 77005-1892 +1 713-348-5221 fax From fum at univ.trieste.it Thu Mar 21 11:15:00 2002 From: fum at univ.trieste.it (danilo_fum) Date: Thu, 21 Mar 2002 17:15:00 +0100 Subject: Competing "retrieval" rule vs compiled "habit" rule? Message-ID: Dear Richard and dear ACTers: here are our two (Euro) cents of thoughts concerning the bread-errand task and related debate. In fact, we filled a whole coffee break discussing this problem; in evaluating our reply, however, please note that we drank Espresso. According to our point of view, to fully grasp (and model) the bread errand task it is necessary to: 1. Get an adequate understanding of how people drive home following a known route. This probably has to do, as many of you pointed out, with the employment of perceptual cues (affordances) that activate motor programs. 2. Get an adequate understanding of the role of situated cognition in perspective memory tasks. A good starting point could be represented by Erik's proposal, in which the perceptual cue triggers the retrieval of a suspended intention (analogously to what happens in the Patalano & Seifert model). In our case, the specific goal "stop at the bread shop" could have been associated with the bread shop sign. A set of "perspective" productions (maybe a general retrieval mechanism?) could retrieve the suspended intention after being triggered by the perceptual cue, provided that it is able to win the competition with the other productions related with the driving and conversation tasks. It is also possible to hypothesize that, in this and similar situations, an explicit set of "rehearsal" productions could be used to increase the activation level of an intention, and to make its retrieval more likely. This could be similar to what happens in the WM phonological loop, or in the strengthening mechanism of Altmann and Trafton TOH model. Mike raised an interesting issue concerning the retrieval costs in strategy selection. There is some evidence (Rickard, 1997) for the view that if two strategies compete for the choice of the next problem solving operator, one being based on the explicit retrieval of previous instances and the other relying on a procedural mechanism, only one of them could be used due to attentional limitations. Data from a new experiment we carried out with the toads-and-frogs puzzle (paper submitted at the CogScie conference) seem to support this point, too. Any way, the retrieval cost issue is deeply related with the new buffer-oriented architecture. Maybe it could be untangled only by fully understanding the ontological status of the buffers themselves (psychological structures, computational devices, neural-based constructs?) We eagerly look forward to learn something more at the next ACT-R workshop. ======================================= Fabio Del Missier Danilo Fum Andrea Stocco Department of Psychology University of Trieste via S.Anastasio, 12 I - 34134 Trieste Italy e-mail: {delsmisfa,fum,stocco}@univ.trieste.it Phone: +39 040 676-2708 Fax: +39 040 676-2757 ================================================== From hepu.deng at rmit.edu.au Sun Mar 24 18:56:27 2002 From: hepu.deng at rmit.edu.au (Hepu Deng) Date: Mon, 25 Mar 2002 10:56:27 +1100 Subject: [ICONIP02-SEAL02-FSKD02] Call for Papers [Apologies if you receive this announcement more than once.] Message-ID: ---------------------------------------------------------------------- 9th International Conference on Neural Information Processing (ICONIP'02) 4th Asia-Pacific Conference on Simulated Evolution And Learning (SEAL'02) International Conference on Fuzzy Systems and Knowledge Discovery (FSKD'02) ---------------------------------------------------------------------- November 18 - 22, 2002, Orchid Country Club, Singapore http://www.ntu.edu.sg/home/nef Organized by: School of Electrical and Electronic Engineering Nanyang Technological University, Singapore Sponsored by: Asia-Pacific Neural Network Assembly SEAL & FSKD Steering Committees Singapore Neuroscience Association In Co-Operation with: IEEE Neural Network Society International Neural Network Society European Neural Network Society SPIE Supported by: Lee Foundation Singapore Exhibition & Convention Bureau ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ CALL FOR PAPERS, SPONSORSHIPS, AND SPECIAL SESSION PROPOSALS ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ICONIP'02, SEAL'02, and FSKD'02 will be jointly held in Orchid Country Club, Singapore from November 18 to 22, 2002. The conferences will not only feature the most up-to-date research results in natural and arti- ficial neural systems, evolutionary computation, fuzzy systems, and knowledge discovery, but also promote cross-fertilization over these exciting and yet closely-related areas. Registration to any one of the conferences will entitle a participant to the technical sessions and the proceedings of all three conferences, as well as the conference banquet, buffet lunches, and tours to two of the major attractions in Singapore, i.e., Night Safari and Sentosa Resort Island. Many well- known researchers will present keynote speeches, panel discussions, invited lectures, and tutorials. About Singapore --------------- Located at one of the most important crossroads of the world, Singapore is truly a place where East and West come together. Here you will find Chinese, Indian, and Malay communities living together, their long established cultures forming a unique backdrop to a clean and modern garden city. English is spoken everywhere and is the common business language of all. Few places on earth promise such a delight for the palate, with gourmet cuisine from over 30 countries. Exotic resorts in neighboring countries are only a short bus/ferry ride away. Orchid Country Club (OCC) ------------------------- The venue for this year's conferences is at one of Singapore's premier country clubs, a 25-minute bus ride from the city. Away from the hustle and bustle of downtown Singapore, the tranquil setting of the resort is ideal for serious technical discussions with an accommodating space and ambience for relaxation. Not to miss out on the splendor of downtown Singapore, the organizer has also secured good quality and affordable accommodation in the heart of the city with pre-arranged transport to/from the OCC. For golf enthusiasts, OCC is equipped with the largest computerized driving range in South East Asia and boasts of a 27-hole golf course with facilities for night golfing, ideal for relaxation after each day of technical discussions. Visit the OCC website at http://www.orchidclub.com Night Safari and Sentosa Resort Island -------------------------------------- It is said that a visit to Singapore is not complete without making a trip to two of the Republic's famous attractions. The only one of its kind in the world, the Night Safari provides a setting for visitors to experience what it is like to observe animals in their nocturnal habitat. The island of Sentosa offers some unique attractions and a visit there will also provide a glimpse and imagery of Singapore's past and present. Visits to these two attractions will be included as recreation for the joint conference. (Websites: http://www.zoo.com.sg/safari/, http://www.sentosa.com.sg) Topics of Interest ------------------ The joint conferences welcomes paper submissions from researchers, practitioners, and students worldwide in but not limited to the following areas. ICONIP'02: ~~~~~~~~~ ARTIFICIAL NEURAL MODELS - Learning algorithms, Neural modeling and architectures, Neurodynamics NATURAL NEURAL SYSTEMS - Neuroscience, Neurobiology, Neuro- physiology, Brain imaging, Learning and memory COGNITIVE SCIENCE - Perception, emotion, and cognition, Selective attention, Vision and auditory models HARDWARD IMPLEMENTATION - Artificial retina & cochlear chips HYBRID SYSTEMS - Neuro-fuzzy systems, Evolutionary neural nets, etc APPLICATIONS - Bioinformatics, Finance, Manufacturing, etc. SEAL'02: ~~~~~~~ THEORY - Co-evolution, Coding methods, Collective behavior METHODOLOGY - Evolution strategies, Genetic algorithms, Genetic programming, Molecular and quantum computing, Evolvable hardware, Multi-objective optimization, Ant colony, Artificial ecology EVOLUTIONARY LEARNING - Artificial life, Bayesian evolutionary algorithms HYBRID SYSTEMS - Evolutionary neuro-fuzzy systems, Soft computing APPLICATIONS - Scheduling, Operations research, Design, etc FSKD'02: ~~~~~~~ THEORY AND FOUNDATIONS - Fuzzy theory and models, Uncertainty management, Statistical & probabilistic data mining, Computing with words, Rough sets, Intelligent agents METHODS AND ALGORITHMS - Classification, Clustering, Information retrieval & fusion, Data warehousing & OLAP, Fuzzy hardware, Visualization, Decision trees, Data preprocessing HYBRID SYSTEMS - Evolutionary neuro-fuzzy systems, Soft computing APPLICATIONS - Control, Optimization, Natural language processing, Forecasting, Human-computer interaction, etc. Special Sessions ---------------- The conferences will feature special sessions on specialized topics to encourage in-depth discussions. To propose a special session, email the session title, name of the conference under which the special session will be organized, contact information of the organizer(s), and a short description on the theme and topics covered by the session to Xin Yao, Special Sessions Chair (x.yao at cs.bham.ac.uk), with a copy to Lipo Wang, General Chair (Cc: elpwang at ntu.edu.sg). Sponsorship ----------- The conferences will offer product vendors a sponsorship package and/or an opportunity to interact with conference participants. Product demonstration and exhibition can also be arranged. For more information, please visit the conference website or contact Tong Seng Quah, Sponsorship/Exhibition Chair (itsquah at ntu.edu.sg), with a copy to Lipo Wang, General Chair (Cc: elpwang at ntu.edu.sg). Keynote Speakers ---------------- Shun-ichi Amari, RIKEN Brain Science Institute, Japan David Fogel, Natural Selection, Inc., USA Mitsuo Kawato, ATR, Japan Xin Yao, The University of Birmingham, UK Lotfi A. Zadeh, University of California, USA Registration Fee ---------------- The registration fee for regular participants before August 15, 2002 is S$680 (approximately US$370 as at February 6, 2002), which includes the proceedings, lunches, banquet, and tours. Submission of Papers -------------------- Authors are invited to submit electronic files (postscript, pdf or Word format) through the conference home page. Papers should be double-column and use 10 pt Times Roman or similar fonts. The final version of a paper should not exceed 5 pages in length. A selected number of accepted papers will be expanded and revised for possible inclusion in edited books and peer-reviewed journals, such as "Soft Computing" and "Knowledge and Information Systems: An International Journal" by Springer-Verlag. Important Dates --------------- Paper/Summary Deadline : April 30, 2002 Notification of Acceptance : July 15, 2002 Final Paper/ Registration : August 15, 2002 Honorary Conference Chairs -------------------------- Shun-ichi Amari, Japan Hans-Paul Schwefel, Germany Lotfi A. Zadeh, USA International Advisory Board ---------------------------- Sung-Yang Bang, Korea Meng Hwa Er, Singapore David B. Fogel, USA Toshio Fukuda, Japan A. Galushkin, Russia Tom Gedeon, Australia Zhenya He, China Mo Jamshidi, USA Nikola Kasabov, New Zealand Sun-Yuan Kung, USA Tong Heng Lee, Singapore Erkki Oja, Finland Nikhil R. Pal, India Enrique H. Ruspini,USA Harcharan Singh, Singapore Ah Chung Tsoi, Australia Shiro Usui, Toyohashi, Japan Lei Xu, China Benjamin W. Wah, USA Donald C. Wunsch II, USA Xindong Wu, USA Youshou Wu, China Yixin Zhong, China Jacek M. Zurada, USA Advisor ------- Alex C. Kot, Singapore General Chair ------------- Lipo Wang, Singapore Program Co-Chairs ----------------- ICONIP'02: Kunihiko Fukushima, Japan Soo-Young Lee, Korea Jagath C. Rajapakse, Singapore SEAL'02: Takeshi Furuhashi, Japan Jong-Hwan Kim, Korea Kay Chen Tan, Singapore FSKD'02: Saman Halgamuge, Australia Special Sessions: Xin Yao, UK Finance Chair ------------- Charoensak Charayaphan, Singapore Local Arrangement Chair ----------------------- Meng Hiot Lim, Singapore Proceedings Chair ----------------- Farook Sattar, Singapore Publicity Co-Chairs ------------------- Hepu Deng, Australia Chunru Wan, Singapore Li Weigang, Brazil Zili Zhang, Australia Sponsorship/Exhibition Chair ---------------------------- Tong Seng Quah, Singapore Tutorial Chair -------------- P. N. Suganthan, Singapore For More Information -------------------- Please visit the conference home page or contact: Lipo Wang, ICONIP'02-SEAL'02-FSKD'02 General Chair School of Electrical and Electronic Engineering Nanyang Technological University Block S2, 50 Nanyang Avenue, Singapore 639798 Email: elpwang at ntu.edu.sg Phone: +65 6790 6372 Fax: +65 6792 0415 Conference Secretariat ---------------------- ICONIP'02-SEAL'02-FSKD'02 Secretariat Conference Management Center/CCE, NTU Administration Annex Building #04-06 42 Nanyang Avenue, Singapore 639815 Email: nef at ntu.edu.sg Fax: +65 6793 0997 _______________________________________________ Iconip02-seal02-fskd02 mailing list Iconip02-seal02-fskd02 at spock.bf.rmit.edu.au http://spock.bf.rmit.edu.au/mailman/listinfo/iconip02-seal02-fskd02 From cl at andrew.cmu.edu Mon Mar 25 17:04:22 2002 From: cl at andrew.cmu.edu (Christian Lebiere) Date: Mon, 25 Mar 2002 17:04:22 -0500 Subject: 2002 ACT-R summer school and workshop Message-ID: 1) The deadline for summer school applications is a week away (April 1). 2) The mailing list was recently updated. If your address was incorrectly included or if you would like to unsubscribe, please send email directly to me at cl at cmu.edu or to act-r-users-request at andrew.cmu.edu Thanks, Christian] EIGHTH ANNUAL ACT-R SUMMER SCHOOL AND WORKSHOP ============================================== Carnegie Mellon University - July/August 2002 ============================================= ACT-R is a cognitive theory and simulation system for developing cognitive models for tasks that vary from simple reaction time to air traffic control. The most recent advances of the ACT-R theory were detailed in the recent book "The Atomic Components of Thought" by John R. Anderson and Christian Lebiere, published in 1998 by Lawrence Erlbaum Associates, and in the online proceedings of the 2001 ACT-R Post-Graduate Summer School (act.psy.cmu.edu). Each year, a week-plus summer school is held to train researchers in the use of the ACT-R system, followed by a three-day workshop to enable new and current users to exchange research results and ideas. The Eighth Annual ACT-R Summer School and Workshop will be held at Carnegie Mellon University in Pittsburgh in July/August 2002. SUMMER SCHOOL: The summer school will take place from Wednesday July 24 to Thursday August 1, with the intervening Sunday free. This intensive 8-day course is designed to train researchers in the use of ACT-R for cognitive modeling. It is structured as a set of eight units, with each unit lasting a day and involving a morning theory lecture, a web-based tutorial, an afternoon discussion session and a homework assignment which participants are expected to complete during the day and evening. Computing facilities will be provided. Summer school attendees are expected to attend the workshop as part of their training. To provide an optimal learning environment, admission is limited to a dozen participants, who must submit by April 1 an application consisting of a curriculum vitae and a statement of purpose. Demonstrated experience with a modeling formalism similar to ACT-R will strengthen the application. Applicants will be notified of admission by April 15, with early notification upon request. Admission to the summer school is free. A stipend of up to $750 is available to advanced graduate students for reimbursement of travel, housing and meal expenses. To qualify for the stipend, students must be US citizens and join to their application a letter of reference from a faculty member. WORKSHOP: The workshop will take place from the morning of Friday August 2 to Sunday August 4 at noon. Mornings will be devoted to research presentations, each lasting about 20 minutes plus questions. Participants are invited to present their ACT-R research by submitting a one-page abstract with their registration. Afternoons will feature more research presentations as well as discussion sessions and instructional tutorials. Suggestions for the topics of the tutorials and discussion sessions are welcome. Admission to the workshop is open to all. The early registration fee (before July 1) is $100 and the late registration fee (after July 1) is $125. The summer school and workshop were scheduled to coincide with the Cognitive Science conference hosted by George Mason University in Fairfax, Virginia from August 8 to 10 (www.hfac.gmu.edu/~cogsci). Housing and computing facilities will be available at CMU from August 5 to 7 for participants to the summer school or workshop who wish to stay on to work on their ACT-R projects and collaborate with other researchers until the start of Cogsci2002. A registration form is appended below. Additional information (detailed schedule, etc.) will appear on the ACT-R Web site (http://act.psy.cmu.edu/) when available or can be requested at: 2002 ACT-R Summer School and Workshop Psychology Department Attn: Helen Borek Baker Hall 345C Fax: +1 (412) 268-2844 Carnegie Mellon University Tel: +1 (412) 268-3438 Pittsburgh, PA 15213-3890 Email: helen+ at cmu.edu ________________________________________________________ Eighth Annual ACT-R Summer School and Workshop July 24 to August 4, 2002 at Carnegie Mellon University in Pittsburgh REGISTRATION ============ Name: .................................................................. Address: .................................................................. .................................................................. .................................................................. Tel/Fax: .................................................................. Email: .................................................................. Summer School (July 24 to August 1): ........ (check here to apply) ==================================== Applications are due APRIL 1. Acceptance will be notified by APRIL 15. Applicants MUST include a curriculum vitae and a short statement of purpose. Demonstration of experience with a modeling formalism similar to ACT-R, such as a completed model, should also be included in the application. A stipend of up to $750 is available for the reimbursement of travel, lodging and meal expenses (receipts needed). To qualify for the stipend, the applicant must be a graduate student with US citizenship and include with the application a letter of reference from a faculty member. Check here to apply for stipend: ........ Workshop (August 2 to 4): ........ (check here to register) ========================= Presentation topic (optional - include one-page abstract with registration): ........................................................................... Registration fee: Before July 1: $100 ... After July 1: $125 ... The fee is due upon registration. Please send checks or money orders only. We cannot accept credit cards. HOUSING ======= Housing is available in Resnick House, a CMU dormitory that offers suite-style accommodations. Rooms include air-conditioning, a semi-private bathroom and a common living room for suite-mates. The rates are $267.75/week/person or $38.25/night/person for single rooms and $192.50/week/person or $27.50/night/person for double rooms. Housing reservations will be taken after acceptance to the summer school. Do not send money. See http://www.housing.cmu.edu/conferences/ for further housing information. To reserve a room in Resnick House, fill in the dates and select one of the three room options: I will stay from ................ to ................ 1. ... I want a single room 2. ... I want a double room and I will room with ................ 3. ... I want a double room. Please select a roommate of ....... gender ROOM PAYMENT IS DUE UPON CHECK-IN. DO NOT SEND MONEY. A block of rooms has been set aside at the nearby Wyndham Garden Hotel at the discounted rate of $99.00 per room per night (parking $12.00). For reservations, call 877-662-6242 before July 11 and mention that you are attending the ACT-R Workshop. Shuttles and buses from the hotel to the university are available. Send this form to: 2002 ACT-R Summer School and Workshop Psychology Department Attn: Helen Borek Baker Hall 345C Fax: +1 (412) 268-2844 Carnegie Mellon University Tel: +1 (412) 268-3438 Pittsburgh, PA 15213-3890 Email: helen+ at cmu.edu From mailman-owner at spock.bf.rmit.edu.au Sun Mar 31 14:18:29 2002 From: mailman-owner at spock.bf.rmit.edu.au (mailman-owner at spock.bf.rmit.edu.au) Date: Mon, 1 Apr 2002 05:18:29 +1000 Subject: spock.bf.rmit.edu.au mailing list memberships reminder Message-ID: spock.bf.rmit.edu.au mailing list memberships. It includes your subscription info and how to use it to change it or unsubscribe from a list. You can visit the URLs to change your membership status or configuration, including unsubscribing, setting digest-style delivery or disabling delivery altogether (e.g., for a vacation), and so on. In addition to the URL interfaces, you can also use email to make such changes. For more info, send a message to the '-request' address of the list (for example, iconip02-seal02-fskd02-request at spock.bf.rmit.edu.au) containing just the word 'help' in the message body, and an email message will be sent to you with instructions. If you have questions, problems, comments, etc, send them to mailman-owner at spock.bf.rmit.edu.au. Thanks! Passwords for act-r-users at andrew.cmu.edu: List Password // URL ---- -------- iconip02-seal02-fskd02 at spock.bf.rmit.edu.au taatep http://spock.bf.rmit.edu.au/mailman/options/iconip02-seal02-fskd02/act-r-users%40andrew.cmu.edu