From cl+ at andrew.cmu.edu Fri Nov 1 13:38:07 1996 From: cl+ at andrew.cmu.edu (Christian J Lebiere) Date: Fri, 1 Nov 1996 13:38:07 -0500 (EST) Subject: Away next week. Message-ID: Dear ACT-R users, I will be away from the office (and any email connection) from tomorrow to Sunday November 10th. Email to act-r-users-request at andrew.cmu.edu will be answered promptly upon my return. Christian From cl+ at andrew.cmu.edu Fri Nov 15 17:10:58 1996 From: cl+ at andrew.cmu.edu (Christian J Lebiere) Date: Fri, 15 Nov 1996 17:10:58 -0500 (EST) Subject: Fourth Annual ACT-R Workshop and Summer School Message-ID: Here is the preliminary announcement for next year's workshop and summer school. Don't register quite yet since some details (e.g. housing) still need to be settled. Feel free to send me any feedback you might have before the announcement is finalized. Christian FOURTH ANNUAL ACT-R WORKSHOP AND SUMMER SCHOOL ============================================== Carnegie Mellon University - July/August 1997 ============================================== 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. Each year, a workshop and summer school is held to train researchers in the use of the system and to enable current users to exchange results and ideas. The fourth Annual ACT-R Workshop and Summer School will be held at Carnegie Mellon University in Pittsburgh from July 21 to August 5, 1997 (just before the Cognitive Science conference). The summer school will take place from Monday July 21 to Friday August 1, with the intervening weekend free. This intensive 10-day course is designed to train researchers in the use of ACT-R for cognitive modeling. It is structured as a set of 10 units, each involving a theory lecture, a web-based tutorial, a homework assignment and a discussion session. Computing facilities for the tutorials, assignments and research projects are provided. Due to space limitations, admission is limited to a dozen participants, who must submit an application consisting of a curriculum vitae and statement of purpose by MAY 1. Applicants will be notified of admission by MAY 15. A stipend of up to $750 is available for graduate students. To qualify for the stipend, students must be US citizens and join to their application a letter of reference from a faculty member. Admission to the summer school is free. The workshop will take place from Saturday August 2 to Tuesday August 5. Participants are invited to give a short presentations of their related research (completed, ongoing and planned projects are all acceptable). Mornings will be devoted to research presentations, discussion sessions (e.g. standardization of ACT-R terminology), and instructional tutorials (e.g. ACT-R parameter estimation). During afternoons and evenings, participants are encouraged to work on their research, exchange ideas and engage in unstructured discussions. The admission fee to the workshop is $50. Additional information (e.g. registration and housing forms and detailed schedule) will be available on the ACT-R web site at http://act.psy.cmu.edu/ or can be requested at: 1997 ACT-R Workshop and Summer School 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 cl+ at andrew.cmu.edu Wed Nov 20 14:30:00 1996 From: cl+ at andrew.cmu.edu (Christian J Lebiere) Date: Wed, 20 Nov 1996 14:30:00 -0500 (EST) Subject: "Working Memory" Message-ID: <8mYpksW00iWX46gs9E@andrew.cmu.edu> ACT-R has occasionally made a somewhat casual use of terminology, and one of the most commonly mentioned sticking points is the (ab)use of the term "working memory". In ACT-R, it is used as a synonym for declarative memory, i.e. the entire set of chunks encoding declarative knowledge, which is at odds with its use in much of the literature and some other production systems. In order to increase the clarity of the ACT-R educational materials and improve our consistency and communications, we are considering dropping the term "working memory" and instead use exclusively the term "declarative memory". Use of the term "wme" would also be discontinued and the term "chunk" would be used to refer to elements of declarative memory. Of course, this change would only bind ourselves and we would only set an example, not engage in an Orwellian attempt to purge the confusing words from the vocabulary of every ACT-R user. Probably the most salient impact of this change would be in the use of the system itself. The acronyms for the terms "working memory" and "working memory element" appear in many ACT-R commands (e.g. addwm, wmfocus, etc). The proposal is to continue providing the existing commands, but also to supply functionally equivalent alternatives with names more consistent with the new terminology, and to teach those alternatives in the educational material. The change would involve replacing the acronyms "wm" and "wme" by "chunk(s)" in most commands and "goal" in those that deal with the goal stack. For example, addwm would become addchunks and wmfocus would become goalfocus. Again, this change would be entirely backward-compatible and probably wouldn't take effect until next year. Let us know what you think, either by replying to us directly or to the list. Thanks, Christian From akyurek at bart.nl Thu Nov 21 08:22:51 1996 From: akyurek at bart.nl (Aladin Akyurek) Date: Thu, 21 Nov 1996 15:22:51 +0200 Subject: "Working Memory" Message-ID: >ACT-R has occasionally made a somewhat casual use of terminology, and >one of the most commonly mentioned sticking points is the (ab)use of the >term "working memory". In ACT-R, it is used as a synonym for >declarative memory, i.e. the entire set of chunks encoding declarative >knowledge, which is at odds with its use in much of the literature and >some other production systems. > That's right. Suggested change appears appropriate. Aladin From cl+ at andrew.cmu.edu Tue Nov 26 10:11:18 1996 From: cl+ at andrew.cmu.edu (Christian J Lebiere) Date: Tue, 26 Nov 1996 10:11:18 -0500 (EST) Subject: "Chunk" In-Reply-To: <8mYpksW00iWX46gs9E@andrew.cmu.edu> References: <8mYpksW00iWX46gs9E@andrew.cmu.edu> Message-ID: It seems that the approval for dropping "working memory" and using "declarative memory" exclusively is unanimous. This implies also dropping the use of the term "wme", which would become even more obscure that it is now. Most people support switching to "chunk", which we already use as a synonym for "wme". A couple of people expressed concern that our use of the term "chunk" is somewhat different from its use in the literature and in Soar and that confusion may result. At least one Soar user reported adapting well to the term after some initial confusion, but the concern is legitimate. It has been suggested to parallel the switch from "working memory" to "declarative memory" by replacing "wme" with "dme", for "declarative memory element". This has an appealing simplicity, and would make for a straightforward change in command names. I must confess a personal aversion to this new acronym, but I have gotten over worst things. "Chunk" and "dme" have symmetrical pros and cons. "Chunk" is a well known term in the literature, but it comes with a somewhat different shade of meaning than ours. "Dme", on the other hand, arrives with a clean slate but is a completely unknown word to impose on the community. I think that the familiarity of "chunk", both among ACT-R users where it has already been used as a synonym for "wme" and in the wider community, is the deciding factor. On a lighter note, it has been pointed out that "chunk" makes for longer command names and carries more potential for typos. The use of "wm" vs "wme", on the other hand, has been fairly confusing (Is it "addwm" or "addwme", "wmtype" or "wmetype"?). Finally, there would be the question of how to pronounce "dme". Would it be 'dimi', the straightforward generalization from "wme"? 'dime' has also been suggested. 'demi' would raise ACT-R's celebrity quotient. Or perhaps 'deme', to piggyback on the popularity of 'meme'. Possibilities are endless. As usual, let us know what you think, either by replying to us directly or to the list. Thanks, Christian From gray at gmu.edu Tue Nov 26 12:49:48 1996 From: gray at gmu.edu (Wayne Gray) Date: Tue, 26 Nov 1996 12:49:48 -0500 Subject: "Working Memory" In-Reply-To: <8mYpksW00iWX46gs9E@andrew.cmu.edu> Message-ID: Christian and folks, The need for a change from WME to something else is VERY much needed. However, let us not make Soar's mistake of using the word "chunks." Chunks has an informal but long-standing meaning in the psychological literature. Why not coin our own term, one that does not carry the surplus meaning of "chunks" or "wme." I propose calling these things exactly what they are: >>declarative memory elements<<< AKA DMEs, pronounced, "DiMES" The advantages are: (1) This is exactly what they are in act-r theory (2) No surplus meaning, no confusion with other uses of "chunks" or "WMEs" (3) A pronouncable acroynm. (4) Easy change from "wm" and "wme" to "dm" and "dme." **** The idea of going from wmfocus to goalfocus and changing other references to the goal stack seems like a good one to me. Cheers, Wayne #=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#= Wayne D. Gray Human Factors & Applied Cognitive Program George Mason University m/s 3f5 Fairfax, VA 22030 Phone: +1 (703) 993-1357 http://web.gmu.edu/departments/psychology/Wgray.html Work is infinite, time is finite, plan accordingly. #=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#= From tj at medinfo.ohio-state.edu Tue Nov 26 15:35:22 1996 From: tj at medinfo.ohio-state.edu (Todd R. Johnson) Date: Tue, 26 Nov 1996 15:35:22 -0500 Subject: Paper comparing Act-R and Soar Message-ID: <3.0.32.19961126153519.00a09760@earth.medinfo.ohio-state.edu> A non-text attachment was scrubbed... Name: not available Type: text/enriched Size: 949 bytes Desc: not available URL: From schunn+ at CMU.EDU Tue Nov 26 13:35:36 1996 From: schunn+ at CMU.EDU (Christian Schunn) Date: Tue, 26 Nov 1996 13:35:36 -0500 (EST) Subject: "Working Memory" In-Reply-To: References: Message-ID: <8manVsG00iV4I381RN@andrew.cmu.edu> I agree with Wayne that Soar did make a mistake in using the word 'chunk'---their notion of chunk has almost nothing to do with the historical, psychological sense of the word. However, I don't think we should generalize to the conclusion that no architecture should use the word 'chunk'. While the declarative memory element in ACT-R may have some subtle differences from the general meaning of chunk, it is very close in meaning: -Chunks refer to declarative knowledge stored in long-term memory -Chunks are built up through experience -Chunks have direct implications for memory span and retrieval times While it is true that previous incarnations of ACT-R lead to somewhat arbitrary uses of the old WMEs so that they frequently did not correspond to the common meaning of chunk, I think this is becoming less and less true as we think more seriously about the origins of knowledge in ACT-R models. The visual interface is one big step in that direction. Perhaps the best resolution of this issue is to ask non-users of ACT-R what they think of our use of 'chunk'. For long-time ACT-R users, it does not really matter what term we use---it is the novice user or reader for which this is an issue. My expectation is that psychologists will treat DMEs as a hacker's invention; whereas they will bring a lot of useful and relevant baggage to bear if they see the word 'chunk'. And on a more amusing note, 'dimes' reminds me of drug dealers. -Chris From klahr+ at andrew.cmu.edu Tue Nov 26 13:46:49 1996 From: klahr+ at andrew.cmu.edu (David Klahr) Date: Tue, 26 Nov 1996 13:46:49 -0500 (EST) Subject: "Working Memory" In-Reply-To: References: Message-ID: <0mangNq00iWWQ77Ww3@andrew.cmu.edu> This is all quite fascinating, and I hope that someone is archiving this exchange. For folks who write about this kind of stuff (eg. Bazerman, "Shaping written Knowledge") it may someday provide a interesting example of the influence of terminology of the development and propagation of concepts in a field. In fact, it is surprisingly difficult to pinpoint exactly when certain of our basic terminology was invented. What was the first occurrence -- in print -- of "means-ends-analysis" or "problem-space" or "weak methods" or "cognitive architecture" or "mental model"? One thing that is clearly unique about the current issue is both its explicit self-consciousness and its collective nature. I think that this did NOT apply to any of the other terminological decisions made earlier. My two cents worth is to suggest that folks consider the following list of synonyms for "chunk": lump, clump, clod, hunk, gob, nugget, and wad. --dk From rc7 at prism.gatech.edu Wed Nov 27 13:19:18 1996 From: rc7 at prism.gatech.edu (Richard Catrambone) Date: Wed, 27 Nov 1996 13:19:18 -0500 (EST) Subject: "Working Memory" In-Reply-To: References: <96Nov26.213057pst."27654"@bart.parc.xerox.com> <96Nov26.213057pst."27654"@bart.parc.xerox.com> Message-ID: Well, I don't know about the unambiguous pronunciation part. I could imagine pronouncing "deme" as DEEM or DEEMEE or demEE. Richard At 11:12 AM -0500 11/27/96, Christian J Lebiere wrote: ..... >Unlike dme, deme has an unambiguous pronounciation. We are trying to >replace our synonymous use of working memory and declarative memory, wme >and chunk, by a unique designation. Adopting a word that everyone >pronounces differently is not exactly a recipe for great communication. ______________________________________________________________________ Richard Catrambone School of Psychology Associate Professor Georgia Institute of Technology (Tel) (404) 894-2682 Atlanta, Georgia 30332-0170 (Fax) (404) 894-8905 USA Internet: rc7 at prism.gatech.edu WWW: http://www.prism.gatech.edu/~rc7/ ______________________________________________________________________ From gray at gmu.edu Tue Nov 26 22:52:42 1996 From: gray at gmu.edu (Wayne Gray) Date: Tue, 26 Nov 1996 22:52:42 -0500 Subject: "Working Memory" In-Reply-To: <8manVsG00iV4I381RN@andrew.cmu.edu> References: Message-ID: At 13:35 -0500 11/26/96, Christian Schunn wrote: >I agree with Wayne that Soar did make a mistake in using the word >'chunk'---their notion of chunk has almost nothing to do with the >historical, psychological sense of the word. However, I don't think we >should generalize to the conclusion that no architecture should use the >word 'chunk'. While the declarative memory element in ACT-R may have >some subtle differences from the general meaning of chunk, it is very >close in meaning: > >-Chunks refer to declarative knowledge stored in long-term memory >-Chunks are built up through experience >-Chunks have direct implications for memory span and retrieval times > Maybe . . . but perhaps the view from CMU is different than that of the rest of the world. So, for example, what would a non-cognitive modeler call a learned part of a procedure? I would call it a production and assign it to the act-r category of "procedure memory." Someone with a different bias/background (and there are a few of these in cognitive psychology) might want to call it a "chunk." So, for a non-modeler, would any of Chris' three points apply JUST to act-r declarative memory elements and not to act-r procedures? Perhaps we should follow Dave Klahr's suggestion and go with clod, or something. #=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#= Wayne D. Gray Human Factors & Applied Cognitive Program George Mason University m/s 3f5 Fairfax, VA 22030 Phone: +1 (703) 993-1357 http://web.gmu.edu/departments/psychology/Wgray.html Work is infinite, time is finite, plan accordingly. #=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#=#= From pirolli at parc.xerox.com Wed Nov 27 00:30:57 1996 From: pirolli at parc.xerox.com (Peter Pirolli) Date: Tue, 26 Nov 1996 21:30:57 PST Subject: "Working Memory" Message-ID: <96Nov26.213057pst."27654"@bart.parc.xerox.com> I sent a very terse vote for dme = "deme" but thought I should elaborate my choice in the context of the current discussion. I agree that "chunk" has unfortunately acquired some baggae from common usage in the memory community and the Soar community. I can forsee papers on "The Different Conceptions of 'Chunk' in...", which is reason enough to avoid it. dme = "dimee" or "dime" has no such baggage, is technically precise, but has the same air of hacker-jargon as "wimee" "wad" or "clog" might work just as "spin" "charm" "color" etc lost their original meanings when applied to physics and brought in a bit of whimsy. dme = "deme" on the other hand could be considered as a contraction of "declarative meme" which is very precise both theoretically and technically, and carries the right baggage. From Dawkins, a "meme" is the smallest unit of replicable cultural knowledge. This is very similar to the notion of a proposition (smallest unit of meaning that can be assigned a truth value). It also implies the notion of "replicable" with is very important within the physical symbol system hypothesis. This also provides a subtle tie from the theory of the Adaptive Character of Thought to one of the best known public writers on adpatation and evolution. --Pete From kirsch at c223.npt.nuwc.navy.mil Wed Nov 27 09:04:56 1996 From: kirsch at c223.npt.nuwc.navy.mil (Kirschenbaum_Susan) Date: Wed, 27 Nov 1996 09:04:56 -0500 Subject: FW: "Working Memory" Message-ID: Into the semantic frey: I sent the attached message to Christian shortly after receiving his first message. Now, given Dave Klahr's note about documenting the birth of a technical term/new technical meaning for an old term, I add it to the discussion. More than that, I wanted to respond to Chris Schunn's suggestion that "dme" might be seen as "a hacker's invention" -- I disagree, it is a useful shorthand for declarative memory element. Isn't that exactly what the term is intended to mean? It is more percise to invent a new term than to try to redefine and old one, even if some of the associated baggage is "useful and relevant." Chris's suggestion that non-ACT-Rers be consulted is a good way to resolve this question. Does chunk carry too much baggage or the wrong baggage (see Wayne's comments re procedural memory)? Does/can dme, lump, clump, clod, hunk, gob, nugget, or wad best carry the right, percise, technical meaning? My personal vote is for the new term, dme, but I am not a casual or novice user or reader of this literature. Susan _______________________________________________________________________________ From: Kirschenbaum_Susan on Mon, 25 Nov 1996 9:12 am Subject: RE: "Working Memory" To: Christian J Lebiere Christian, I think this is a good idea. My personal objection is that the new terms "addchunk," etc. are longer, giving me more opportunity to make typos. ;-) Seriously, the only real problem that the term "chunk" is used with a slightly different meaning by other architectures. This could cause confusion, both for the researcher who uses both architectures, and for the reader of this scientific literature. I wish there were an alternative term such as "declarative memory element," typed as "DME." What do you think? Susan ----------------------------------------------------------- Susan S. Kirschenbaum, Ph.D. Code 2214, Building 1171/1 Naval Undersea Warfare Center Division Newport, RI 02841 Tel: (401) 841-3354 Fax: (401) 841-4749 Kirsch at c223.npt.nuwc.navy.mil From tony.simon at psych.gatech.edu Wed Nov 27 08:33:46 1996 From: tony.simon at psych.gatech.edu (Tony J. Simon) Date: Wed, 27 Nov 1996 09:33:46 -0400 Subject: "Working Memory" In-Reply-To: <96Nov26.213057pst."27654"@bart.parc.xerox.com> Message-ID: Having now worked with both Soar and ACT I did initially have trouble with the use of the term chunk but, as Christian mentioned in his first message, I felt I adapted quite quickly. However, I am convinced that the points that have been made argue strongly against its adoption in a new formulation for ACT-R terminology. Why adopt something that already has a meaning that may (or may not) conflict with what it is intended to mean. As cognitive psychologists, we should know that we interpret new information in terms of what we already know. If there is anything that could confuse that interpretation it should be avoided. That suggests a vote against "chunk". Whimsical and amusing as clod, wad (or even splod) may be, I guess the same argument might apply. Therefore, by the same token, we should probably go with dme. As Pete Pirolli points out (how alliterative!), dme would inherit exactly the required meaning if any past knowledge can be bought to bear. By analogy from wme, it is a declarative memory element. Besides, it gives us all a great deal of fun comparing how the diverse cultures that make up this community choose to pronounce it. As they kept telling me when I first got to CMU, my "English" English led me to put the emPHAsis on the wrong sylLABle. So it could me d'mee or demmy (as in Moore), dimmee (as in wimee), deemee (as in Mimi (as in Recker)) {whoops got into lisp there}, etc etc etc. --Tony ------------------------------------------------------------------------ Tony J. Simon Assistant Professor of Psychology & Cognitive Science School of Psychology (Tel) (404) 894-2681 Georgia Institute of Technology (FAX) (404) 894-8905 Atlanta, GA 30332-0170 Internet: tony.simon at psych.gatech.edu WWW page http://www.prism.gatech.edu/~as53/ ------------------------------------------------------------------------ From cl+ at andrew.cmu.edu Wed Nov 27 11:12:27 1996 From: cl+ at andrew.cmu.edu (Christian J Lebiere) Date: Wed, 27 Nov 1996 11:12:27 -0500 (EST) Subject: "Working Memory" In-Reply-To: <96Nov26.213057pst."27654"@bart.parc.xerox.com> References: <96Nov26.213057pst."27654"@bart.parc.xerox.com> Message-ID: First of all, I think that it has been an excellent, very useful discussion. This is exactly what the act-r-users mailing list was intended for but until now it hadn't fulfilled its potential. Keep it up. I initially prefered "chunk" because I thought (still do) that "dme" (like "wme") is a lame, unpronouncable, funny-looking acronym. It is just not a word. However, I found Pete's argument for "dme" = "deme" so convincing that I would suggest to go one step further: instead of using "dme" and pronouncing it "deme", let us use "deme" itself as the word. This has a number of advantages. Deme is a much better word than dme. RDAR (Radio Detecting And Ranging) might have been a fine acronym for techies. But radar caught on among the public because it looks like a real word and has a clear pronounciation. So did laser, nylon, meme, etc. They are catchy. To paraphrase Woody allen, 90% of life is marketing. Unlike dme, deme has an unambiguous pronounciation. We are trying to replace our synonymous use of working memory and declarative memory, wme and chunk, by a unique designation. Adopting a word that everyone pronounces differently is not exactly a recipe for great communication. Deme can stand for both DEclarative MEmory and DEclarative Memory Element. This means that the past inconsistency in commands, with some using wm (e.g. addwm) and others using wme (e.g. modwme), can be solved. Addwm becomes adddeme and modwme becomes moddeme (or add-deme and mod-deme, but that is a debate for another day). Christian