From lionel.marquis at unil.ch Thu Sep 12 19:19:50 2013 From: lionel.marquis at unil.ch (Lionel Marquis) Date: Fri Sep 13 01:19:50 2013 +0200 Subject: Question on a specific difference between ACT* and ACT-R Message-ID: Hi, I have only basic level in ACT-R or similar models of memory functioning. The aspect I am most interested in is the "spreading activation" part of ACT-R. Here is my question. In earlier versions of the model (i.e., ACT*) there was a parameter p to account for the fact that "the total activation emitted by a node is less than its own activation, that is, if ai is the activation of node i, it sends p*ai units of activation to all connected nodes where p<1. Because of this factor, the amount of activation decreases exponentially with distance from source" (Anderson & Pirolli 1974: 794). In later versions (i.e., ACT-R), this parameter p had dropped from the equations. What are the (theoretical or empirical) reasons for removing p? Or is the idea behind factor p somehow still accounted for, but in a different form? Thanks a lot in advance for your help on this particular point ! Best regards, Lionel Marquis ================================== Lionel Marquis Institut d'Etudes Politiques et Internationales University of Lausanne Switzerland B???iment G???polis 4779 T???: +41 21 692 31 56 Fax: +41 21 692 31 45 E-mail: Lionel.Marquis at unil.ch Web: http://www.unil.ch/iepi/page14150.html From db30 at andrew.cmu.edu Fri Sep 13 13:22:26 2013 From: db30 at andrew.cmu.edu (Dan Bothell) Date: Fri Sep 13 12:22:26 2013 -0500 Subject: [ACT-R-users] Question on a specific difference between ACT* and ACT-R Message-ID: --On Friday, September 13, 2013 1:19 AM +0200 lionel.marquis at unil.ch wrote: > > In later versions (i.e., ACT-R), this parameter p had dropped from the > equations. What are the (theoretical or empirical) reasons for removing p? Or > is the idea behind factor p somehow still accounted for, but in a different > form? > > ACT-R does not have such a parameter because unlike ACT*, it only has one level of spreading activation. The reasons for the change from multiple levels of spreading activation to a single level can be found in "The Adaptive Character of Thought". Hope that helps, Dan _______________________________________________ ACT-R-users mailing list ACT-R-users at act-r.psy.cmu.edu https://mailman.srv.cs.cmu.edu/mailman/listinfo/act-r-users From lionel.marquis at unil.ch Thu Sep 12 19:32:38 2013 From: lionel.marquis at unil.ch (lionel.marquis at unil.ch) Date: Thu, 12 Sep 2013 23:32:38 -0000 Subject: [ACT-R-users] Question on a specific difference between ACT* and ACT-R Message-ID: <52324c16.15c8.ac909700.6e8fca15@unil.ch> Hi, I have only basic level in ACT-R or similar models of memory functioning. The aspect I am most interested in is the "spreading activation" part of ACT-R. Here is my question. In earlier versions of the model (i.e., ACT*) there was a parameter p to account for the fact that "the total activation emitted by a node is less than its own activation, that is, if ai is the activation of node i, it sends p*ai units of activation to all connected nodes where p<1. Because of this factor, the amount of activation decreases exponentially with distance from source" (Anderson & Pirolli 1974: 794). In later versions (i.e., ACT-R), this parameter p had dropped from the equations. What are the (theoretical or empirical) reasons for removing p? Or is the idea behind factor p somehow still accounted for, but in a different form? Thanks a lot in advance for your help on this particular point ! Best regards, Lionel Marquis ================================== Lionel Marquis Institut d'Etudes Politiques et Internationales University of Lausanne Switzerland B?timent G?opolis 4779 T?l: +41 21 692 31 56 Fax: +41 21 692 31 45 E-mail: Lionel.Marquis at unil.ch Web: http://www.unil.ch/iepi/page14150.html -------------- next part -------------- An HTML attachment was scrubbed... URL: From db30 at andrew.cmu.edu Fri Sep 13 12:21:58 2013 From: db30 at andrew.cmu.edu (db30 at andrew.cmu.edu) Date: Fri, 13 Sep 2013 16:21:58 -0000 Subject: [ACT-R-users] Question on a specific difference between ACT* and ACT-R In-Reply-To: <52324c16.15c8.ac909700.6e8fca15@unil.ch> References: <52324c16.15c8.ac909700.6e8fca15@unil.ch> Message-ID: <9ED24BE759BBDDCD4B9FDE65@actr6b.cmu.edu> --On Friday, September 13, 2013 1:19 AM +0200 lionel.marquis at unil.ch wrote: > > In later versions (i.e., ACT-R), this parameter p had dropped from the > equations. What are the (theoretical or empirical) reasons for removing p? Or > is the idea behind factor p somehow still accounted for, but in a different > form? > > ACT-R does not have such a parameter because unlike ACT*, it only has one level of spreading activation. The reasons for the change from multiple levels of spreading activation to a single level can be found in "The Adaptive Character of Thought". Hope that helps, Dan