[ACT-R-users] Question related to the Activation of Chunks and Base-Level learning

[db30 at andrew.cmu.edu]

--On Thursday, August 5, 2021 5:15 AM -0400 Arun Krishna <arunbkris at gmail.com> 
wrote:

>
>
> Hi,
>
> I have a question related to the Activation of Chunks and Base-Level
> learning.
>
> Problem Description:
>
> In the model I am using, the base level learning value is not getting changed
> when I use the chunk merging. I am using add-dm command with python API,
> where I add the chunks with same slot values and was expecting the base level
> learning for the chunks getting increased.
>
> Reference:
>
> ACT-R\tutorial\unit4\unit4.pdf,  4.3.2 Chunk Merging section.
>
> Code:
>
> In the model I am using I had set the sub symbolic computation
>
>  (sgp : esc t) and (sgp :rt -0.5).
>
> The complete setting for the model is given below
>
> "(sgp :v t :esc t :rt -0.5 :lf 0.4 :ans 0.5 :bll 0.5 :act nil :ncnar nil
> :ul t)
>
> (sgp :seed (200 4))"
>
>
>
> I am using the python API for add-dm
>
> actr.add_dm(
>
>             [chunkName, "isa", "navigation", "navigationStart", currentState,
> "navigationEnd", "state-16",
>
>              "navigationTrigger", trigger])
>
> This API is getting triggered with different chunk names. Chunk-1,
> chunk-2,…..chunk 50. All the chunk has same slot values and my expectation
> is that the base level learning of every chunk will be modified.
>
> But when I use (pprint-chunks-plus chunk-3) I get
>
>   SIMILARITIES  NIL
>
>    REFERENCE-COUNT  0
>
>    REFERENCE-LIST  NIL
>
>    SOURCE-SPREAD  0
>
>    LAST-BASE-LEVEL  0
>
>    BASE-LEVEL  NIL
>
>    CREATION-TIME  0
>
>    FAN-IN  NIL
>
>    C-FAN-OUT  0
>
>    FAN-OUT  0
>
>    IN-DM  NIL
>
>    ACTIVATION  0
>
>    BUFFER-SET-INVALID  NIL
>
> Could you please let me know how to modify the base level learning and
> reference count of these chunks (Which supposed to point to a single chunk
> with different reference names since the slot values are same) ?
>



Add-dm always "adds" a chunk to declarative memory - it does not merge it
into declarative memory.  There is a merge-dm command that works like
add-dm except that it merges the chunks into declarative memory.

However, for modifying the base-level the set-base-levels command allows
one to directly adjust the base-level parameters for a chunk.  It's use
varies depending upon how the parameters for base-level learning are set
(:bll and :ol), and you can find the details in the reference manual.

Hope that helps,
Dan