During the call Ben asked about my system.. So here is an ultra simplified description.. First couple paragraphs start out as the same mediocre sounding cognitive platforms we get too much of ..
* but what is different is the idea of instincts as motivated actions (based on static memories.. not based on the usual "preferences" system everyone else talks about).* *. here we go* *Sensory registers* When an environmental stimulus is detected by the senses, it is briefly available in what Atkinson and Shiffrin called the sensory registers (buffers). composed of multiple registers, many for each sense. The sensory registers do not process the information carried by the stimulus, but rather detect and hold that information for use in short-term processing. For this reason Atkinson and Shiffrin also called the registers ``buffers". *EventCalc Sequence (secondary registers - EC) * When events are emitted from the motivational system, (below) it generates artifacts in several sets of registers. As noted above, sensory registers do not process information. How sensory processing is done is that these registers are compared and updated to a specific level of detail. The supplementary EventCalc that was emitted by the motivational system are stored as a record of the VSpace (instead of the sensory content) This is how EventCalc Sequencing contains the required information to recreate the sensory data. (Logicmoo's internal "game" server implements a copy of this) *Proto-narrative* *Sequencing (ternary registers - PN) * Refers to information that is registered by an imaginary Echoic system. Which is part of the output that comes from the motivational system (that accompanies the previous) It corresponds to the Phonological loop as described by Baddeley (1968). As with EventCalc Sequencing, ProtoNarrative Sequences only hold superficial aspects of a narrative. (Just enough to trigger EventCalc Sequences .. (there are at least 10-30 ECs per 4-7 PNs ) Specifically, Protolanguage Sequencing has a duration of between 1.5 and 5 seconds depending on context. *Motivational System Sequencing * The system bootstraps with several EventCalc sequences. Such sequences are emitted into secondary registers, as noted above, and cause LM489 to observe sensory another agent doing tasks. These tasks may seem high level such as “searching” or doing “experimentation” or “listening” but LM489 does not in any way understand these sensory anomalies. (they merely play out into the secondary buffer.) LM489's* motivation is only to replicate* (in their world) a copy of an imaginary agent (well itself) doing them. Not doing of this is uncomfortable to LM489 ( Later on LM489 believes “they’ thought of the task and they no longer believe they are copying an innate memory) RE-Generation of the above three types of sequences must already exist to be played into their respective buffers. Each sequence is triggered by another sequence (no requirement they be from the same type) Important to understand that these are not triggered by sensors the way an attention system would work.. But instead by the need for LM489 "to go along" with the secondary buffer … A wiki page for this system is https://logicmoo.org/xwiki/bin/view/Main/Psychology/InstinctualKnowledge/ (i have a few 1000 pages of docs that go into much better details than i gave here.. (and even more code for it)) My system is admittedly optimized for carrying out longer operations ( There are several of these game copies (each contain the 4 part system) running at once in the brain ).. It is probably harder for it to learn conditions/responses .( but isn't that true of animals and humans?) What was important about it: This allows resources to be "some exact size" . It is the level of the 3 sequencing systems that set the "level of detail the world comes in" (not some attention system) From the outside human it only "appears" it was using an attention system. This allows sequences to hold their positions for longer and shorter periods of time and allows the "experience of time" to be set by the agent's sequences (instead of a world clock) -Douglas On Thu, Aug 12, 2021 at 11:59 PM Ben Goertzel <[email protected]> wrote: > A couple blog posts, posted just now, summing up some of the ideas I > plan to discuss in the group tomorrow/l > > Paraconsistent Interzones: > https://bengoertzel.substack.com/p/paraconsistent-interzones > > Open-Ended Motivations: > https://bengoertzel.substack.com/p/open-ended-motivations > > ;) > ben > > On Thu, Aug 12, 2021 at 8:56 PM Ben Goertzel <[email protected]> wrote: > > > > August 13, 2021, 7:30AM-9AM US Pacific time > > > > TOPIC: Paraconsistent Models of Open-Ended AGI Motivations. (leader: > > Ben Goertzel) > > > > URL for video-chat: https://singularitynet.zoom.us/my/benbot > > > > ... open to all AGI enthusiasts/researchers who want to listen and/or > > converse on the theme... > > > > ben > > > > -- > > Ben Goertzel, PhD > > http://goertzel.org > > > > “He not busy being born is busy dying" -- Bob Dylan > > > > -- > Ben Goertzel, PhD > http://goertzel.org > > “He not busy being born is busy dying" -- Bob Dylan > > -- > You received this message because you are subscribed to the Google Groups > "opencog" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To view this discussion on the web visit > https://groups.google.com/d/msgid/opencog/CACYTDBfS%2BvEt9BzQmDpGYBSur7MCdTow2Fvg00dHwFXhFj3Wfw%40mail.gmail.com > . > -- You received this message because you are subscribed to the Google Groups "opencog" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/opencog/CAER3M5kvhuB-5eHpkw%3D-qq5GS86a9Drpaqab4_fB0T8cx-OFsw%40mail.gmail.com.
