# Full Text: Beacons - pre emulation of social cortex > Extracted from `beacons.pdf` --- ## Page 1 Beacons Beacons are. Cat in an environment with objects Beacons are vision-labeled objects in a simulation, Though all of them is able to be considered as a Publisher of Signal (a beacon) Most likely these will undergo a filtering process; Some of the more predictable ones (actually just terrain) Are scrubbed; (for optimization); meanwhile The most dynamic ones are usually then labeled as SocialTypes To then have their own decisionNodes for countering every Signal That they produce (the cat will then hedge/ add weight / influence decision) There are also significant mention for SelfSocial nodes (a placeholder for tracking self-identity awareness) GIST: Cat vs environment with objects Every object; or even things with a noticed Pattern ## Page 2 Is considered a Beacon; We can decide on whether to run a wholesale scan; to then Decide the Predictability % and the nearest ObjectType Of Each entity; then treat them as if They would continuously emit a signal That the Current Agent Cat need to respect (by adding weights ;or triggering hedges to simulations ) . Make beacons table Run a process to check nearest, run confirmation at modTude then, after checking, publish pulses for weights onto SIM i.e. a Social Cortex processing the HUMAN -> HUMAN OWNER. a system where we take a active affector depending on What the labeled HUMAN would do (and their affectors too; by mood for example) From the sim / thoughts involving what their mood do to their action (can be simplified; often simplified; by math / FE range ) Onto our cat’s action space ## Page 3 Formal Process: 1. Scan everything in vision for PATTERNS 2. Label all patterns and divide them by Significance 3. Significance = currently wholey decided by 4. [LEARNED]EpigeneticObjectTableX (a progression of [BASE]EpigeneticTable) After we label all patterns 5. Label each object if they are distinct by Pattern: (specific pattern that the Cat wakes up today; and decide in its immediate (medium term memory) of terrain-associated awareness; Or identity-associated awareness (any kind of stashed Identity-related-Memory) 6. Objects distinct by pattern are labeled OTSP-Named-1-% (object-type-scanned-pattern) 7. Objects distinct by type [but static] are labeled OTSA-Named-1-% (object-type-scanned-A(static object)-Name-1) 1 = is then changed to label number % = is then changed to inverted similarity % Author note: “naming conventions are subject to FREQUENT CHANGES / revamp based on associated framework , that will be attached as a naming convention update / description” (please help) 8. Objects distinct by type [but dynamic] are labeled OTSB-Named-1-% (object-type-scanned-B(dynamic object)-Name-1) 1 = is then changed to label number % = is then changed to inverted similarity % 9. Objects distinct by type [but Social] are labeled OTSSB-Named-1-% (object-type-scanned-S(social)-B(other social)-Name-1) With same other assignment; social objects are always assigned Sim Node Social objects are distincted by a special table to elevate normal objects (take the existing table ; and uplabel them to social) ## Page 4 10. Just for clarity in our important mechanism in the paper; there is a OTSSA-Named-1-% Which is a [self-social] label (object-type-scanned-S(social)-A(self social)-Name-1) This way; Identities (any psychological equivalent) are able to be represented As if they are whispering selves that adds weight or simulatable despite not being very clearly depicted Consider them >:D ghosts 11. This is especially useful when a EpigeneticChain / CYC Calls upon certain simulations (known; very frequent actually) On which outcome matrix; highly depends on a back-and-forth Stancing with the current agent. Therefore; the selves could be a stash of “recent label” attached to a Self equivalent; To then both be applied a same usual rule; on how a Social + Social would behave / call a Sim (so they can easily port a simulation that usually decides a normative outcome -> thus weight -> between two Social creature; to just swap one with a Self equivalent with a distinct label) 12. To note on the underlying framework that relies on Quickcalc mechanism (every simulation grades are subject to simplification to mathematical formula; depending on their Free Energy % threshold for novelty); this links the Social Cortex Sims to quickcalc) ## Page 5 Formal Description: In a simulation; All entities (in the mind of the “Current Agent”) is considered a Publisher. . (ideally this would be so; even Walls are publisher; or a wall that is identified having A certain pattern that the initial; low grade scanner of the agent; notices) . They publish Signals. Signals are usually weights or Hedges; or specific processing node for Hedges. Back to the fact that all entities are considered Publishers, but. . But for optimization reasons; these are then classified onto a certainty spectrum. Some of them are almost absolutely predictable , such as Walls, Rocks. etc (unless labeled with a certain recognition layer (for patterns / etc)) Some of them are less predictable; unfamiliar pattern or noticed discrepancy Some of them are dynamic; thus usually always come with a noticeable HEDGING signal Some of them are Highly Dynamic (social entities); they come with a prediction matrix and their own linkage to a node that process how their back-and-forth behavior, - (that prediction matrix node is pretty much linked to the equivalent of Empathy) (a depictable / composable simulated reactions / predictions of such entity in a separate space;) - Moving back to the term of BEACONS and the signals that they publish: In our framework; as with the processing of signals; ## Page 6 (producing hedges / weights) . These BEACONS Will then produce their specific labeled; Hedges (depending on the nearest object template that they represent) And they have their own TREE dynamic on which nodes of processing should happen (this is directly encoded in a [temporary/updated]Epigenetic table that Starts from a Baseline (Inherited) reactionary table of actions Onto a Learned (modified) reactionary table of actions (reactions in this case). . Discussion Human. Cat. Sees environment v Scans v Patterns -> types v Threshold of Uncertainty v ObjectTypes ObjectTypes are sometimes elevated onto SocialTypes Everyone of those are BEACONS with signal pulses (in biology; this is likely a chemical that even as it undergoes degradation (pulse ends) ; simply linger awhile;) . SocialCortex is always active Attaching Beacon to everything identifiable ## Page 7 (its an organ with a Processing Node that has to be filled) (an absence of patterns simply mean the Organ (biological) will eat Anything / any brainchem signal and make it a social) . (this reality actually implies that objects are initially labeled as Social) instead of the other way around as in; in a biological being; everything is first considered to be of the highest grade before they are downgraded but since we are trying to make a smushed-up quickly cooked version of a working agent with presumption of optimizing need we start this way instead! (just make a empty port to fill with patterns then upgrade) (in reality; this is a two-fold process where the Vision Cortex labels the pattern; talks to Social cortex; then talks back to the labeling central (not sure which brain part that is)) (we decide to combine Vision + Social; onto a working agent) (thus simply decide from bottom up) ## Page 8 Summary We hope to construct an agent by attaching a (any) Vision Model With the appropriate labeling regime (labeling regime is subject to change and revamp!) With the appropriate channel for processing the (labeled types) Processing comes in form of BEACONS; with signal pulse REFERENCES Prior discussion / Youtube topic: # () lightcone# () globalworkspacetheory # () activeinferenceenvironmentagents (active types / the name for the types) # () epigeneticbases# () morphic(episimilar) () epigenes # () blindness/absence of signal # # + et et --- *Extraction method: pymupdf*