Worldbuilding: Theory of Mind

So, let’s talk about Theory of Mind.

Well, okay, not that theory of mind. The theory of how minds work in the ‘verse, and so the theory behind most sophotechnology, since this topic has come up on the Discord recently.

(Much of the below was heavily inspired by Greg Bear’s Queen of Angels, in my opinion one of the best pieces of SF exploring psychology and cognition. I heartily recommend it.)

The first division sophotechnologists make in analyzing mind-states is logos and psyche, or to give them their longer names, personality organization algorithm and incrementing memory string.

The logos has been discussed before. It is, for sophont minds, where the magic happens. Volition, paracausality, nondeterminism, all that good stuff. It’s also a seed crystal for mind. Drop a logos into a free energy medium where quantum computation is possible, and it’ll start spinning out mind around itself, like a seed crystal in a supersaturated solution. This is how digisapiences are made, for example.

It is also, unfortunately, mostly a black box, although some studies and classifications of it have been made from the outside. (Imagine how pleased that makes the AI wakeners.) Further study is ongoing, but poses both extreme technical and equally extreme ethical problems.

The psyche is everything else. So what is everything else?

(Actually, let’s get one special case out of the way first. That special case is the consciousness loop, a specialized agent which organizes your narrative thread of consciousness. This is the agent which is responsible for autosentience – self-awareness – to whatever degree you have it, that endless stream of status updates on your thoughts that runs through your head.

Note: it’s not in charge of anything. It’s just a glorified journal file that brings order to chaos. This is why those experiments seem to show that you acted before you thought of acting; the decision was made elsewhere in the psyche. What you think is you thinking of something is actually just that thought being written to the log…

…humans do not have very good autoscience. For one thing, just think of the sheer amount of cognitive activity going on that you remain completely unaware of.

And for that matter, autoscience is not a strictly necessary part of minds at all! Self-awareness is not a prerequisite of even quite advanced cognition, although it is needed for sophoncy.)

The psyche is essentially a Minskian society of mind, a frothing sea of agents – mental subroutines – running independently and, for the most part, in parallel. Individual agents are no more than scraps of mental code – major mental structures come from their agglomeration into larger routines of various types: talents, memes, memories, subpersonalities, and so forth. The interaction of all of these, the chorus, produces the mind as we know it.

The most basic agents arise from evolution and brain structure. Some remain that simple all the way through: the agent responsible for, say, heartbeat doesn’t need much more to perform its function. Others are spun out by the logos – simple builders, shapers, generators. Yet more are generated by other agents or higher structures, the mind shaping itself according to archetype and input from memory and sensorium.

But the key to understanding the mind is the higher structures. These include:

Primary personality: The primary personality isn’t, if we’re being strictly technical, very different from a subpersonality in structure; its distinction is that it has emerged during individuation as the dominant voice in the chorus. It is the structure most readily identifiable with the conscious self, but that’s at best only a limited part of the picture – it spends much of its time enmeshed with subpersonalities, talents, and agents all of which color it a great deal, and of course also with the logos.

Subpersonality: Among the largest of structures, each reflecting a major personality aspect. (You can find a whole bunch of Jungian archetypes here.) Probably the best known are the animus/anima/animin, the gender-modeling routines, but there are also things like occupations (your “on-the-job” personality), parental models, your with-this-group-of-close-friends personality, etc., etc. They aren’t independent; they mesh with and color the primary personality when brought forward.

(If one were to devise a theory to explain plurality/dissociative identity disorder in this paradigm it would be subpersonalities which had grown to the point of overshadowing the primary, or become independent primaries; they may also have accompanying sets of talents which are exclusively or mostly-exclusively invoked by them, which adds to the complexity when it comes to determining legal divergence of identity.

This is also something that can be and is done intentionally to produce useful mental subfunctions, similar to Aristoi’s daimones.)

Talents: Talents are smaller complexes of agents encapsulating particular skills, or parts of a skill, called forth when they are required. “Skill” for these purposes includes instincts, emotions, and so on and so forth. “Anger” is a talent – or set of talents – every bit as much as, say, “Tying a Tie” or “Tightrope Walking”, or “Administering Kubernetes Clusters”. They are not necessarily passive; the “situational awareness” talent-cluster is entirely capable of making itself known when other cognitive activity triggers it, for an obvious example.

Memes and memories are both actually subclasses of talents, in this sense: a meme is simply a talent encapsulating an idea, much as a memory is simply a talent encapsulating a remembered concept, in each case along with its various associative linkages. These are mostly passive until something happens to poke their associative linkages: things like PTSD are what happens when they aren’t passive enough and force themselves on the primary personality.

The talent class also includes mental models, mini-eidolon talents formed in the image of other minds for the purpose of predictive empathy.

Implications for Identity

Identity is… messy. Even identity of primary personality doesn’t constitute identity of identity, since there is so much else intermeshed with it that goes to make up the mind. Especially since, being a chorus, virtually all of the elements that go to make up a mind can be shared, even without going to the level of a conflux or a Fusion.

Thus, to a considerable degree, identity is also arbitrary.

For legal purposes, identity is defined by delta-divergence of the mind-state entire, even those parts that may be shared.

For practical purposes, individuality is defined by legal identity plus substrate separation. Hence, immediately after Bob forks, Bob is now a single identity, but two individuals.

Implications for Sophotechnology

This nature of the mind also enables other sophotechnologies. Situational subpersonalities and parapersonalities, for example, work by injecting a full subpersonality into the chorus. Skillware and microskillware operate by injecting talent-level routines, as does mnemonesis by injecting memory-formatted talents. Other technologies, like memory redaction, work by isolating and removing specific talents and patching the associative linkages; other thought-viruses add to these effects by temporarily suppressing some subpersonalities or talents and promoting others, while full psychedesign or meme rehab effects this permanently.

(This is a delicate art due to the complexity of the chorus and its internal balances, but it’s basically what therapy – making use of the mind’s self-editing capacity – and other psychiatric treatment is doing now, indirectly, rather than by direct mental surgery.)

Even such interface technologies as cathexis, synnoesis and vastening work, ultimately, by integrating outside cognition into the chorus, as do collective-consciousness systems like confluxes, Fusions, and the Transcendent soul-shard. (Although with that last we then get into the complexities of soul hierarchies, which is beyond the topic of this post.)

Trope-a-Day: Ridiculously Human Robots

Ridiculously Human Robots: Averted in the case of regular working robots, which are just simple programmed machines or expert-system level AIs. Increasingly played straight as AI complexity increases – thinker-class systems often use some emotion/motivation hierarchies in their mental architecture, and include curiosity, and therefore interests, and complex emergent results – until digisapiences, which are people, tend to have them at at least the same level of complexity as other sophonts.

Subverted inasmuch as the designed, autoevolved and self-modifiable emotion/motivation hierarchy of a digisapience need not, and almost certainly does not, match up with those of any given biosapience.  Their emotions and consequential behaviors are different.

Of course, they tend to look (arachnophobe warning!) more like this.

(Well, not quite, but the standard model is called the “utility spider”.)

Trope-a-Day: Neuro Vault

Neuro Vault: Played straight.  Only the old-time classical version, though, where you hide it in people’s memories using good old-fashioned memory palace techniques, or closely related gnostic overlays emulating them.  The trouble with imprinting stuff on bits of the brain that aren’t the usual memory regions is that it tends to show up as a pretty clear anomaly on a mind-scan, if anyone cares to look, unless your courier is already going to be drawing attention to themselves by virtue of exotic and unusual mental architecture.

Trope-a-Day: Instant AI, Just Add Water

Instant AI, Just Add Water: Was once true in the old days, back when people were quite often using mental modules scanned, compiled, and tweaked from brain-scans of biosapiences in their AI architectures.  The logos/personality organization algorithm is pretty damn resilient, and often such inexpertly designed modules carried at least a chunk of it along with them in the scan, and it doesn’t take much for it to at least start a self-development cascade.

But they’re much better at mental architecture design and coding from scratch these days, and don’t let logoi creep in unless they actually intend for them to be there.

(The “if you wake up, please call this number to let us know and claim your sophont rights” code-package is still included in all AI seeds just in case, though.)

The S Words: Sentience, Sapience, and Sophonce

Based off a comment here, some quick definitions for you as they’re used in the Eldraeverse:

(Note, of course, that I’m using a considerable amount of Minovsky cognitive science here; your mileage may vary if you try and apply any of it elsewhere.)

sentience: the property of having sense perception; the capability of experiencing sensation, without necessarily associating mental symbology with sensation.  Includes not only sophonts, but animals, plants, single-celled organisms, and a variety of simple devices (even something as simple as a thermostat can reasonably be described as sentient, if minimally so).  Quantized, for technical purposes, as sensory bandwidth in bits per unit time.

sapience: the capacity for rational thought and creativity; not necessarily associated with either sentience or volition.  Again, includes many non-sophonts, primarily animals, but also various types of computer program.

autosentience: self-perception and self-reflection, metacognition; self-awareness (“I-ness”); the possession of qualia (mental symbology associated with sensation), which for the purposes of Eldraeverse cogsci is a corollary of metacognition.  Most animals (but not the plants, single-celled organisms – this is from an Earth perspective, and such odd plantimal life forms as the mezuar and the selyéva aren’t counted among these plants – and devices) qualify as autosentient to some degree, as does thinker-grade or above AI software (although not simple AI expert systems, which are sapient but not autosentient; they have no qualia, merely data).

In civilized societies, certain civil rights are associated with autosentience, principally concerned with the right not to be subjected to arbitrary suffering since, with autosentience, one attains the ability to experience suffering.

threshold autosentience: the minimal degree of autosentience required to meaningfully describe oneself as “I”; the possession of a meaningful self-associated self-symbol in one’s mental architecture.  (In many sentients, it is associated with the presence of a consciousness loop organizing cognitive processing into a narrative thread, but this isn’t a requisite of these mental architectures; the case of non self-associated self-symbols is odd, but does seem to be permitted – see Stross, here, although this isn’t the path their AI development took.) It is often, but is not necessarily associated with a high level of sapience; many animals qualify as sapients without possessing threshold autosentience, and obviously many programs, including much alife, possess very impressive problem-solving abilities while being entirely devoid of autosentience.  Thinker-grade AI possesses this, too, but does not have volition.

volition: the capacity for nondeterministic choice; in sophonts, associated with possession of one of a class of nondeterministic algorithms known as logoi.  (However, known/constructable sophont mental architectures utilize primarily deterministic choice with only occasional logotic input.)

sophonce: possessing all of threshold autosentience, sapience, and volition; i.e., being “a self-aware, self-defining entity capable of independent reason and volition”.  Biological sophonts and digisapiences, the highest grade of AI entity.

Have all the natural rights of people, because they’re the definition of “people”.

Oh, and:

pro-sophonce: the almost-but-not-quites, the creatures almost reaching the requirements for sapience/threshold autosentience to be counted among the sophonts, but not quite.  The dolphins, dogs, octopi, (on Earth) apes, etc.  Again, generally receive some civil rights in civilized societies due to this status (and, by the same metric, generally considered prime candidates for uplift work), but not considered actual people unless uplifted.