(Going to need to piecemeal this together, so bear with me it’s in progress.)
This is a look at high level processing, somewhat closer to what most of CogSci/Neuroscience is focused on. Underlying this high level processing are the core conceits which assume that:
a) All functions of an organism are a specialization/complexification of features which exist at the cellular level.
b) “Emergent” properties are not unique properties of any particular intercellular interactions.
c) Cells do not “know” the state of other cells, they must be transmitted (or “assumed”).
d) All inter-cellular communication is metabolic and requires metabolic products.
(There’s probably a few I’m not listing, will try to list them as they occur).
e) All “life” requires “memory” in order to express “behavior”. That memory does not require adaptivity, however all “learned” behavior requires adaptive “memory”.
We are also considering: That “memory” and “behavior” are equivalents, with “behavior” being an “expressed” form of “memory”.
All “behavior” is a stimuli response. All vertebrates have at least two discrete stimuli response units, which operate interdependently of each other.
For the scope of this model, we are limiting our ethological scope to humans particularly, however most of this model should apply across at least all primates, very likely all vertebrates, and possibly all “complex” multicellular organisms.
Expressed behavior at the top level is the an artifact of discrete functional units. These functional units are flexible, and other than core response units can be added or removed.
These functional units are largely independent of other functional units, and process all information in a specialized way.
These functional units can be stacked and sorted in various ways to produce discrete behavioral output.
We should be able to modify and predict high level behavioral output by understanding what the discrete function of each of these nodes are, and measuring their output relative to other modules.
Each functional module “learns” or stores information independently of other modules (there are no “universal” engrams).
The core modules work as master schedulers, which flexibly allocate the order and “strength” which other modules are activated in a behavior response.
The core modules do not store stimuli information themselves, instead they work as a sort of lookup table of existing functions to “translate” stimuli into behavior.
“Target behavior/goals” are discrete functional modules. Humans have at least four discrete goal modules.
These goal modules feedback to the core modules to update the “default routing” based on changes in stimuli state.
Each functional module can run asynchronously, and it’s contribution is calculated in “on the fly”.
All functional modules provide “strength” signal, this strength signal (valence) modifies expressed behavior.
The order in which the modules are scheduled modifies the total contribution of the module to behavior.
Every individual has a unique lookup table, based on construction and “memory” within each functional module.
Visualizing This: I find it useful to borrow CogSci “matrix” conceits like this, but instead imagine each of those modules/spheres is a discrete functional module. These functional modules extend all the way down the spine, and all discrete behavioral feature requires a sphere to support it. Each finger and toe requires a module for instance, each type of “cognitive” function requires one.
In this conceit, neurons work exclusively to transmit stimuli information. Since all “behavior” is initiated by stimuli, they are the starting points of behavior. All neurons are fed into function modules, which themselves contain a map of stimuli and responses (nuclei). Each nuclei has a “mini” lookup table (astrocytes), which modifies behavioral responses specific to the nuclei’s function, based on prior input (or genetic biases/default construction) from our “goal” modules.
The output of these functional modules is expressed primarily in “strength”. Higher “strength” metabolic responses (strength in quotes here because sometimes no response at all is a “high strength” response) and are fed into comparator modules (e.g. putamen/accumbens) where differentials between current state and “goal” states are generated.
Macro state computation occurs in the deep cerebellar nuclei on the “incoming” side of the loop, and dentate gyrus on the “outgoing” side of the loop. The incoming/outgoing differentials are “subtracted” into “awareness”, which serves to re-weight the “incoming” side of the process. Essentially, human brains are “aware” or “conscious” because it’s an attempt to additively tune the “incoming” side of functional modules.
This gets a little tricky because our “goal” modules have an “incoming” and “outgoing” side as well (let’s call these “thoughts” for the sake of visualization) which are independent of the “direct” stimuli receivers. For example, we can “feel” “pain”, however our goal modules can re-weight the high level look up tables ordering so that it pushes that “pain” further down in the stack (or weight goal state higher).
Further, the current “functional network” conceits do not describe the functions of the functional modules, at all. Functional models in nervous systems are not tied to behavior specifically, they are tied to a specific type of calculation which is tied to stimuli directly, rather than behavior. For example there is no “speech” module in a nervous system (e.g. Broca’s), speech is a stimuli reaction produced by an interaction between several modules. The modules themselves are way more generalized than currently understood, and any module can contribute memory/behavior for any particular task.
Application: This is already the basis of psychosurgery/neuro-stimultion techniques, attempting to modify the weighting of functional modules in particular areas of the brain. What we are missing is an understanding of what each of those functional modules actually contributes (we’re stuck in a lot of bad assumptions here).
For example, we can mess around in the nucleus accumbens and either increase or decrease expression of “addictive” behavior, but from a processing standpoint what we are doing is tweaking the weights of inputs into a comparator. When we get better at this, we will be able to follow those inputs upstream and tweak it there.
So how do “therapy” models work under this conceit? Again, there’s two things to consider, order of integration and strength of signal when generating output behavior.
When goal/target is not achieved, it attempts to continue weighting up it’s “incoming” side weighting (anxiety) until a metabolic breakpoint is reached and the brain as a whole starts attempting to shut output in goal modules which are contributing to lack of goal state achievement (depression).
Sometimes, goal state is actually being achieved, but the order in the higher order lookup tables was modified down previously to account for temporary external/stimuli side conditions. What a therapist can do is help walk through the process of re-weighting those inputs against current experience, or train you how to “artificially” re-weight certain inputs (just think about it this way…).
For some people, a large amount of distress is caused due to their default “ordering” being significantly different than the individuals around them, which results in significant differences in behavior or thought. Programs like CBT can train in a default ordering closer to external/social expectation.
When we talk about “psychiatric conditions”, usually what’s being referenced is “disorder”, or processing that diverges greatly from “expectation”/goal. And the goal of the field is to “re-order” (or re-weight the inputs, but mostly re-order) the stack to achieve an output closer to “goal”.
Another common mechanic is simply “self-acceptance”, which is “programming” a differential into your goal state which “down-regulates” the intensity of signal generated by mismatch of social module functions.
Behavioral Examples: Something discussed a lot recently is the effect of social media, especially on younger individuals (it’s not at all limited to them though). Social media represents a really interesting effect in that it the modules which we use to process the stimuli are a lot more spread out than most social stimuli. The more functional modules stimuli touch, the more weight that stimuli ends up having in our state comparisons.
A unique/insidious aspect of social media is that it “trains” us to push these inputs ever higher in the stack, giving them more weight, and ultimately creating a feedback loop that depends on the feedback itself to properly weight our processing stack. It’s pretty dramatic to see the differences in anxiety for people who break the feedback loop by either discontinuing it altogether or “consciously” pushing it down the stack.
A lot of video games have this mechanic as well, but even worse since the stimuli can touch even more functional modules.
“Drug Use”. Why do people “use drugs” which have a high negative risk associated with them (e.g. illegal, “unhealthy”)? It’s an attempt to re-weight/re-power the stack. Smoking weed for example may lower the “strength” on particular inputs (e.g. physical pain being transmitted from functional modules in the spine), and taking hallucinogens may temporarily “decouple” the stack ordering altogether. Once deconstructed, it’s easier to reconstruct the stack (or for some people, they get a stack which has even “worse” output).
Why does a hug have such a dramatic impact on cognitive processes? Same process here, we are adding stimulation, which is getting assigned a positive valence in the evaluation centers (probably the Olives?), and that weight modifies all downstream processes.
How about something like post partum depression? In this case the model would explain this as receiving a constant stream of positively valenced input from the body, which suddenly cuts out before the rest of the nervous system has a chance to re-assert homeostasis. We go from “balanced/slightly high” to “NOTHING“, which is exactly the same as a traumatic life event. (I wonder if there’s a correlation between pitocin use and PPD?)