模倣子 How do Immunomemes Function?

Memetic Index

This Essay is In Progress

A Meme is as a Meme Does
We have a couple of basic properties of memes, and whatever immunomenes do in the course of their functioning, they at least do these two things:
  1. Trigger other memes
  2. Change the memetic state of the cohort
This implies that there is a state that the memeplex goes into when a foreign meme tries to invade it, and this transition is mitigated by immunomemes. This means that each state that is able to defend itself against invaders contains at least one immunomeme as an enactment candidate in its state matrix. Again, the concept of what an entry in a row or column of a state matrix actually means in a practical sense is still in flux. We can think vaguely of the collection of minds in the local memetic fabric and the memes in the memetic inventory of each as a system of matrices of some dimensionality and

Memetic Drift and Mutation
We can imagine that over time the memes in a memeplex may "drift around" in a number of ways. These involve the meme drifting to another function, changing its form, leaving the memeplex (or being culled -- really the same thing), or entering the memeplex (effectively a memetic incursion), or a combination of these.

Static ΔFunction Leave Enter
No Change Same meme performs
new function
(state transition matrix change)
Meme leaves
(column deletion)
Meme invades
(column insertion)
Meme changes
Role same
Meme changes and
takes on new role
(state transition matrix change)

Here we see that  we need some kind of model for a mapping of state transitions based upon which memes are deployed within a given state. This gives us our model of memetic pathways, which the probability matrix (or whatever we decide it is and what to call it) of the memetic inventories of the members of the cohort. All the memetic inventory matrix gives us is which individuals are likely or able to deploy memes (in a given existing state) that will transition the cohort to another state. In practice this may be impossible to characterize a priori, except to get a (rough) idea of the memetic inventories of individual cohort members. This provides a description of which states may actually be transitioned to, so if it's possible to characterize possible states, which probably depends upon whether there are transition regions where the instantaneous contiguous states are, to borrow from calculus, "continous" or "non-chaotic".

A Challenge for Quantitative Memetics
Here is an important point for the development of quantitative macromemetics. If in order to perform meaningful computations we must characterize things like transition matrices where the data are effectively impossible to measure (or construct statistically from large samples, for example), then we are effectively stuck in the same situation as economists or sociologists, for example. That is, we are dependent upon untenable assumptions such as "perfect information", "supply-demand curves", and "individual beliefs".  We want measurable quantities, perhaps gathered as a census might be, such as presence/absence of a meme or a connection or what-have-you, or measurable thresholds for meme deployment (based on such as physiological measurements and preferably as properties of memes themselves rather than individuals), deterministic and falsifiable characterizations of transition matrices, and so forth, and as needed units of measurement for same. We should be able to achieve arbitrary precision with increased efforts at measurement. Our development of theory should head in this direction.

State Transition Matrix
A transition matrix should resemble something like the following. A state is a collection of memes (and cohort members, which for the sake of simplicity remains static) forming the columns, and the cohort members forming the rows. A state transition matrix has states as columns and memes that cause such a transition and the new status transitioned to. In principle, each transition point is a 100% certainty, i.e., if a given meme if deployed in a given state, the next state of the cohort is specified with total certainty.

S1 S2 S3
m1 t1,1 t2,1 t3,1
m2 t1,2 t2,2 t3,2
m3 t1,3 t2,3 t3,3

We have a few problems:
An immunomeme represents a state change, but to where?
Where does the cohort transition to from the immunomemetic state?
How does the immunomeme know to fire (given the novelty of the input)?

According to this model, it looks like immunized memeplexes change to a state specified by the given immunomeme when an incursion into the system is detected, i.e., a "foreign" meme which would "threaten" its stability.

Dealing with Novelty
 One could imagine, of course, that the memeplex goes into a totally different submemeplex in the case of certain insults, and this might look like going to war, going on "witchhunts", widespread panic, or as economists might enjoy, frantic buying or selling on the market, decrease in spending, or, as Greenspan wrestled with, a propensity to remain in the same employment and put up with lower wage growth. Do these memetic states "exist already" somewhere, are do they appear spontaneously in response to an invader or novel situations. Is there even such a thing as a "novel situation" (philosophical point)? Perhaps something truly novel will make a society "collapse" rather like the Roman Empire or any number of other civilizations throughout history (or any number of companies). It might be interesting to look at the Fall of the Roman Empire, or the French and American Revolutions, or the Japanese Meiji Restorations for example of how "novel" memetic structures appeared in a very short time and where these new memetic structures came from (spontaneously congealed, leaned on history, etc.).

It could also be that the action of an immunomeme could be something along the lines of inserting new rows into the state transition and cohort inventory matricies in response to a fairly "trivial" insult. This might preserve the ability of the system to "remember" the old state and continue to be able to transition to other known states as before. The question is, simply put, is what might the destination state matrix and cohort inventory matrix look like. It seems likely that there are multiple answers and there may be more than one right one, and this may depend partly on the type of invasion the system is trying to repel. Omniphagic immunomemes may be characterized by a common state, no matter the original state when the insult occurred, and the resultant state may look very similar to the original, or it may even be a "loop" or circuit directly back to the original state.

Do memetic systems have "memory" or "counters" such that they can keep track of how many times a stimulus has occurred or how long the state has persisted. Perhaps an interesting question, but probably only a shorthand for a large number of possible states.

Another possibility may be for "parametric immunomemes" to create a new state based upon the parameters of the attacker, or rather, the content of the attacker might be used to construct a set of immunomemetic state and inventory matrices based on those contents. Depending upon the complexity, this presumably would take time, and would no doubt depend upon the collection of memes, MIAOs, and even existing states of the memeplex at the time, but of course there are plenty of examples of memeplexes borrowing from one another (religion, Catherine the Great and Voltaire, Peter the Great and Western Europe, the Mongols and the Han Chinese, the Japanese generally, the Nordic peoples and Christianity, the Americans and Western Europe (Germany and England initially), and the list goes on and on, and of course we wouldn't need copyright and other intellectual property laws if this weren't true). But again, all this takes time, sometimes centuries.

Oligarchies, Star Chambers, and Templates
One possible explanation for the immunomemetic response of a memeplex is that by some action a "template" of a state matrix and an inventory matrix is somehow generated as the state to which the immunomene "sends" the system upon detection and reaction to an invading meme. A "template" and the limits of allowed changes which may be made to said template constitute, then, the flexibility of the associated immunomeme(s) to adapt to and absorb attempts at invasion.

It might look like, or at least have the same effect as, an oligarchy of power-holders who in practice can only make certain decisions within a limited set of possibilities. Or rather, the system automatically responds to normal events, and may even be able to deal with simple invasions, but when something requiring a more complex response by the system occurs, a new matrix must be developed, but only within the scope of available resources and possible configurations for the system. It may be possible to set up a matrix that calls a totally new set of memes, but it is in principle not possible to change what those memes call once the immunomatrix is exited by the next incoming meme.

To say that there is some kind of template that serves as the basis for producing the state that an immunomeme jumps to in response to an invader doesn't really add much more than just saying that an immunomeme causes a state transition, which is fairly prosaic, since all memes do that. However, the difference is that immunomemes are reacting to "novel" memes(1). This does not really represent a problem in terms of the destination state and its attendant matrices, however. These matrices may exist already (and probably do in the case of omniphagic immunomemes).

One thing we can say, in terms of the invocation of an immunomeme, is that whatever the current state of the memeplex is, the incoming alien meme will not be in the matrix of state transitions to other states (matrices), nor will it be, effectively, in the instantaneous inventory matrix of the cohort. That is perhaps the difference between an immunomeme and all of the other memes in the instantaneous state matrices, that is, that the alien meme is not in the matrix and that in order to determine what fires next, the alien must be recognized by one (or more) immunomemes.

Can we say that whichever immunomeme recognizes the alien is the one to fire? Perhaps. We have not yet put forward a theory for how the person and deployed meme combination is selected for each transition. It all depends upon perceived reward opportunity and then some as-yet-vague idea of who gets to "go first", which may have to do with some kind of as-yet-unclarified notion of "status". For now we can imagine that most every state has at least one immunomene and when an incoming alien meme appears, deployment of one of those immunomemes presents a reward opportunity, and somebody has some chance of taking it, while if the alien succeeds in "masquerading" as another meme, then the cohort will jump to the state pointed to by that memetic transition. The invasion is then effectively successful, since the new meme has managed to give somebody a reward in the course of making a known transition within the current system, and therefor it will be known that future invocations of the novel meme by that same person or another may produce a similar reward.

Resonance, response of other cohort members, which produces the reward, is also what causes the memeplex to go into another state, i.e., all cohort members update their memetic inventories and their deployment "probabilities"(2). In the face of a novel memetic input, this process may become unstable, produce unknown results.

In the case of an alien meme that is "too alien" to produce any response, even a "negative" one, we have something which could be termed "alienation". The invader must trigger some existing pathway, or somehow insert itself as a new row in the state transition matrix, i.e., as a new transition option, and as a column in the inventory matrix, i.e., as a meme that at least one person may now deploy.  By the way, given these two matrices and a vector of the cohort members, the multiplication of all three of these may produce the final output state (or some value to be applied against a threshold or suchlike).

Memes must be able to insert themselves into state matrices based on whether they seem likely to produce a reward. This may become a circular argument and frustrate quanitification. If a meme tries to put itself on the menu and is ignored (has zero firing probability), it looks like alienation. However, if it insinuates itself with a non-zero weight, then it's in the system. For survival, it doesn't matter where it transitions to, but for it to change the system, it does. It must cause transition to a different state,  i.e., expand the graph branches, or create a transition to an altogether new state, i.e., expand the state inventory. The fun thing here is that the new transition state could be pre-existing and completely outside the memeplex under attack. Of course, cohort members would have to be aware of the memes making up that new, external memeplex. This comes back to the idea of a template. What sorts of new state matrices are able to be formed based on the overall inventory of the cohort, and the pathway structure of the existing memeplex? Assuming matrix insertion is possible, it should be possible to insinuate a new meme into the matrices such that it does exactly the same thing as some existing meme in the same state matrices, effectively stealing some of its function. What we want is that the memeplex make different transitions, do things with different order and/or frequency, and also do completely different things or stop doing old ones.

Add diagrams of memetic trasnsitions
Add matrices and vectors showing novel meme insinuation, and possibly novel states

(1) There is also the possibility that an immunomeme need to act against environmental, non-memetic incursion, e.g., global warming, etc. Whether this is the purview of (immuno)memetic theory and how that might work are topics for future research. Of course, there are plenty of immunomemetic responses to the idea (MIAO) of global warming, e.g., on the American political scene, but actually dealing with the problem, or with its fallout, may be another matter.

(2) What this means is as yet unclear. There is a matrix of all cohort members and all of the memes available to each, and the intersection may represent the chance that that person deploy the meme at that juncture, but it may not work that way (it might be one everywhere).

No comments:

Post a Comment