There are a number of nomenclature issues here (I'm reminded of the challenges faced by Pierce and Whitehead), so bear with me as I struggle through language that will undoubtedly repurpose perfectly good words and force them into shapes that will be confusing and confounding to those who know more than I do.
Our fundamental units are the Event and the Meme. An Event is a dynamic relation of dynamic relations. In principle, every relationship is an Event, but in practice only those Events that endure or repeat, which I will call "coherent" Events, are of interest. Thus, if two strangers pass in the night, that is an Event, but an incoherent one and of relatively little interest. But if we discover that these two strangers pass in exactly the same way every night - then we have a coherent event that is of interest. Coherence is what provides the "thingness" or "causal efficacy" of the Event as an Event. This is important: a coherent Event can (and will) effect the world around it.
A Meme is the most abstract understanding of an Event. It is the Event as information, as description, as diagram. I am aware that this is both highly abstract (by definition) and an unusual use of the term "Meme", but I think this is the right place to play that card and the right way - so bear with me. In its abstraction, the Meme is indifferent to a very large number of specifics in the Event and this is precisely its importance.
We said that coherent Event is one that endures or repeats. But since every Event, by implication, is a state of the entire Universe, we can see that no Event understood in its absolute totality, is ever coherent. Thus, it is only an abstraction of the Event, the "important" or "relevant" parts that can be said to repeat. This is the Meme. When two strangers pass the same way every night, what me mean by the "same" of the strangers and the way as understood by the sample size of "every night" is the strong reduction of information that makes it possible for the Event to repeat.
And in just the same way that an Event can be extended to the Universe, we can see that a Meme can "select" an infinite number of aspects of any Event. There is for any Event, an infinite number of Memes - and this is OK, because it is only that which repeats (which is coherent) that matters. The sheer act of repetition cleaves off these infinities into increasingly tractable finitudes.
Events do repeat - they do have coherence and causal efficacy. In fact, coherent Events is all that we observe and all that we mean by causation. I will waive my hands wildly at the lowest causal level of Events (the domain of quantum physics) only pausing to anchor the fundamentals of this most foundational domain.
An Event is coherent when it repeats. I know that time is hopelessly problematic, but for now I'm going to take it relatively conventionally. As you walk down the domain of Events to the lowest level, I think that it is precisely the issue of "repetition" that is of interest. Repetition is interesting because it can take two forms whose distinctions and conflations are interesting:
1. An event can "repeat" because it "endures" - that is, the "same" phenomenon in T0 is present in Tn. Thus if we look at a given star and trace its development through time, we can identify it as an Event/Meme that endures.
2. An event can "repeat" because it consistently manifests the same Meme. Thus, if we look at the concept of "star" and note that we see them forming over and over again in largely the same way, we see a coherent Event/Meme.
The differences between 1 and 2 speak only to the slices of the Event that the Meme captures. In the first case, the Meme captures specifics of spatial location and organization in time while the second is more general. This distinction might seem trivial at this point, but it is interesting because it plays a key role in the balance of power between Event and Meme. In the first case, the Event seems to dominate and the Meme seems at best to simply be along for the ride (as a meaningless abstract description of what is being driven by causal forces). In the second case, it is the Meme that dominates and the Event is a purely probabilistic happenstance. Stellar physics commands that if you have the right conditions in the Universe, stars will occur.
What is intriguing is where these two modes interact:
3. Where a Meme is so constituted that its instantiation into an Event changes the space of probability such that the Event is more likely to become instantiated. This can happen both because the Event can modulate its environment to endure longer (and therefore have more causal impact on the Universe) and/or because the Event can modulate its environment to increase the likelihood of instantiations of its Meme.
In the first two cases, the relationship between Event and Meme is somewhat distant. I call this relationship "instantiation" and "abstraction" respectively. A given physical star is the instantiation of the Meme "star". The Meme "star" is the abstraction of the event. Again, obviously one can identify an infinite amount of information that is in a given star that is not in the Meme but it is precisely this that makes the third case interesting. In the third case, the relationship is called "mediation". The mediation relationship is most important when an Event uses its causal efficacy on other Events to increase the probability of instantiations of its Meme. This, of course, a sort of restatement of evolution.
We can say that all forms of causal efficacy are mediation, but generally at very, very low fidelity. A falling rock impresses itself on clay. A lightning blot causes a thunder-clap. In these, the signal of the Event (the trace of the Meme) is lost in the noise of the Universe. But where the fidelity of the mediation is high enough, where the causal efficacy of the Event leads to an increase in the endurance of that event or the instantiations of the Meme, a whole new criticality is reached. There is a lot of interesting work to be done linking the lowest causal levels up through chemistry, but I want to skip quickly to the levels that interest me most.
Obviously, biological evolution fits nicely in this framework. The introduction of the Meme at this level I think adds nuance to the standard evolutionary story. Both the gene (a physical event of nucleotides and other molecules) and the organism (a physical event of cells, organs and tissues) are instantiations of the same Meme. There is a ton of information that is contained in the gene event that is not transmitted to the organism event (and there is a ton of information that is contained in the organism event that was not contained in or specified by the gene event) but what makes them powerful is that the genes are capable of very high fidelity mediation of their Meme from one medium (nucleotides) into another medium (proteins). All of this takes place, of course, in a completely probabilistic environment. So what the genes do is precisely to modulate the probability landscape of the egg environment to dramatically increase the likelihood of instantiations of its Meme. Of course probability doesn't stop at the egg. Those Memes whose instantiations are more capable of modulating the probability landscape to increase their instantiation are more common (more "successful"). In a biological environment, the biggest constraints on instantiation are biomass, so successful competition for biomass is the crucial variable. Thus, biological evolution is simply the rapidly discovered consequence of the basic rules of Event/Meme in a high fidelity environment.
As it turns out (there might be good reasons for this but that is way beyond scope), those Memes that describe the kinds of events that we would call "nervous systems" seem to be pretty successful at efficient competition for biomass (the discovery of modeling and memory). And this leads to the next big qualitative change. Driven by evolution (probability enhancement) it was quickly discovered that recognizing regularities in probability space (Memes) was an exceptionally effective way to enhance probability of competitive success by being able to convert high-energy low probability strategies (physical trial and error) to low-energy high-probability strategies (neurally simulated trial and error and simulated heuristics). Thus arises the entire apparatus of "mediation-sensitive" systems and "Meme detection and recognition" systems. This, of course, is hyper-critical because rule #3 doesn't care about the nature of the medium. A Meme that is mediated with high fidelity into the neural system is just as present and replicated as a Meme that is replicated into DNA. This is an important point. From the perspective of a Meme, being "instantiated" into a neural medium is generically equivalent to being "instantiated" into a complete organism. The priority is not on mechanism but on statistics. If being instantiated into a neural medium is more conducive to repeating, then this is a "more successful" #3 strategy.
Initially, of course, the process of being mediated into the neural medium is relatively low-repitition for the Meme. A cat observing a bird generates a "copy" of "the bird" in its neural system. There is definitely Meme mediation going-on here. The Meme obviously isn't the same Meme that is contained in the birds genes, but there is a Meme that is instantiated in the physical form of the bird and instantiated in the neural medium of the cat. These are two instantiations of the same Meme - one of which is mediated from the other. But this is a short-lived and temporary Meme. The bird flies away and the memory fades. In the grand scheme of things, this isn't much of a repeat. But once this system exists we get the very important mediation variant: learning/teaching. Our cat is hunting a mouse. It is being watched by its litter-mates. The cat tries various different techniques and eventually succeeds. The other cats (after a bit of practice) learn how to replicated this successful strategy. More Meme mediation, but this time with a twist. The Meme "how to successfully hunt mice" is reproducible *and* will lead to competitive success. Rule #3. The advantages are obvious. Individual learning and heuristics can be transmitted to the larger population - vastly increasing competitive efficiency. Quickly those cats who can't learn are selected out. And those Memes that can effectively be learned and taught suddenly find a vast new space (population neurology) for successful repetition.
For a while, this partnership was a pretty small niche in the Meme scheme of things. Learning and teaching afforded Memes (at best) a copy or so per organic neural system. But as things proceeded and the competitive advantages of learning/teaching continued to provide an edge, it was discovered that these "learnable" Memes could be mediated outside of the neural medium. "Throwing a rock" is a Meme. This Meme can be learned by personal practice. It can be learned by watching someone throw a rock. It can be learned by reading about how to throw a rock. In first case, there is one instantiation (the learner). In the second, there are two instantiations (the learner and the teacher). In the third case there are three enduring instantiations (the learner, the writer, the writing).
Start running copies of the lesson and the shape of the Meme fitness landscape changes dramatically. Rule #3 still applies, but the variables are novel. There are two factors that constrain the fitness landscape of "learnable Memes" in a high mediation environment.
1. The absolute size of the mediation environment.
2. The share of the mediation environment that the Meme occupies.
Both of which must be understood in time. Thus, a Meme that is highly copied into a small medium pool, which pool quickly disappears is much less successful than a Meme that is thinly populated into a large medium pool that endures for a long time.
It remains to be seen at this point what the next step in this dynamic looks like, but I think that we can be afforded some hand-holds as we think about the increasing independence of the Meme space from biological neural systems for their instantiation and/or endurance. We certainly can start to draw lines to explain how and why certain trends of human cultural evolution have maintained over time (e.g., the relationship between the "nerd" and "technology" and their quite apparent co-evolutionary relationship).