This post addresses a somewhat technical but quite critical point in eurycosm theory: how does the assumption of a peaked pattern-notability distribution lead to "morphic resonance"?

And also another poing: Where does this peaked distribution come from on the first place?

I refer to

as given above...

So -- Given a universe with a bunch of patterns in it (or, to say it differently -- given a set of patterns that are considered to together comprise a "world"), we can look at the probability distribution of pattern-notabilities in this universe.

One approach would be to characterize the distribution of pattern-notabilities in a RANDOM universe, and then posit that the distribution of pattern-notabilities in the actual universe is different from that. Specifically, the hypothesis is that in the actual universe, the distribution is more concentrated on a relatively small number of patterns.

One problem with formalizing the above is that it's not so clear, in this very general setting, what comprises a "random universe."

An alternative approach is to think about random mutations to the world observed by some complex observer. Suppose we take the observation-set corresponding to a certain complex observer, and mutate it randomly a little bit. Then, if the hypothesis holds, this should generally result in an observation-set with a slightly flatter pattern-notability distribution.

What does it mean to "mutate an observation-set randomly a little bit"? Relative to an observer O, it means to replace the observation-set Obs1 with another observation-set Obs2 so that O will judge Obs1 and Obs2 to be similar. (If we want to really get relativistic we can posit a meta-observer O1 who is making inferences about O's similarity judgments about hypothetical worlds...) ....

For instance, one could form Obs2 by shuffling around the elements of the observations in Obs1 in minor but random ways. This is similar to "permutation analysis" in statistical validation.

So, OK, suppose we have an observation-set (aka world) that has peaked pattern-notability in this sense. How do we get morphic resonance type phenomena out of this?

Suppose pattern P has been observed somewhere in world W, by an observer O who has partial knowledge of W. Suppose O knows that W has a peaked notability distribution. Then the observation of P should increase the odds that O would give for P to be observed elsewhere in W.

Note that this kind of "morphic resonance" does not carry implications of causality. That is, we're not saying that (in any usual sense) the observation of P in one place CAUSES P to appear in some other place. Rather, we're saying that the observation of P in one place in a world, increases the odds that the world being observed is one where P occurs in another place.

For instance, consider two identical twins, T1 and T2. We can assume T1 and T2 have lots of common patterns in their minds. Let's call these common patterns P. Suppose some new pattern P1 arises in T1's mind. Suppose the new pattern M1 emerges from the combination of P1 and P. Then the peaked notability distribution means there will be a bias for M1 to occur elsewhere in the world. But this will imply there is a bias for some pattern P2 to occur in T2's mind, so that M1 can emerge from the combination of P2 and P. So twin telepathy, in a basic form, follows from the peaked-notability variant of morphic resonance.

Does this explain why twin telepathy occurs sometimes and not others? Not exactly. But we can grapple toward an explanation for this, perhaps. It's hypothesized above that pattern notability and attentional intensity tend to be correlated. If so then perhaps when P1 is more attentionally intense, M1 will end up being a more notable pattern. This in fact seems plausible -- a more attention-grabbing event will cause more significant patterns to emerge in a person's brain.

There is a long way from these general notions to a real theory of twin telepathy. But the direction seems plausible.

The next natural question is: Where does the peaked pattern-notability distribution come from? Why would the world have this property?

I don't have a very good answer for this at the moment, but will share some musings.

If one accepts the fundamental observer-dependence of the world, AND accepts that real observers are biased, then it seems a form of peaked notability distribution emerges naturally. But this obvious observation leads to some subtle considerations.

Most real-world observers are biased to perceive patterns that they already know, and bad at perceiving patterns that are new to them. Thus, if one is constructing a world or world-model based on the patterns perceived by some particular pattern-recognizing mind that has finite resources at its disposal, the odds seem high that this world or world-model will have a peaked notability distribution. Once it has recognized a pattern, the observer will be biased to recognizing that same pattern in other places, and will be less likely to observe other new patterns it doesn't know about (because recognizing new patterns takes more energetic/computational resources, which we are assuming to be limited).

On the other hand, if we consider this as an explanation for transparently morphic resonance ish phenomena like twin telepathy, we run into a funny conceptual problem.

We can't explain the morphic aspect of twin telepathy via bias in the twins' minds. If one twin breaks their leg and the other one gets a strong feeling something is wrong, this isn't because of a cognitive bias on the part of either twin.

But if we posit a limited-resources mind looking at a huge library of possible worlds, and choosing which ones to include in their short-list, it seems to be true that this mind is by default more likely to include worlds with peaked notability distribution -- because a mind with limited resources is going to be biased to recognize the patterns it already knows.

So the conclusion is that a peaked notability distribution could emerge from a lazy-minded God, in essence...? Or to put it less dramatically -- a finite-minded God. In this sense peaked notability distributions are highly compatible with some kind of Simulation Hypothesis.

From a world-engineering view, peaked notability distributions save computational/energetic resources (by re-using patterns over and over more often), and also provide worlds that encourage emergence of intelligence (because minds like to do induction, and these are worlds in which induction works). But this is a weak argument (at least without further supporting arguments), as there may be many other ways to create worlds that conserve computational/energetic resources.

Alternately, one might think to make an anthropic argument. Intelligences are much more likely to exist in universes that support induction. Since we are intelligent, the odds are high that we live in a universe that supports induction. But then we have to argue that the most likely way to get a universe that supports induction, is to create a universe that has a peaked notability distribution generally. So this anthropic argument doesn't work that cleanly, so far as I can tell.

However, interesting tweak on the anthropic argument would go as follows:

And also another poing: Where does this peaked distribution come from on the first place?

I refer to

**Principle 18: A characteristic of the eurycosm, or at least of large portions of the eurycosm within which humans have tended to exist, is that the distribution of pattern notability tends to be more peaked than one would expect from naïve assumptions of probabilistic independence among different entities. That is, once one observes a certain pattern P in one part of a set S that is part of the eurycosm, this surprisingly-much increases the probability of observing that pattern P in some other part of S. Further, this phenomenon seems to occur for sets S that are defined as spatiotemporal regions (though not only for such sets S). Generally, one seems to have a certain set of patterns that occur a bit more than one would expect, and the others that occur less.**as given above...

So -- Given a universe with a bunch of patterns in it (or, to say it differently -- given a set of patterns that are considered to together comprise a "world"), we can look at the probability distribution of pattern-notabilities in this universe.

One approach would be to characterize the distribution of pattern-notabilities in a RANDOM universe, and then posit that the distribution of pattern-notabilities in the actual universe is different from that. Specifically, the hypothesis is that in the actual universe, the distribution is more concentrated on a relatively small number of patterns.

One problem with formalizing the above is that it's not so clear, in this very general setting, what comprises a "random universe."

An alternative approach is to think about random mutations to the world observed by some complex observer. Suppose we take the observation-set corresponding to a certain complex observer, and mutate it randomly a little bit. Then, if the hypothesis holds, this should generally result in an observation-set with a slightly flatter pattern-notability distribution.

What does it mean to "mutate an observation-set randomly a little bit"? Relative to an observer O, it means to replace the observation-set Obs1 with another observation-set Obs2 so that O will judge Obs1 and Obs2 to be similar. (If we want to really get relativistic we can posit a meta-observer O1 who is making inferences about O's similarity judgments about hypothetical worlds...) ....

For instance, one could form Obs2 by shuffling around the elements of the observations in Obs1 in minor but random ways. This is similar to "permutation analysis" in statistical validation.

So, OK, suppose we have an observation-set (aka world) that has peaked pattern-notability in this sense. How do we get morphic resonance type phenomena out of this?

Suppose pattern P has been observed somewhere in world W, by an observer O who has partial knowledge of W. Suppose O knows that W has a peaked notability distribution. Then the observation of P should increase the odds that O would give for P to be observed elsewhere in W.

Note that this kind of "morphic resonance" does not carry implications of causality. That is, we're not saying that (in any usual sense) the observation of P in one place CAUSES P to appear in some other place. Rather, we're saying that the observation of P in one place in a world, increases the odds that the world being observed is one where P occurs in another place.

For instance, consider two identical twins, T1 and T2. We can assume T1 and T2 have lots of common patterns in their minds. Let's call these common patterns P. Suppose some new pattern P1 arises in T1's mind. Suppose the new pattern M1 emerges from the combination of P1 and P. Then the peaked notability distribution means there will be a bias for M1 to occur elsewhere in the world. But this will imply there is a bias for some pattern P2 to occur in T2's mind, so that M1 can emerge from the combination of P2 and P. So twin telepathy, in a basic form, follows from the peaked-notability variant of morphic resonance.

Does this explain why twin telepathy occurs sometimes and not others? Not exactly. But we can grapple toward an explanation for this, perhaps. It's hypothesized above that pattern notability and attentional intensity tend to be correlated. If so then perhaps when P1 is more attentionally intense, M1 will end up being a more notable pattern. This in fact seems plausible -- a more attention-grabbing event will cause more significant patterns to emerge in a person's brain.

There is a long way from these general notions to a real theory of twin telepathy. But the direction seems plausible.

**But Where Does the Peaked Distribution Come From?**The next natural question is: Where does the peaked pattern-notability distribution come from? Why would the world have this property?

I don't have a very good answer for this at the moment, but will share some musings.

If one accepts the fundamental observer-dependence of the world, AND accepts that real observers are biased, then it seems a form of peaked notability distribution emerges naturally. But this obvious observation leads to some subtle considerations.

Most real-world observers are biased to perceive patterns that they already know, and bad at perceiving patterns that are new to them. Thus, if one is constructing a world or world-model based on the patterns perceived by some particular pattern-recognizing mind that has finite resources at its disposal, the odds seem high that this world or world-model will have a peaked notability distribution. Once it has recognized a pattern, the observer will be biased to recognizing that same pattern in other places, and will be less likely to observe other new patterns it doesn't know about (because recognizing new patterns takes more energetic/computational resources, which we are assuming to be limited).

On the other hand, if we consider this as an explanation for transparently morphic resonance ish phenomena like twin telepathy, we run into a funny conceptual problem.

We can't explain the morphic aspect of twin telepathy via bias in the twins' minds. If one twin breaks their leg and the other one gets a strong feeling something is wrong, this isn't because of a cognitive bias on the part of either twin.

But if we posit a limited-resources mind looking at a huge library of possible worlds, and choosing which ones to include in their short-list, it seems to be true that this mind is by default more likely to include worlds with peaked notability distribution -- because a mind with limited resources is going to be biased to recognize the patterns it already knows.

So the conclusion is that a peaked notability distribution could emerge from a lazy-minded God, in essence...? Or to put it less dramatically -- a finite-minded God. In this sense peaked notability distributions are highly compatible with some kind of Simulation Hypothesis.

From a world-engineering view, peaked notability distributions save computational/energetic resources (by re-using patterns over and over more often), and also provide worlds that encourage emergence of intelligence (because minds like to do induction, and these are worlds in which induction works). But this is a weak argument (at least without further supporting arguments), as there may be many other ways to create worlds that conserve computational/energetic resources.

*An Anthropic Argument?*Alternately, one might think to make an anthropic argument. Intelligences are much more likely to exist in universes that support induction. Since we are intelligent, the odds are high that we live in a universe that supports induction. But then we have to argue that the most likely way to get a universe that supports induction, is to create a universe that has a peaked notability distribution generally. So this anthropic argument doesn't work that cleanly, so far as I can tell.

However, interesting tweak on the anthropic argument would go as follows:

- Peaked notability distributions encourage kindness, because they increase empathy, by increasing the degree to which various minds are similar
- Kindness encourages greater intelligence, because it allows social groups to become more cohesive, enabling collective cognition to work better
- Thus, the observation that we are intelligent and highly empathic increases (anthropically) the odds that we live in a world with the kind of peaked notability distribution that increases similarity between minds

## No comments:

## Post a Comment