7.09.2012

Meta Evolution: fitness beyond reproduction


In the most general sense, evolution is any complex process that has some definable direction. There are three such trends on the highest ontological level: entropic equalization in physics, improving quality of reproduction in biology, and increasing predictive power in cognitive (psycho-social) systems. They are statistically irreversible within corresponding domain or system, and lower trends also apply in higher domains. I think these trends can be generalized into a meta-trend, defining transition or metaevolution from lower- to higher- complexity domains.

Many other observers (such as Ray Kurzweil) also see matter, life, and intelligence as phases in the evolution of the universe. But they use scale and complexity as a measure of progress. That presents a problem: galaxies are bigger than brains and nothing is more complex than random noise. Of course, they mean functional complexity, but neglect to clearly define what that function is.

Any system is defined by some relatively conserved core, such as genotype in biology. Propagation of that core is effective function of a system, which serves as a relatively fluid or adaptive “phenotype”. To be more effective, such phenotypes grow and functionally sub-differentiate, while the core becomes relatively smaller, isolated from environment, and less direct in controlling the system. Since only the core is ultimately conserved, I propose that direction in metaevolution is defined by the second trend.

In effect, environment or ecosystem differentiates into increasingly deep “conservation food chain”, with growing disparity between the most conserved cores and the fastest-adapting peripheral systems.
Competition between cores favors more broadly instrumental elements within each. This leads to evolution of more “abstract” cores, that define lesser proportion of their more “bottom-up” phenotypes.
For example, initial conserved cores in biology were autocatalyzing proteins and RNAs, later displaced by DNA genes, more flexible and coding for multiple and changing during phylogeny RNAs & proteins.

Superficially similar to my interpretation are The Major Transitions in Evolution and Meta-Systems Transition Theory. From John Stewart’s (hopelessly utopian) Evolution’s Arrow:
Two attributes that increase as evolution proceeds are the scale of cooperative organization and evolvability through the discovery of effective adaptations“. But pure cooperation is a product of group selection: a notoriously inefficient mechanism. In terms of conserved traits, what’s far more likely is preferential selection of a some common subset, while individual specifics become dispensable.

Hence, I focus on the evolution of increasingly refined/ abstracted conserved traits, though they do depend on correspondingly complex phenotypes. The later however is a cost, while core propagation is a benefit (net result) of this trend. These propositions are mostly tautological, but so are the concepts of thermodynamics, survival of the fittest, and algorithmic complexity theory, not to mention all of math.
My core types are conceptual and the transition between phases is gradual, with expanding overlaps:

Entropy growth: equalization and stabilization of matter and energy distribution across space-time for all interacting entities. Maximized fitness here is equality of distribution or continuous recurrence.
Biological evolution: restoration and reproduction of genomes by selective acquisition of their constituents. The fitness here is discontinuous recurrence: mediated by differentiated phenotype.
Cognitive exploration: recognition and projection of correspondence between inputs and predictions. The fitness is hierarchically projected match between environment and a model: cognitive phenotype.

Living organisms metabolize matter and energy to propagate their pattern (information), while cognitive systems “metabolize” patterns: learn and forget information, to increase their predictive power.  
Other posts here elaborate on each phase of core abstraction during “history of time".

Entropy growth as an *increase* of order


The only universal trend in physical world is the growth of entropy, toward some kind of equilibrium.
Entropy is usually considered to be a measure of disorder, antithetical to reproduction & prediction.
It is well known that physical entropy growth increases complexity of the exact description of micro-states in all interacting systems. But that complexity is irrelevant for overall behavior of these systems because it is random: the differences between micro-states cancel-out on a macro-level. What matters for whole-system prediction is macro-gradients, which don’t cancel out. Such gradients are reduced in interactions, increasing effective order or predictability of the system. This interpretation is my own.

Physical disorder appears to increase in interactions because Information Theory measures it by the number of bits in the description, which is compressed, rather than by the difference in physical values, which is not. Variation within contiguous objects is lower than that between them because interactions follow S-T continuum. Interaction between the objects will reduce overall differences, but increase those within each. Total variation will be reduced in magnitude, but increased in the number of smaller differences, representation of which is less compressible. So, the description of closed system may require more bits as a result. But physics study empirical values, not their descriptions.


The above describes effects of repulsive interactions. For attractive interactions (gravity) the disparity in concentration across Space is increasing, but so does stability of the results, - the change over Time is decreasing (down to a “timeless” black hole). The distinction of Time from Space is its irreversible direction. Any reversal can be defined only in terms of a higher-dimensional coordinate. For a top dimension, there is no higher coordinate, so a reversal is inconceivable. Irreversibility of Time means that it is a top dimension, thus equalization predominates for both repulsive & attractive interactions. However, Plank scale is a limit for any continuous interaction, thus also for resulting entropy growth.

So, both the growth of entropy & that of reproduction represent increasing stability over time & similarity across space. The difference between them is in scope: entropy increases in all interacting systems (no spatially distinct adaptive interface), while reproduction is restricted to a genotype or an equivalent. Obviously, the growth of entropy drives all biological processes. Still, simple averaging is infinitely inferior to biological evolution, & most would never consider it as a form of meta evolution.
On the opposite, the "heat death" seems be life's worst nightmare. But evolution is just as inferior to technological progress driven by cognition, considering its speed & potential. And reproduction at its ultimate extreme, "the gray goo" scenario, seems just as abhorrent as the heat death. Some rational version of “Omega point” would be a final state for representation, - as depressing as anything final.


Indefinite transition to higher core types is possible only if the universe is open, - meta evolution in a closed universe will eventually run out of variation & die a heat death. The universe is closed if it can be fully described by a fixed number of patterns / physical constants. In that sense, a known universe is always closed, it just seems to get larger all the time. 
A real question is whether unknown universe is empty, - will there ever be any new information / external impact? Bayesian take on the unknown is that all possibilities should be assigned equal probability. Since empty is one distinct possibility, & not empty is an infinite number of them, a priori chances that the universe is open are infinity-to-one.
Of course, any ecosystem can be effectively closed, - evolution may be reversed by entropy (lack| excess of free energy) or reproduction (food chain exhaustion or predators & parasites). Any given system has only an infinitesimal chance of evolving forever, but meta evolution should prevail overall.

From within the known universe, it does appear to be "closed” in terms of GR & QM constants, - no limits or exceptions were found in a hundred years of intense research. But, however fundamental, these cover only a tiny part of our knowledge. All other known patterns do decay/terminate with distance, in the frame of reference used to define them (smaller patterns are normally stronger/ more evolved, as larger systems evolve out of them later). 

This negative meta-pattern of decay can't be interrupted by any known law, - there can never be an evidence of infinite range for any pattern. So, this decay meta-pattern blocks projection of GR & QM into infinity, although from outside of their frame of reference. Cosmological predictions are theoretically interesting, - that’s how things should develop without intelligent intervention. But such intervention will be armed by science of the future. Modern science is only ~300 years old, in a hundred years it should be as relevant as medieval alchemy is today.

1.08.2012

Biological evolution: maximization of reproduction


Pre-biological types of broadly-defined reproduction are:
Restoration: ability of fermions, atoms, & molecules to maintain their structure while exchanging energy & constituent particles. That’s a spatially- or temporally- discontinuous pattern preservation: replication over time, but not over space (which would be reproduction in conventional sense).
Replication: atoms/molecules are selected by template from an environment, as in restoration. But the new layer then selects the next layer of constituents from the environment, as in crystals, or is detached & then each layer repeats the cycle, as with prions, RNA & perhaps some other nucleic acids.
The selection on these two stages is by direct replicational fitness, with no differentiated phenotype.

Moving up to biology, a good start is "Evolution in Four Dimensions: Genetic, Epigenetic, Behavioral, and Symbolic Variation in the History of Life” by Eva Jablonka & Marion Lamb. The book reviews types of “heritable” traits of increasing complexity & variability, & their transmission mechanisms.
I'd quibble that these are
stages rather than dimensions: they did not appear all at once but evolved sequentially. More importantly, accelerating variation on higher stages requires new selection criteria for the traits, which E4D addresses only tangentially. Selection by reproductive fitness works for genetic variation, - mutation rates are comparable to the duration of reproductive cycle. But starting with epigenetics, & increasingly so for behavioral & symbolic traits, variation is much faster than reproduction. These traits must be selected by increasingly abstract representational criteria, such as memory of usage, proximity/ similarity to such memory, & predictions thereof by projecting patterns:

Epigenetic modulation: reinforcement of frequently used, & inhibition of the unused functions. Partly heritable, epigenetic memory of usage modulates gene expression in ontogeny of somatic phenotype.
Neural conditioning of stimuli recognized as reliably associated with usage via some sort of proximity. Conditioning drives the development of more complex & dynamic behavioral patterns, enabling an increasingly differentiated food chain, - an ecological phenotype.
Cortical cognition: discovery & projection of patterns (mutually corresponding stimuli), potentially predictive of usage or its proximity. Such patterns are encoded into predictive models, accumulated & refined across generations via language & recorded media. Symbolic models enable flexible functional differentiation & technological phenotype of human society. To elaborate on each of the above:

Memory of usage: Life forms evolve a differentiated somatic phenotype, which selectively accumulates material & energy to facilitate preservation & reproduction of a genotype. Phenotypic functions are selectively reinforced/inhibited by Lamarckian use & disuse within a reproductive cycle via epigenetic variation. Moreover, such variation is often heritable, shifting initial activity levels across generations (Lamarck has been partly vindicated, see E4D). E4D lists four types of epigenetic inheritance: self-sustaining feedback loops, structural inheritance, chromatic marking, & RNA-mediated inheritance, & there're probably many others. Given that complexity of genotype in phylogeny is growing very slowly, the much faster increase in complexity of somatic phenotype must be driven largely by epigenetics:

Basic life adds an alternative gene expression/transcription layer to a replication cycle, with the genes also serving as templates for RNA, proteins & so on. Gene expression changes in sequence controlled by master genes, modulated by environmental impact via methylation & other types of "imprinting". Such "memory" of usage alters sensitivity to future stimuli by changing the number & responsiveness of corresponding "receptors". When heritable, this memory speeds up ontogenetic development.
Eukaryotes add a more adaptive "cell body" phenotype layer to a relatively conserved nucleus, further sub-differentiating reproductive cycle. The cell body is less "conserved", - it has much faster & more adaptive replacement cycle than the nucleus. In fact, the cell body & some organelles (mitochondria & chloroplasts), were probably acquired by the nucleus (former archaea) from foreign bacteria.
Multicellular organisms add differentiated somatic cells, & corresponding alternative gene expression cycle, to the original germ cells. Somatic phenotype can be hugely more complex & adaptive through the reproductive cycle, & some of this adaptation is also epigenetically heritable.
Sexual dimorphism seems to add a "test" phenotype. Reproductive phenotype is largely female, while males of many species only contribute their genes, serving as "one half of a germ cell". On the other hand, that half is less conserved than the female half: human Y chromosome differs from that of a chimpanzee by ~30%, as opposed to ~2% for the rest of genome. Apparently, males function as a still more adaptive layer of phenotype: a test vehicle for externalized & then sexually selected variation.

Conditioning: Sufficiently advanced somatic phenotype gains controlled spatial mobility, initially guided by inherited & usage-modulated reactions. Mobility leads to the development of higher levels of ecosystem: animals feeding on plants & carnivores feeding on herbivores. In effect, this "food chain" forms an extended phenotype, with "variation" vastly accelerated due to mobility. A correspondingly adaptive new layer of internal phenotype evolves on the higher levels of the food chain: sensors & nervous system. This representational phenotype selects patterns of stimuli to guide behavior by reacting to ecological variation, utilizing initially chemical & then also electrical signaling. However, in a dynamic environment, modulation by past usage alone soon becomes dysfunctional. To be useful, the responses must be modulated by increasingly predictive stimuli. A stimulus is predictive if it consistently co/counter-occurred with inherited, usage-modulated, or previously conditioned stimuli. The former are conditioned to activate or suppress reactions associated with the latter in the past.

Conditioning evolves to potentiate stimuli by increasingly distant & indirect coincidence| precedence:
Direct temporal associations are likely conditioned by amygdale, initially to invoke 4F-type drives.
More complex
spatial associations of memories seem to be formed by hippocampus, which contains a map of "place cells" for spatial orientation. Another function of hippocampus is forming declarative memories, probably by facilitating their transfer from primary into association cortices. So, my interpretation is that hippocampus "conditions" memories by associating them with value-loaded locations (in evolutionary context “valuable” must be, or likely to become, close to the subject). Without such association memories are not recognized as important, thus do not become declarative. Discovery & conditioning of still more abstract associations probably requires neocortex. These include operant conditioning of actions by their consequences, & instrumental conditioning of “tools" by their products. Among higher social animals conditioning is "heritable" by imitation, which accelerates acquisition of behavior & facilitates development of "animal traditions".

Cognition: Predictive modeling adds a new layer to representational phenotype: patterns of value-free stimuli. The selection criterion for these patterns is correspondence, quantifiable as cumulative match among inputs, which indicates their predictive power. Such patterns may later be conditioned if their projection is associated with correspondingly projected value patterns. This probably requires some form of cortical minicolumns. Cognitive phenotype beyond neocortex may include oral & recorded language, as well as institutional culture & science: Jablonka's "symbolic dimension". But a more effective selection criterion, - correspondence vs coincidence, is far more important because it outlast any specific transmission method for its selections. Conditioning can also be transmitted via external media, but by itself it would never produce anything like human civilization. Cognitive prediction & planning enables vastly expanded external phenotype: ecological, by cultivating plants & animals, social, via dynamic functional differentiation, & technological, by manufacturing structures & tools.

I won't get into detail on ecological & technological phenotype hierarchies (as I did with somatic hierarchy). That would be too much of a distraction, & I am sure anyone smart enough to read this can conjure his own versions. I’ll also skip “societal phenotype” – it’s too complex for a brief treatment. Again, these are layers of operational phenotype, acquired by a far more fundamental representational phenotype: mechanisms of conditioning & pattern recognition.
A similar interpretation of ecosystem & technosphere is explored in Richard Dawkins' excellent "The Extended Phenotype: The Long Reach Of The Gene". But beside the genes, I think such "phenotype" also propagates more abstract & functional conserved cores: mechanisms of response reinforcement or inhibition by usage, coincidence, & prediction. Such mechanisms should displace genes as the ultimately conserved core as soon as a system develops more efficient reproduction media.

To reiterate: The genome becomes more stable in higher organisms: mutation rates are the highest in bacteria & decrease as the organisms become larger & longer-living. At the opposite extreme, humans are exceptionally uniform genetically. This is because of improved protection for genome, as well as longer reproductive cycle (mutations accumulate during reproduction). Despite this slowdown of genetic variation, the growth of phenotypic complexity appears to accelerate during phylogeny. This can only be explained by increasing proportion of complexity selectively acquired during ontogeny, according to increasingly mediated representational fitness. 


Reproductive fitness is still the ultimate arbiter in biology, but the shift from innate to conditioned motives begins a transition to correspondence-maximizing phase of metaevolution. This transition would be complete with the next shift: from conditioned motives to purely cognitive curiosity, discussed in my next post:



Human motivation: developmental perspective



The beginning of cognitive phase in metaevolution as we know it is human civilization. Conserved core here is human motivation, while technology and institutions is a more fluid instrumental phenotype. According to their conserved vs. adaptive properties, our motives can be grouped into three categories: instincts, conditioned values, and pure curiosity. Motives of the two higher categories are obviously acquired, what’s conserved here is their value acquisition mechanism:

- evolution selects instincts fit for their own propagation, innate but postnatally modulated by usage,

- conditioning value-loads stimuli preceding (instrumental for) previously value-loaded stimuli,

- cognitive curiosity searches for predictive patterns, even among initially value-free stimuli.

 Higher mechanisms accelerate acquisition of adaptive values to drive increasingly mediated responses: from specific physiological reactions to more general longer-term attention, prediction, and planning. Brain areas that implement these value-acquisition mechanisms likely evolved in the same sequence:

- Instincts, largely physiological and rooted in 4Fs, are encoded mainly in brainstem and hypothalamus

- Conditioning is initiated by basal ganglia and limbic system, then generalized in cingulate cortices.

- Value-free curiosity is an intrinsic drive of neocortex, though heavily modulated by subcortical drives.

This scheme is vaguely similar to triune brain model, but in my interpretation these substrates differ mainly in the mechanism by which they acquire values, while values themselves are relatively transient. Relative strength of these mechanisms themselves is also changed via lifelong mutual reinforcement and inhibition.

Our instincts are similar to those of other mammals. Excellent account of that level of motivation is Jaak Panksepp‘s “Archaeology of Mind: Neuroevolutionary Origins of Human Emotions“. He catalogs seven innate drives: rage, fear, lust, care, grief, play, and seeking. Not included are homeostatic drives, related to feeding, sleep, pain, thermoregulation, etc. They are more basic but mechanically similar.

 Panksepp singles out “seeking” as an exceptionally non-specific drive. It’s the same as my curiosity, but for him the principal driver is dopaminergic areas, as they were in lower animals. I believe this function has been taken over by neocortex in mammals, especially in humans. Now these lower areas are mostly passive mediators of cortical feedback: Dopamine's Role in Unrewarded Learning, or they simply mess you up with some stupid urges.

 The discussion below is mostly on conditioning and cognition: increasingly adaptive mechanisms that seem to strengthen with our personal growth. Until it hits harsh constraints of biological life cycle.


Conditioning by increasingly general instrumentals


Alternative hierarchical schemes are Maslow's hierarchy of needs and ERG. But both treat higher needs as innate, though latent, while I think they develop via instrumental conditioning. Especially the top of both hierarchies: self-actualization or a drive toward excellence in a chosen field. Activity in modern fields can’t be innate, so the drive to “self-actualize” there must come from instrumental conditioning.

Conditioned values competitively inhibit and potentially displace each other, as well as innate instincts. Such "value drift" is necessary because behavioral patterns for effective reproduction of higher animals change too fast to evolve with their genotype, and are far too complex to fit in it. Goal-directed planning greatly accelerates this drift, because conditioning can now be driven by predictions vs. experiences.

This selects for increasingly general instruments, conditioned by a wider range of benefiting values. Even our notion of self is acquired. Basic self is a conditioned identification with one’s body, as a tool to control sensory stimuli that trigger innate drives. Which results in self-preservation and self-promotion drive: childish impulsiveness is substantially displaced by adolescent egocentrism. Next essential “tool” to be conditioned into one’s identity is society. It starts with one’s mother, initially almost as necessary as a body, then extends to family, peers and mates, tribe, nation, humanity. Rough Freudian parallels are ID for the drives, Ego for a body, and Super Ego for social identification.

The broadest human value with innate component is empathy or affinity. Probably via medial frontal cortex, it conditions external phenotypes recognized as similar to self-image, thus building on prior instrumental conditioning. But self-image changes radically during maturation. It is then anchored by relatively stable body and proximate society. But if these change, self image will change too.

Conditioning by similarity probably emerged because evolution doesn’t select for individuals, it selects for genes. Similarity is the proxy for affinity: presence of same genes in a different individual. So, empathy with and resulting support of similar individuals is instrumental to propagation of one’s own genes. “Gene” here is a pattern rather than a substance. Metabolism and reproduction don’t preserve substance: atoms come and go, only their pattern is maintained and propagated.

Empathy becomes more abstract as recognized affinity expands. Tribal identification is loaded with specifics: customs, mythology, language, diet, clothing and housing, ethnic history and homeland. All these details gradually lose significance as one progresses toward universal human values. Which are pretty hard to define. Even the most basic of our drives may be subverted by acquired values: pain may be counter-conditioned into masochism, hunger into anorexia, lust into asexuality, self-image into suicidal ideation, familiar and tribal loyalty into religious fanaticism, etc.

Our evolutionary imperative: reproductive drive, is already profoundly confused: birth rates decline in the wealthiest and the most educated social groups. And of course, no one would spend his life's savings to manufacture ever greater amounts of his DNA. In fact, the genes may go extinct altogether, as soon as we have better tools to produce phenotypes that we currently value.

Increasing respect for freedoms and cultural diversity makes modern social identification ever more tenuous, while formalization and automation reduce the role of empathy in economic interactions. Then there is our biology and developmental patterns, but future life extension and mind expansion should remove even that anchor from our common identity (Francis Fukuyama’s worst nightmare).

The only two drives left in common will be instrumental self-improvement and curiosity. And curiosity should subsume the former because it drives cognition: the only universally instrumental tool. Cognition also drives invention, which is a vastly accelerated evolution. And the concept of evolution is an ultimate target of instrumental conditioning: evolution produced humanity itself. 

In effect, cognition should be recognized and then conditioned as being instrumentally superior to conditioning itself. This is similar to the way conditioning may inhibit and displace instincts, even though they initiated conditioning in the first place. So, given sufficiently deep self-reflection, curiosity should displace all other motives. Which already happened for the best scientists.

So, curiosity is an intrinsic cortical motive, but it competes / combines with all sub-cortical motives. As with other values, relative strength of curiosity is modulated by conditioning. Being the most generally-instrumental motive, pure curiosity has longer and less obvious payoff. Thus, it initially works the background, “researching” subjects selected by other, more urgent motives. But range and depth of such research spills over multiple subjects, ultimately into curiosity about fate of the universe and the meaning of life. This is facilitated by a broad intellectual exposure, if combined with weak specific pressures and temptations, especially during one’s formative years. 

Hippocampus is necessary for declarative memory consolidation, hence for sustained cognition. Evolutionary reason for this constraint is probably that hippocampus also mediates conditioning by spatio-temporal association, via its “maps“ and grid cells in conjunction with its direct connection to amygdala. All behavior is local and memories aren’t be valuable enough for animals to consolidate unless associated with emotionally charged locations: proximate in past, present, or potential future.

Despite such constraints, cognition will displace conditioning because its mechanism will be recognized and conditioned as being instrumentally superior to that of conditioning itself. Mechanically, that means frontal cortex will gain effective control of VTA, NA, amygdala, and such. This similar to the way conditioning inhibits and displaces instincts, in spite of the fact that instincts initiated conditioning.


Popular culture vs. value generalization


Looking around, eventual dominance of higher motives and corresponding expansion of attention span seems quite implausible. Mass media serves our innate preference for 4Fs, intense sensory stimulation, and novelty: the lowest common denominators of curiosity. Constant exposure desensitizes and addicts us to such entertainment, which actually *shrinks* attention span in much of modern society.

Likewise, wealth weakens instrumental conditioning by easily satisfying lower motives, starting with fast food and junk TV / games / social media for children. Still, a proper upbringing could raise motivation to a level where instrumental self-improvement becomes a main motive and inhibits the lower ones.

Also, value generalization process is constrained by the limits and pressures of our life cycle, as well as the range of experience. Most of “identity expansion” I described is over by the end of adolescence, when myelination of neocortex is mostly complete and neuroplasticity declines. Jobs, dating, family and child-raising puts a lot of stress on people, shrinking their purely intellectual.

Cultivating values in children expands, but also stabilizes one's own values: a good actor must believe his act. Also stabilizing is a fixed social role: at work, in extended family, & broader community. Later, aging weakens more general motives as the loss of myelin impedes stimuli propagation to and from higher levels of cortical hierarchy. And constant health concerns further displace less urgent motives.

But overall, there are long-term trends toward extended adolescence, reduction of specific pressures via longer general education, delayed and reduced family formation, more fluid social roles, life extension. Even more important could be extreme educational aids and environments, as well as some form of direct brain stimulation to improve focus on abstract problems, which is likely to become available soon. These trends have no end in sight, they should extend value generalization process indefinitely.

Recapitulation & more “meta”.

Equalization, or simple entropic averaging, forms continuous patterns by increasing similarity between interacting objects, at the expense of decreasing similarity within each (variation drift to smaller scales).
Reproduction increases core/environment differentiation
by selectively replicating internally differentiated genotype pattern, & by introducing externally differentiated adaptive phenotype which collects the "constituents" for such replication. The "ecosystem" for reproduction must be non-random: "stabilized" by prior entropic equilibration.
Correspondence maximization further differentiates a core from environment. Only the quantity of an input is represented: the sensors "modulate" internal "memory" into a quantitative equivalent of an external impact, with minimal incorporation of the constituents. Moreover, only the match among inputs is maximized, by "selecting" only their projected coordinates, not the inputs themselves. The conserved core here is predictive correspondence, patterns & algorithms themselves are adaptively "metabolized" to maximize it. The "ecosystem" for representations, such as brain, develops by previous reproduction
of functionally equivalent memory & recognition processing "cells".

A higher-phase core "exploits" the lower-phase ones within its adaptive phenotype. This
competition among core types is not obvious: life accelerates the growth of entropy, & cognition requires increasing numbers of "reproduced" memory & processor cells. But internal & external differentiation of a genotype "disrupts" S-T continuity (entropy) by increasing specialization & metabolism. And predictive correspondence of a cognitive system is maximized by accelerated "pattern metabolism" in memory, reducing stability of representations. Memory & processor cells don't have to be physically identical, only functionally equivalent, & technical progress accelerates the turnover of computational substrates. In other words, fitness of a more-abstract core comes at the cost of less stable specifics in its adaptive hierarchy, increasing the disparity in recurrence
between a core & its environment.

A more abstract core requires & enables more
functionally differentiated adaptive hierarchy, which is correspondingly resource-intensive. Metaevolution is a universal, rather than local, trend only if the best more complex systems are a more efficient than the best less complex ones. That seems plausible because specialized instruments
produced by functional differentiation are intrinsically more efficient than multi-purpose ones. But this efficiency is realized only in correspondingly diverse environments, with enough targets for such specialization. Such diversity is increased by functional differentiation:
a) larger systems form a new level of ecosystem, b) their resources enable greater range of winner-takes-all competition, & effective diversity is likely to grow with the range. So, differentiation drives a self-reinforcing cycle, providing a competitive advantage for still more differentiated systems.

There are lower forms of recurrence than classical entropy, & higher than cognitive representation:
-
Quantum probability should be l
ower than entropy: matter/energy conservation & resulting entropic equalization stops on the scale of quantum effects. So, the recurrence maximized by classical entropy is not really continuous, not nearly as much as the distribution of probability within a quantum wave. In fact, there can be no ultimate beginning for metaevolution, - perfect randomness is inconceivable.
-
Mathematical compression,
derived by deduction, is more abstract than cognitive modulated correspondence. Math is a body of computational short-cuts, which are not representative of reality per se, but are universally instrumental for discovering predictive representations. Beyond known math, there must be indefinite number of potentially instrumental meta-representative layers.

Meta evolution as proposed here is a process of selective abstraction & spatio-temporal stabilization of
mutual correspondence within any “ecosystem”. Ladies & Gentlemen, the meaning of life, universe, & everything: increasing recurrence of relatively shrinking conserved cores.

Comments from the knol

Todor Arnaudov:

Socialization



>Overall, the value drift “selects” for increasingly general instrumental motives, -
>they simply last longer, being conditioned by correspondingly broad range of associated prior values.

Also, more general instruments could allow doing more with less efforts in less time - generalization can be optimization.

>Childish impulsiveness is substantially displaced by adolescent egocentrism,
>which in turn is displaced by increasingly broad socialization.

Then sometimes it's followed by selective de-socialization/change of the POV to society
for subjects who has reached to understanding of the reasons for their prior socialization, because it's conditioned as well.

>This development of broader "self" is initiated by specific inherited motives: somatosensory feedback,
>imitation & respect for authority, the patterns of human beauty, sexual drives, parental instincts, the need for
>social support & status, empathy or altruism.

I'd question that respect of authority is inherited motive, authority just can punish you or you're afraid it can/will; and you believe it's more capable and more dangerous than you, you respect it because of this and because you believe that others believe that it is, and they may punish you as well.

Respect of authority can be derived very simply: the only "animate object" for a new born mind are mother, father etc. They are dynamic and harder to predict, unlike environment; they change "randomly" sometimes, they are there when needs are satisfied and they are missing when needs are not satisfied (or are there where mind is punished), they are older, bigger, have money and decide instead of you etc. This can be badly conditioned. Further church (where it has this access to people), school, police, army, government etc. make people respect authority by displaying their "authority" == force and superiority.

I think also that empathy, altruism, the need of social support are not inherited as well, they are product of the environment and availability of competing agents around. I guess all of those are initiated the same/similar way as respect of authority and have common source. Also they are a form of optimization/maximization such as team work etc.

...

However in a world with only one person who doesn't know that other intelligent beings have existed or could have existed, there couldn't be altruism or social support at all. If one is capable and secure enough to be "untouchable" in a normal world with billions of agents needing social support, he also may escape from the social support urge, I guess you know an example

I guess patterns of human (physical) beauty and general visual beauty (graphics design etc.) are implicitly inherited, they are rediscovered and shared by many because of our common cognitive algorithm.




Last edited Jan 12, 2011 11:23 PM
Report abusive comment


>Also, more general instruments could allow doing more with less efforts in less time - generalization can be optimization.

Obviously, but I was trying to explain how instruments can become more psychologically important than their original purpose.

> Then sometimes it's followed by selective de-socialization/change of the POV to society
> for subjects who has reached to understanding of the reasons for their prior socialization, because it's conditioned as well.

Right.

>I'd question that respect of authority is inherited motive, authority just can punish you or you're afraid it can/will; and you believe it's more capable and more dangerous than you, you respect it because of this and because you believe that others believe that it is, and they may punish you as well.

Respect for authority is inseparable from imitation instinct, - that’s how you choose who to imitate. Imitation is accellerated aquisition of knowledge learned by others. That’s a core of social instinct, & some animals are obviously more inherently social than others. Another part of it is a “follow the leader” instict, necessary for many group actions, & also common for all social animals.

>I think also that empathy, altruism, the need of social support are not inherited as well, they are product of the environment and availability of competing agents around. I guess all of those are initiated the same/similar way as respect of authority and have common source. Also they are a form of optimization/maximization such as team work etc.

It’s hard to distinguish between inherited & acquired parts, but most basic values have inherent element to speed-up subsequent conditioning. But “inherited” does not mean “permanent”, it can be inhibited just like acquired motives.

>However in a world with only one person who doesn't know that other intelligent beings have existed or could have existed, there couldn't be altruism or social support at all.

In such situation the “person” would not feel as an individual, rather as a society / ecosystem of his constituents.

>I guess you know an example

I am sure there’s more than one, most for a wrong reason. I don’t run on a different algorithm, but my stimulus-to-response delay is much longer than what's decent & proper (if the stimulus is novel).

>I guess patterns of human (physical) beauty and general visual beauty (graphics design etc.) are implicitly inherited, they are rediscovered and shared by many because of our common cognitive algorithm.

There’re many different elements to beauty, the most abstract part is simply a balance between pattern & novelty that one finds intellectually stimulating at the moment.

Report abusive comment
Posted by Boris Kazachenko, last edited Oct 19, 2010 2:25 AM
>Also, more general instruments could allow doing more with less efforts in less time - generalization can be optimization.
>>Obviously, but I was trying to explain how instruments can become more psychologically important than their original purpose.

I meant that association areas may not inhibit primary cortices but serve for optimization of lower ones' tasks, and all can develop together. Going too abstract without connection to reality may lead you to insanity or helplessness and vulnerability, I don't think too strong association areas *without* solid enough lower levels is healthy. Also, in order to create a thinking machine or any machine, one needs physical technology, someones' minds should develop it.

>But “inherited” does not mean “permanent”, it can
> be inhibited just like acquired motives.

OK

>Respect for authority is inseparable from imitation instinct, - that’s how you choose who to
> imitate.
>Imitation is accellerated aquisition of knowledge learned by others. That’s a core of social
> instinct, & some animals are obviously more inherently social than others. Another part of it is a
> “follow the leader” instict, necessary for many group actions, & also common for all social
> animals.
>It’s hard to distinguish between inherited & acquired parts, but most basic values have inherent
> element to speed-up subsequent conditioning. But “inherited” does not mean “permanent”, it can
> be inhibited just like acquired motives.

My concern is about the abstractness and fanciness of terms. “Acquisition of knowledge learned by others” is complex and implies understanding of concepts of others, while I guess it might be simple such as:

- Detect important input (dynamic, novel)
- Record
- Replay / Compensate (follow) to keep it as input

It turns to seem “social” because it always happen around other individuals, while (I guess) it would work with fake stimuli. Imitation (record-replay) is the simplest form of filling procedural memory (or any memory) if it was empty or when mind lacks experience to decide on its own. Choosing what to record/ (what's important) as “respect of authority” could be recognition of the most frequent/strongest/dynamic input etc. and then it's generalized and conditioned to obey.

Baby initially has no choice, but imitating/mapping (recording senses into empty mind) and “respecting authority” (helplessness). Eventually it disrespects it (egocentrism, self-identification), but then it respects it again (can't win to the grown-ups, they are more powerful). Empathy is seeing yourself in the shoes of others, it's praised while it might be a mirrored-egoism. Society also teaches children subordination and social hierarchy (hierarchy of [potential] power/skills/scope), and children learn that they can get some things they need by social activities (collaboration optimization, sex, etc.) and by climbing the social hierarchy (as power/skills...).

>>However in a world with only one person who doesn't know that other intelligent beings have existed or could have existed, there couldn't be altruism or social support at all.
>In such situation the “person” would not feel as an individual, rather as a society / ecosystem of his constituents.

Well, this is true even now (at least for the ones who are smart enough); we're not “individuals” and there's no integrated “self”, it's a degree of generalization and resolution of covered space, time, scope, conceptual clarity etc.

>There’re many different elements to beauty, the most abstract part is simply a balance between >pattern & novelty that one finds intellectually stimulating at the moment.

Agreed.
A “lower” kind of beauty I see is the associations to stimuli of pleasure/need, such as pictures of nudity. The concepts are confused.

Report abusive comment
Posted by Todor Arnaudov, last edited Oct 20, 2010 5:25 AM
> I meant that association areas may not inhibit primary cortices but serve for optimization of lower ones' tasks

I don’t see how it relates to the subject, & “optimization” doesn’t sound very meaningful. More general value-loaded concepts direct lower levels’ spatio-temoral focus, & adjust their detail vs scope balance. Is that what you meant?

>Going too abstract without connection to reality may lead you to insanity or helplessness and vulnerability, I don't think too strong association areas *without* solid enough lower levels is healthy.

“Higher association areas” are in the best position to decide on the balance between induction & deduction. Their “strength” is the degree of control over lower levels’ focus, not necessarily greater degree of generalization overall. Insanity is the *lack* of intellectual integrity, - disruption of higher levels or their connections to the lower ones. Helplessness, vulnerability? Well, you make your best guess & take chances. The outcome is never certain, but the alternative is... mediocrity.

>“Acquisition of knowledge learned by others” is complex and implies understanding of concepts of others,

No it doesn’t, knowledge is always partial, especially second-hand.

> It turns to seem “social” because it always happen around other individuals, while (I guess) it would work with fake stimuli. Imitation (record-replay) is the simplest form of filling procedural memory (or any memory) if it was empty or when mind lacks experience to decide on its own. Choosing what to record/ (what's important) as “respect of authority” could be recognition of the most frequent/strongest/dynamic input etc. and then it's generalized and conditioned to obey.
>Baby initially has no choice, but imitating/mapping (recording senses into empty mind) and “respecting authority” (helplessness). Eventually it disrespects it (egocentrism, self-identification), but then it respects it again (can't win to the grown-ups, they are more powerful). Society also teaches children subordination and social hierarchy (hierarchy of [potential] power/skills/scope), and children learn that they can get some things they need by social activities (collaboration optimization, sex, etc.) and by climbing the social hierarchy (as power/skills...).

Yet, some people seem to be a lot more social than others, even if you control for upbringing & intelligence. But it’s easier to see in animals, just compare different breeds of dogs. And there’s been plenty of evidence on the effects of oxytocin, which you’ve been going off about. That’s genetics, google prairie vole + oxytocin. As for authority, it probably has something to do with testosterone or lack thereof.

>Empathy is seeing yourself in the shoes of others, it's praised while it might be a mirrored-egoism.

No, it’s recognizing yourself in others, & then *conditioning by similarity*, a category different & independent from instrumental conditioning. That’s basic enough to be an inherited mechanism, right? Obviously, on the level of self-identification, it’s a lot stronger in social animals (& in women).

Report abusive comment
Posted by Boris Kazachenko, last edited Oct 21, 2010 5:00 AM
>I don’t see how it relates to the subject, & “optimization” doesn’t sound very meaningful. More general value-loaded concepts direct lower levels’ spatio-temoral focus, & adjust their detail vs scope balance. Is that what you meant?

I'm not sure, sorry to bother you.

> Their “strength” is the degree of control over lower levels’ focus, not necessarily greater degree of generalization overall.

Aha, OK.

>>Going too abstract without connection to reality may lead you to insanity or helplessness and >>vulnerability, I don't think too strong association areas *without* solid enough lower levels is >>healthy.
>“Higher association areas” are in the best position to decide on the balance between induction &
>deduction. Their “strength” is the degree of control over lower levels’ focus, not necessarily
> greater degree of generalization overall. Insanity is the *lack* of intellectual integrity, - disruption
> of higher levels or their connections to the lower ones. Helplessness, vulnerability?
>Well, you make your best guess & take chances. The outcome is never certain, but the alternative is... mediocrity.

I guess hypertrophy of higher levels in parallel with too much of inhibition of the lower ones may disrupt these connections and endanger the existence of the higher “arrogant” level itself.

For example, one may embrace AGI/progress/generalization as a meaningful purpose to exist, but with a little bit of tuning, one may see it purposeless as anything else. Progress? There's no real progress, everything is just a bunch of patterns - particles, configurations of causal forces, sequences of events or whatever. Understanding of intelligence is meaningful to a mind, but the difference to Universe will be local and only at a given scope of measure. For other scopes it will be the same patterns in different configurations.

>Yet, some people seem to be a lot more social than others, even if you control for upbringing &
> intelligence. But it’s easier to see in animals, just compare different breeds of dogs. And there’s
> been plenty of evidence on the effects of oxytocin, which you’ve been going off about. That’s
> genetics, google prairie vole + oxytocin. As for authority, it probably has something to do with
> testosterone or lack thereof.

Sure there is inherited part in the extent of being “social”, I can't question this. I'm speculating about the extent/complexity of what actually is inherited and what is the core of being “social” (etc.) and not in terms of chemicals. I suspect it's coarse and simple, like sort of a tuning of thresholds for “fixating” stimuli. Respect of authority seems too general a term to me. I tried to speculate on some mechanisms regarding “dominance” with more functional terms, I'd better put it in an article or so.

>>Empathy is seeing yourself in the shoes of others, it's praised while it might be a mirrored-egoism.
>No, it’s recognizing yourself in others, & then *conditioning by similarity*,

I know about similarity, I guess this is how people empathize animals and why use to empathize higher animals much more than lower and are less likely to feel guilty about hurting insects or plants than mammals or birds. However I classify recognizing *yourself*“ and similarity to *yourself* as “egoism”.

>it’s recognizing yourself in others, & then *conditioning by similarity*,
>a category different & independent from instrumental conditioning.
>That’s basic enough to be an inherited mechanism, right? Obviously, on the level of self->identification, it’s a lot stronger in social animals (& in women).

OK, it's basic and conditioning by similarity is different from instrumental conditioning, however CbS is one part of the empathy process, which can be seen as relayed operant conditioning - when empathizing a suffering person, you suffer with him and you want to fix his pain in order to fix *your* pain, etc.

Report abusive comment
Posted by Todor Arnaudov, last edited Oct 24, 2010 4:29 PM
>I guess hypertrophy of higher levels in parallel with too much of inhibition of the lower ones may disrupt these connections and endanger the existence of the higher “arrogant” level itself.

I hate to repeat, but optimal degree of inhibition (or otherwise) can only be determined by those higher levels, & the higher the better. The only disadvantage is slower reaction time, - longer feedback loop. That’s bad in an emergency, but there aren’t many of those in modern life, unless you’re asking for it.

>For example, one may embrace AGI/progress/generalization as a meaningful purpose to exist, but with a little bit of tuning, one may see it purposeless as anything else. Progress? There's no real progress, everything is just a bunch of patterns - particles, configurations of causal forces, sequences of events or whatever. Understanding of intelligence is meaningful to a mind, but the difference to Universe will be local and only at a given scope of measure. For other scopes it will be the same patterns in different configurations.

Right, it’s all equaly meaningless. If you’re clueless, gutless, & like to stay that way.

> Sure there is inherited part in the extent of being “social”, I can't question this.

I thought you did: “I think also that empathy, altruism, the need of social support are not inherited as well, they are product of the environment and availability of competing agents around.”

>I'm speculating about the extent/complexity of what actually is inherited and what is the core of being “social” (etc.) and not in terms of chemicals.

Initial drives + conditioning by similarity (imitation & empathy) + instrumental conditioning (obedience & social identification).

> I suspect it's coarse and simple, like sort of a tuning of thresholds for “fixating” stimuli. Respect of authority seems too general a term to me. I tried to speculate on some mechanisms regarding “dominance” with more functional terms, I'd better put it in an article or so.

More time wasting.

>I guess this is how people empathize animals and why use to empathize higher animals much more than lower and are less likely to feel guilty about hurting insects or plants than mammals or birds.

Right.

>However I classify recognizing *yourself*“ and similarity to *yourself* as “egoism”.

I though you realize that egotism is meaningless concept, the real question is how broad & abstract the “ego” is. And conditioning by similarity increases both.

Report abusive comment
Posted by Boris Kazachenko, last edited Oct 25, 2010 4:52 AM
> Sure there is inherited part in the extent of being “social”, I can't question this.
>>I thought you did: “I think also that empathy, altruism, the need of social support are not inherited >>as well, they are product of the environment and availability of competing agents around.”

Yes, I don't think "social" is inherited in your terms, oxytocin etc. are inherited, which is very far from empathy or respect of authority. If there is no environment with "animate objects", interactive unpredictable input expected to be capable to cause harm or be useful etc. the dog or human will not be "social". "Social" is a vague high-level classification.

Report abusive comment
Posted by Todor Arnaudov, last edited Oct 25, 2010 12:44 PM
>>I thought you did: “I think also that empathy, altruism, the need of social support are not inherited >>as well, they are product of the environment and availability of competing agents around.”
> Yes, I don't think "social" is inherited in your terms, oxytocin etc. are inherited, which is very far from empathy or respect of authority...

The mechanism of the drive to imitate & empathize is conditioning by similarity. That’s a core of being social, & is inherited. The outcome obviously isn’t, I said it many times.
This is more fundamental than I realized, I’ll post a new edit in a few days to elaborate on that, here’re some drafts:

This development of broader "self" is initiated by specific inherited motives: somatosensory feedback, the urge to imitate & the need for authority, the drive for status to gain a reciprocal imitation & authority, patterns of human beauty, sexual drives, parental instincts, empathy & the need for social support: reciprocal empathy...

The broadest inherited drive appears to be empathy / altruism. However, it’s not independently definable: empathy is mediated by recognition of self-image & values in others. This is actually a different type of conditioning: by similarity rather than by instrumental utility. Just like instrumental conditioning, it started as a vehicle for “selfish gene”: similarity is the only way animals can recognize genetic affinity, & then act to preserve & propagate common genes. This is particularly imprtant for social animals, such as us. Affinity also guides imitation, - we’re more likely to imitate those similar to us (other things being equal). That’s a “charisma” element of authority, - competence & power is not enough, to be effective a leader must seen as “one of us”...

And cognition is a higher form of evolution,- a concept conditioned by affinity as a progenitor of humanity & beyond...

Report abusive comment
Posted by Boris Kazachenko, last edited Oct 26, 2010 4:32 PM
B>The mechanism of the drive to imitate & empathize is conditioning by
B> similarity. That’s a core of being social, & is inherited. The outcome
B> obviously isn’t, I said it many times.

Well, so I've written about empathize as conditioning by similarity, imitation and other social concepts like 8 years ago... It seems the entire Cognitive Algorithm is a sort of “conditioning by similarity” - matches to a template are reinforced.

B> it started as a vehicle for “selfish gene”: similarity is the only way animals can recognize genetic affinity, & then act to preserve & propagate common genes.

To me this sounds like finding purpose in something that is a by-effect. Many animals *cannot* see themselves, they see *other* animals (in the earliest age – their mother and siblings) and associate/condition their own feelings and sensations with the ”motion pictures” and smells of the other animals, they are fooled by each-other. Also there are cases of different kinds of animals trying to mate, even cats and dogs, or mammals and non-flying birds. :) In normal conditions they just don't meet and I guess there are simpler smell-driven means to recognize genetic affinity.

Human babies also can't see their own face for a long time, they see mother's and others' faces and probably find correlation between their own facial-muscles and visual feedback.

I agree imitation is fundamental, but “my” purpose is different. It's drive of evolution, higher level systems/virtual universes/”control units” are aiming at mimicking/repeating principles of operation of lower levels – because lower levels are example of working system for them and it's the lower level that *builds* the higher. Mind is a universal simulator of virtual universes, it's a “Multiversor”, it simulates what you feed it, but it can't do it exactly, can't cover all combinations; higher the level, lower the resolution, higher compression and match (generality). Also older versions of brains could have been just “Universors” - simpler simulators and recorders.

Mind imitates also, because record and replay (even with some mapping and adjustments) is still way simpler than inventing everything from scratch, and *repetition* is one of the basic means to detect and define patterns, “imitate/simulate” and “pattern” are correlated.


B>I though you realize that egotism is meaningless concept, the real question is
B>how broad & abstract the “ego” is. And conditioning by similarity increases
B> both.

I don't think “egotism” is a meaningless concept, it's “selfish gene” related, but it's a kind of by-effect. Mind cannot do anything but what its “program” drives, otherwise it's under other mind control, then it is another mind. Altruism are selfish drives which *match* others' selfish drives, so they are “cool”, because they match *your* drive (drive valuable to you), which reinforces selfishness.

Conditioning by similarity could be extremely general. For example when I discussed with my students about AGI and ethics, I asked them:

- What are you going to do if you're in the middle of a battle, and your fellow warriors are thinking machines (even they don't need to be androids), and you're attacked by humans who shoot at you and are trying to kill you? Will you empathize “humans”, because “they are like you”, “have soul” etc. or you will embrace your war-machines and robot friends who are covering you with their chests? Would they be savage heartless machines if they kill people who are trying to kill you, aren't people more heartless and savage?

It is similarity of drives/goals that matters above, and actually people do empathize *non-animate* objects as well (in biological and in dynamics/complexity sense), even in normal, non-war conditions - such as cars, computers or just dolls; people love them, call them names, hug them, kiss them, cry if they loose them. The objects just have to be associated with something related to satisfaction – you use this object often, it's an extension of your limbs, muscles, brains etc. so you associate it with “the image of yourself”.

B>The broadest inherited drive appears to be empathy / altruism.

Ironically I sort of agree, because I believe egoism is fundamental and altruism is worse than genuine egoism, a stronger form. :)

B>the real question is how broad & abstract the “ego” is

The broadness and abstractness of ego is subjective and questionable and it can be mind fooling. A serial killer may believe he was sent by God to punish sinners, his “self” covers The Omnipotent and he may believe he's doing good to humanity, because he will sent them to Heaven. Politicians, Green Peace, Church, Corporations - they all believe they work “for the good of humanity” or "society", which often actually means "to widen the range of their power and/or to replicate as much as possible".

Report abusive comment
Posted by Todor Arnaudov, last edited Oct 29, 2010 5:49 AM
> Well, so I've written about empathize as conditioning by similarity, imitation and other social...
Yeah, & a warehouse of videotape has all that info too. It’s the signal-to-noise that matters.

>It seems the entire Cognitive Algorithm is a sort of “conditioning by similarity” - matches are reinforced.
Cognition is necessary for conditioning, but not vice-versa, cognitive template can be value-free.

>Many animals *cannot* see themselves, they see *other* animals...

OK “recognize genetic affinity” is contrived. I realized that proximity & similarity are both criteria for instrumental conditioning, - by external & internal association. Empathy is conditioned similarity on the highest (for animals) level of identity. Social animals may have specific circuits that mediate it, by directly connecting the highest levels of somatosensory cortex to those of sensory cortex: temporal/parietal. These circuits probably have oxytocin receptors, & may be disrupted in some psychopaths & autistics.

>I agree imitation is fundamental, but “my” purpose is different. It's drive of evolution...

“Mind”, “virtual”, “universe”, “control”, unit”.., - holistic terms, lazy thinking, no progress.
This is so wrong & vacuous, I am getting an allergy. The “drive” is preservation / propagation of whatever forms capable thereof. “Imitation as a purpose”: http://en.wikipedia.org/wiki/Memetics, is based on psychological proclivity to imitate + mindless extrapolation of genetics. For you, it’s a mindless extrapolation of programming, - even less relevant. Replication is a purpose in cellular biology, not in cognition, & definitely not in behavior. Behavior evolved as a high-level *adaptive phenotype* for genetic reproduction, not vice-versa. Cognition exists to *modify* behavior in response to environmental changes, by maximizing *predictive correspondence*, not by replicating the data. Cognition doesn’t copy & “simulate” the same shit over & over just for the hell of it. Even in the most basic form, it’ll “forget” stimuli as soon as they stop being predictive. Representation is *not* replication, - selective abstraction is *the* difference between higher & lower phases of metaevolution. You’ve read this knol zillion times & still don’t get the most basic point.

>Mind imitates also, because record and replay (even with some mapping and adjustments) is still way simpler than inventing everything from scratch, and *repetition* is one of the basic means to detect and define patterns, “imitate/simulate” and “pattern” are correlated.

Just because “replay” is simple doesn’t mean it has any intrinsic value for the “replayer”, repetition can only come *after* recognition, simulation is for prediction, & experimenting by imitation is a ludicrously expensive way to do it. Will you ever grow out of your teenage “virtual universe”?

>I don't think “egotism” is a meaningless concept, it's “selfish gene” related, but it's a kind of by-effect. Mind cannot do anything but what its “program” drives, otherwise it's under other mind control, then it is another mind. Altruism are selfish drives which *match* others' selfish drives, so they are “cool”, because they match *your* drive (drive valuable to you), which reinforces selfishness. I believe egoism is fundamental and altruism is worse than genuine egoism, a stronger form. :)

Incoherent.

B>the real question is how broad & abstract the “ego” is
>The broadness and abstractness of ego is subjective and questionable and it can be mind fooling...

Right, don’t get too abstract if you aren’t smart enough, & don’t get too smart unless you have the guts to stand alone.

Report abusive comment
Posted by Boris Kazachenko, last edited Nov 2, 2010 11:54 PM
B>Social animals may have specific circuits that mediate it, by directly connecting the highest levels of somatosensory cortex to those of sensory cortex: temporal/parietal. These circuits probably have oxytocin receptors, & may be disrupted in some psychopaths & autistics.

OK.


B>You’ve read this knol zillion times & still don’t get the most basic point.

Well, you've read my real "knols" about this *NULL* times, but you're convinced that you get my point all correctly and I'm just an idiot who doesn't get even the basic premises of yours.
I guess the rest of the thousands of readers have understood everything right, so that's why they have nothing different to say.

By saying things that "doesn't" you imply I say "they do", while I'm not sure where exactly I've done.


B>Just because “replay” is simple doesn’t mean it has any intrinsic value for the “replayer”, repetition can only come *after* recognition

Yes. Everybody imitates/replay. The worst "replayers" replay literary input data, because they can't find patterns, patterns are too complex for their mind.
The best "replayers" replay the patterns they discover/recognize in data, the more advanced mind, more abstracted patterns it replays.

>simulation is for prediction

Of course it is, "virtual universes" function is to predict.

Report abusive comment
Posted by Todor Arnaudov, last edited Nov 7, 2010 1:58 PM
>Well, you've read my real "knols" about this *NULL* times, but you're convinced that you get my point all correctly

We've been through this quite a few times, Todor. I don't think you really get your own point either, otherwise you would use terms that actually mean something.

> and I'm just an idiot who doesn't get even the basic premises of yours.

It's not enough to be not-idiot, & "basic" doesn't mean easy to understand.

>I guess the rest of the thousands of readers have understood everything right, so that's why they have nothing different to say.

I think ~half of the hitcount on this knol is you & me. The problem is not that you disagree. It is that you don’t understand enough to disagree, - you can’t translate your terms into mine.

>By saying things that "doesn't" you imply I say "they do", while I'm not sure where exactly I've done.

You said "replaying" is the purpose of evolution, in my terms that would mean a conserved core. I explained that, even if you really stretch "replaying", that would only be true for basic reproductive phase, not for conditioning, let alone cognition. But that seems to be too analytical for you. Your "terminology" seems designed to mean anything, to make sure you're never wrong ("virtual universe" is a perfect example). But that means you're never right either. You're commenting on this knol, & seem to like it, so you need to explain how your opinion is different from mine.

>Yes. Everybody imitates/replay. The worst "replayers" replay literary input data, because they can't find patterns, patterns are too complex for their mind.
The best "replayers" replay the patterns they discover/recognize in data, the more advanced mind, more abstracted patterns it replays.

You need to stop talking about “everybody” & “everything”, or “mind” & “universe” - that’s infantile. Imitating & replaying are “macro” terms that suggest & encourage lazy analysis. If I replace them with input or pattern projection, then that’s only a dispensable tool for cognitive prediction. The purpose (the drive, the core, the criterion) is future correspondence. And behaviour is driven by motivation, in which imitation is only a minor element. Thus translated, I don’t see how this adds anything to my interpretation of cognitive phase, aside from putting the carriage in front of the horse. But of course, you’re not even talking about cognitive phase, or anything definable. Reality is too much of a constraint for a free artist.

Report abusive comment
Posted by Boris Kazachenko, last edited Nov 8, 2010 7:31 PM
Something you miss about programmers is that they know that every concept must be defined before used. I don't think my terms mean everything, discussion is of course more precise in the works and terms are derived from each other and are more advanced forms of the previous one.
Mind is built of virtual universes (simulators, cognitive hierarchies), they are built of control units (basic input-output-memory-compute-comparison-predictors).

>you need to explain how your opinion is different from mine.

I explained why I disagree about altruism, which for you is fundamental, while I think it's a spin-off and mutual ego-fooling.

There is something related to broadness of self in my POV, though, it's that the more evolved control units have collected the more complexity/algorithms/principles of the operation of universe
and can predict in wider range and higher precision and this is sort of objective measure of the "broadness of their self".

>You said "replaying" is the purpose of evolution, in my terms that would mean a conserved core.

Imitation and replaying as drives of evolution are not in that sense like in reproduction, it doesn't imitate and replicate itself. The ultimate trend in all times is building of "control units" which do predict at maximum precision and range input data (model real universe, which for a higher level virtual universe is a virtual universe), using parts already accumulated in themselves.

Perfect prediction is finally perfect simulation of lowest level laws of physics of the lowest level input from "reality", this is trend towards "omnipotence".

You just mention "technological phenotype", but cognition is a tool for it. Obviously Universe evolves means for higher precision repeatable access and "voluntary" manipulation of ever lower constituents, which is not higher generality, but higher specialty.

(Voluntary ~ representation of the target effect in the memory of control unit matches the real effect in the output environment as close as possible, a kind of maximizing correspondence).

Even computer games, CGI, filmmaking display this trend - ever better and literal simulation of "reality", i.e. the "texture" of the patterns at the lowest level inputs mind perceives to be replicated and mastered completely.

Report abusive comment
Posted by Todor Arnaudov, last edited Nov 9, 2010 2:16 PM
You don't understand the concept of conserved core. Cognition is not a means to create technology, it's the other way around. Cognitive algorithm is conserved & maximized, technological phenotype is changing at accelerating pace. Core is about stability, not shear volume. We create more precise representations of meaningless fluff because it doesn't cost anything anymore, not because they get more important. Just because proportion of junk in DNA is 95% & rising doesn't mean adding more junk is the purpose of evolution, - that junk is far less stable than the functional 5%.

Meaningless fluff aside, all you're talking about is cognitive algorithm. There's a knol for that, but you don't comment there because your theaterminology would be way out place.

Report abusive comment
Posted by Boris Kazachenko, last edited Dec 14, 2010 10:09 PM
I see, you say that actually I do recognize, project or search for trends in cognition, incrementing precision is because of self-improvement and accumulation of complexity. However I want to see "metaevolution" and see effects of development and improvement of cognition as part of metaevolution, I abuse "computerese" terms (and the knol), etc.

Also I see that your explanations are centered on the conserved core, if anything could drive or build something, it is it and it's for its "good", therefore it should be important for it. "I shouldn't think holistically", "what I'm talking about does not exist for real", etc.

Fine, I shouldn't abuse your knol, but I don't think my metaphysics is complete junk and that there is not a relationship between cognition and trends in Universe, "my theory" intended to find such.

Report abusive comment
Posted by Todor Arnaudov, last edited Dec 17, 2010 9:54 AM
> Also I see that your explanations are centered on the conserved core,

Without a conserved core, there is no identifiable system.

> Fine, I shouldn't abuse your knol,

It’s not an abuse, I just don’t have much patience for distractions. And frankly, this whole knol is mostly a distraction.

> but I don't think my metaphysics is complete junk

It’s not enough to get *something* right, a theory is worthless if it’s worse than competition. And the latter includes real sciences, which you’re not sufficiently familiar with.

> and that there is not a relationship between cognition and trends in Universe, "my theory" intended to find such.

That’s what I was doing in this knol, but the relationship I found seems to be too abstract for your taste. You want a relationship you can write a story about, & impress a girl with. Tough shit, the “universe” doesn’t care.

Report abusive comment
Posted by Boris Kazachenko, last edited Dec 18, 2010 7:11 PM
>Without a conserved core, there is no identifiable system.

I also have cores and identifiable systems. But it's better to write about it on my own territory, if ever do.

>You want a relationship you can write a story about, & impress a girl with. Tough shit, the
>“universe” doesn’t care.

Your sense of humor is appreciated. :) First of all I've not really worked on this since the past (except to leave comments), second the girls who're really impressed of stuff like this have high probability of personality and emotional issues - according to my experience: 100% and this is not surprising, because normal girls are definitely not interested in megalomanic Universal writings of mad scientists.

Report abusive comment
Posted by Todor Arnaudov, last edited Jan 12, 2011 4:34 AM
>I also have cores and identifiable systems. But it's better to write about it on my own territory, if ever do.

Right, & it must be *your own* theory, your self-esteem depends on it. Too bad you can't have your own universe.

>normal girls are definitely not interested in megalomanic Universal writings of mad scientists.

But that won't keep you from trying.

Report abusive comment
Posted by Boris Kazachenko, last edited Jan 12, 2011 7:06 PM
>Right, & it must be *your own* theory, your self-esteem depends on it. Too bad you can't have your own universe.

You could just ignore me or delete comments, it's your territory, you should keep it safe. I've got lots of fields of interest to boost my self-esteem.

> *your own* theory, your self-esteem depends on it.

No, my own territory in order not to spam yours and to stop bothering you, and because I don't care that much for your (or anyone's) approval - I am convinced I'm right in my POV and it'd work if it's formalized; and I was interested in effects you're not.

I'm tired to try to fit things to yours, to explain (what's worth if you understand or agree?), and of all those fragmentary discussions. Finally I've always known such stuff won't worth until it's formalized or made "running". Your readers seem not to like your meta-evolution either, according to the average rating.

>But that won't keep you from trying.

I prefer to use my art.
 

Issues about biological evolution

Hi Boris, I like your articles and your version of the metaevolution. Evolution is a great concept, I'm exploiting it in my version of metaevolution and theory of intelligence, too. Regarding the part after humans did exist, and intelligence as internalized evolution/accumulation of complexity etc. - this sounds great and is solid.

However, regarding biological evolution, there are issues, what is your explanation of this/do you care?

(How about - biological evolution as a concept, created in order eventually to suggest and give rise to the invention of artificial intelligence... :) )

You might be better in genetics sciences than me, but the following issues easily arise, even if you're not a specialist in genomics/genetics and stuff:

- In order a positive mutation to reproduce, a male and a female should mutate synchronously in a compatible matter, and they have to mate. Now there are 6 billions humans, there were tenths of billions in the past centuries, and it doesn't seem that such mutations did occur. (Correct me, if wrong).

- If a mutation is small, it could give an insignificant advantage, and eventually disappear, because there are no compatible individuals for reproduction. As bigger is the change, as lower the probability of a male/female to mutate in parallel and to mate.

- Even a big mutation, say higher intelligence with 15-20 IQ points in todays measures, could not bring a significant advantage in the wild proto world. Einstein would hardly invent the General Relativity there.

- How natural language was born, including the vocal system - how and why it developed. How does the use/disuse of an organ is turning to DNA - there is no feed-back, except reproduction.

- Assume two synchronized mutants manage to mate and produce an offspring. However, who their offspring will mate afterwards? Therefore, evolution needs a row of parallel compatible mutations of big enough populations, and they should be in a close enough area in order to mate.

- There could be number of evolution lines with similar human-level intelligent beings of different biological kinds, but there is just one kind.

- How does the number of chromosomes change between the mother and a child, without resulting in a still-born or infertile child? What about the milk of the mother and his baby of a new kind?

They say "millions of years pass", to me this is a cheap excuse - regarding proto-human part, there was not so much time, and the population have been pretty small.

DNA seems to be so messy and "spaghetti-coded", in my opinion the probability of producing positive advantageous and accumulating mutations randomly, as errors, in parallel etc. is very questionable.

I know, you actually do mention something related to this, that the bacterias are changing DNA faster and humans seem genetically uniform.


My hypothesis, if biological evolution did really happen, is that DNA of the previous organisms was designed to follow a series of changes, maybe including chromosome number changes, until reaching to stable forms as we know them today. There might be "attractors", where DNA has been aiming to.


Regards
Tosh

Last edited Mar 8, 2010 5:05 PM
Report abusive comment
Hi Todor!

>> In order a positive mutation to reproduce, a male and a female should mutate synchronously in a compatible matter, and they have to mate.

That's wrong, I don't know where you got it from, look up "sexual reproduction" in wikipedia.

>> Now there are 6 billions humans, there were tenths of billions in the past centuries, and it doesn't seem that such mutations did occur. (Correct me, if wrong).

There'd been plenty of genetic changes even very recently, check "Ten thousand year explosion".

>> Even a big mutation, say higher intelligence with 15-20 IQ points in todays measures, could not bring a significant advantage in the wild proto world. Einstein would hardly invent the General Relativity there.

The biggest application of human intelligence is in facilitating complex social cooperation.

>> How natural language was born, including the vocal system - how and why it developed. How does the use/disuse of an organ is turning to DNA - there is no feed-back, except reproduction.

It wasn't "born", it developed gradually. Use/disuse doesn't have to change DNA, epigenetics can be heritable too.

>> There could be number of evolution lines with similar human-level intelligent beings of different biological kinds, but there is just one kind.

There's always "one" at the top, even as the top keeps raising.

>> How does the number of chromosomes change between the mother and a child, without resulting in a still-born or infertile child? What about the milk of the mother and his baby of a new kind?

That's is inane, you should read before you write.

>> DNA seems to be so messy and "spaghetti-coded", in my opinion the probability of producing positive advantageous and accumulating mutations randomly, as errors, in parallel etc. is very questionable.

I think there have been minor genetic changes in humans that produced a major increase in intelligence, mostly through the growth & folding of neocortex. The genetic changes that produced civilization mostly relate to predisposition to learn by immitation. Our intelligence is social, the main inheritance mechanism here is culural conditioning.

>> My hypothesis, if biological evolution did really happen, is that DNA of the previous organisms was designed to follow a series of changes, maybe including chromosome number changes, until reaching to stable forms as we know them today. There might be "attractors", where DNA has been aiming to.

That would the dumbest possible way to implement a "design".
And the forms "now" are less stable than ever.

Boris

Report abusive comment
Posted by Boris Kazachenko, last edited Mar 6, 2010 7:31 PM
Good, you did care. :) The evolution after intelligent beings existed is fine.

>I think there have been minor genetic changes in humans that produced a major increase in >intelligence, mostly through the growth & folding of neocortex

Nice.

>> How natural language was born, including the vocal system - how and why it developed. How does the
use/disuse of an organ is turning to DNA - there is no feed-back, except reproduction.
>It wasn't "born", it developed gradually. Use/disuse doesn't have to change DNA, epigenetics can be heritable too.

We weren't born as well, we were "developed gradually" in our mothers, but we say "born" anyway, implying a significant point in the "development" for the given context. Anything wasn't "invented" as well, it was developed gradually for anyone who knows the history of science/engineering/an invention, but there are important moments anyway.

>> My hypothesis, if biological evolution did really happen, is that DNA of the previous organisms was designed to follow a series of changes, maybe including chromosome number changes, until reaching to stable forms as we know them today. There might be "attractors", where DNA has been aiming to.
>That would the dumbest possible way to implement a "design".

Evolution or development is a form of unfolding and decompressing, the "design" is implied from the birth of the Universe, unless you do believe in true randomness.

If you quote parallel universes - if any possible thing happens (what a waste), but in a separate universe that does not affect the other ones, this is the same as if just one of the possibilities happen without parallel universes, it does not explain why anything did happen exactly this way in this particular universe.

We were "prenataly developed" as well, and it may seem like the evolution at a fast pace, however this development can be perfectly called also a decompression and unfolding, and the process passes through significant points, attractors.

Cognitive development seems also as "evolution", and it can also be called decompression - it is impossible for the system to encode itself in the biological part, so the "program" is compressed and can be decompressed by an appropriate set of stimuli. It develops, but the phases are predictable and predetermined, it's like an in-place sorting algorithm which has to pass all iterations.

And obviously there are DNA combinations and sets of combinations that that produce functionally similar systems, which are more stable than others; and I'm not talking about a change in a gene number N that alters a protein ABC to EFG.

T

Report abusive comment
Posted by Todor Arnaudov, last edited Mar 7, 2010 12:56 PM
>> Good, you did care. :) The evolution after intelligent beings existed is fine.

I don't know if I care enough. Your terminological mix-ups suggest that you're after rhetorical games rather than understanding.
Just because ontogenetic development (look it up) partly follows genetic "design" doesn't mean that phylogenetic evolution does too. If phylogeny had a "design" to follow, it would be as fast as ontogeny (which is millions times faster). Yes, you can say that any initial condition is a "design" for whatever follows, but that's an abuse of language. Design is the compressed part of initial condition. Genetic design determines part of phenotypic complexity, though that part gets relatively smaller for higher organisms.
Yes, phylogeny has that part too, - the genome of whatever species you care to start from, but it determines a decreasingly significant percentage of complexity for descending species. In fact, phylogeny is mainly a compressive process, - it evolves increasingly compressed genotypes.
That's even more the case for cosmology,- universal physical constants (design) define an infinitesimal part of complexity in modern universe. The rest is determined by, depending on a theory, either quantum fluctuation or initial configuration of particles. Neither of these is compressed, so there's not much to "unfold".
So, what you say is not literally wrong, just meaningless. Biology & Cosmology are real sciences (as distinct from AI), you need to study them before you start having "theories". In AI, there's not much worth studying, but you still need to make sense as a full-time job, unless you want to be just another blind tinkerer.

>> Cognitive development seems also as "evolution", and it can also be called decompression - it is impossible for the system to encode itself in the biological part, so the "program" is compressed and can be decompressed by an appropriate set of stimuli. It develops, but the phases are predictable and predetermined, it's like an in-place sorting algorithm which has to pass all iterations.

You got it exactly backwards, cognitive development is *compression* of the stimuli, & it's only as predictable as its environment. Prediction is what it's for, cognition that's easy to predict is out of job. Prenatal development of the brain is, largely, ontological "decompression" of corresponding genes, but that's before cognition even starts.
I did call it "internalized evolution", but that's a loose analogy, cognition is as different from phylogeny as the later from the growth of entropy (I bet you didn't understand that part).
Do me a favor, think about it for few days before replying. It seems to me that you're more interested in wining an argument than in understanding the issues.

Boris.

Report abusive comment
Posted by Boris Kazachenko, last edited Mar 7, 2010 10:14 PM
>Do me a favor, think about it for few days before replying. It seems to me that you're more interested in wining an argument than in understanding the issues.

Regarding rhetorics and human behavior - the same goes to you. Usually humans do a shallow search for the semantics of other people talking. You don't search for interpretations which are meaningful and don't assume that you don't understand the other, but you assume that the other is talking what you expect, but he's just wrong and dumb.

>If phylogeny had a "design" to follow, it would be as fast as ontogeny (which is millions times faster).

I don't think so. If you want to build today's 6 billions of humans (with their precise molecular structure, allowing their actual cognitive capabilities) and all their technology, and all the history that produced it, and all processes are interrelated and everything should happen despite the interrelation and because of them.

And you're starting out from a big melted stone, and some stars and planets around - it's perfectly understandable that the compression rate is huge and needs many many computations and transformations and search, even if it's a search, destined to find.

Why reproduction across space happens for evolved entities? Оne reason can be because you cannot encode 6 billion really different and as complex individuals and all around with that spaghetti code - not DNA, the melted stones, a star and some other stones around + those physical laws.

>You got it exactly backwards, cognitive development is *compression* of the stimuli, & it's only as
>predictable as its environment. Prediction is what it's for, cognition that's easy to predict is out of job.

Of course I'm not talking about this, unless you assume I'm an idiot - but I know you do, OK. :)

Regarding the compression, this is from my own article, see the last word, I was 17 then, "...При човека, повечето обекти (от всякакъв вид) не се запомнят "фотографски", а се "преразказват" в мозъка, записват се най-характерните особености на информационните обекти, входната информация се *компресира*."

Improving the precision and range of predictions and intentional causality (control) is a trend of the evolution in my theories, prediction and compression are linked, humans (high level) compress, whereas the lowest level of Universe does not compress and predict, it cannot envision what will happen, it does things, but Universe "predicts" with probability 1 and max possible resolution what will happen next - that's what "virtual universes" at different levels are aiming to achieve.

But I was talking about predictability of the stages of cognitive development of an individual.


>I did call it "internalized evolution", but that's a loose analogy, cognition is as different from phylogeny
>as the later from the growth of entropy (I bet you didn't understand that part).

:))

It is a decompression, if you look on the Universe as the memory of a computer, and the important thing is the end result data, the final system with its exact physical structure, not abstract/compressed.

Anything needs to be encoded with the "machine language" before it can run. Every abstract core have to be decompressed to physics before it is capable to work in this Universe, each lower preserved core is an interpreter down to the physical level. And this "decompression" goes from the implicit memory of Universe to the explicit memory of Universe (what actually does exist).


>In AI, there's not much worth studying, but you still need to make sense as a full-time job,
>unless you want to be just another blind tinkerer.

Check the feeds in my blog - I know you won't, but you're talking bullshit, since I will teach AGI in the University, and aiming at creation of a start-up, not that that latter would be easy. The original course includes an overview of the AI and its mistakes, brain architectures/biology/developmental psychology, human behavior and all the mathematics, machine learning and philosophy that are required for a mind to reach and explain theories of intelligence and Universe such as Hawkins', Hutter's, yours and even my meaningless one. ;)

T

Report abusive comment
Posted by Todor Arnaudov, last edited Mar 8, 2010 2:55 PM
Sorry, I want to change this reply, it was way too harsh.

I appreciate that we agree on many issues, that's rare enough. But it's the disagreements that matter, & I disagree that your interpretation "compresses" or "predicts" anything. Anyway, metaevolution is just a hobby for me, & obviously for you. So, it's perfectly fine that you don't know much about biology, as long you focus on the stuff that matters.
I do get feeds from your blog (via bloglines), & teaching AGI course sure beats the show business. But I feel that there's too much of a show business left in your postings. And I see no indication that you're doing any theoretical work, which is the kind that matters in this field. I would be delighted to be proven wrong.

Boris.

Report abusive comment
Posted by Boris Kazachenko, last edited Mar 10, 2010 9:14 PM
OK, аccepted, I'm sorry too.

>But it's the disagreements that matter, & I disagree that your interpretation
> "compresses" or "predicts" anything.

OK

>Anyway,metaevolution is just a hobby for me, & obviously for you.
>So, it's perfectly fine that you don't know much about biology,
>as long you focus on the stuff that matters.

I think an issue in my metaevolution is that while it is somehow about metaevolution, it includes a lot of metaphysics, philosophy and digital physics/systems theory/control theory/cybernetics. Mine was somehow an early attempt of a "digital metaphysical theory of everything", it searched for unified concepts which are applicable to many systems and display trends, e.g. why there are computers, why they work the way they work and how/why are they similar to entities such as societies/political states and the Universe, there are shared principles, and computers are closer to a formal demonstration of the principles.

Regarding biology, actually I do study some biological stuff, also biochemistry sometimes, but not too often and not too much; mostly neuroscience. My bad that I didn't check for a minute some of the genetic issues, but was just speculating aloud.

>I do get feeds from your blog (via bloglines), & teaching AGI course sure beats the show business.
>But I feel that there's too much of a show business left in your postings. And I see no indication
>that you're doing any theoretical work, which is the kind that matters in this field. I would be
>delighted to be proven wrong.

Thanks for watching! I do some theoretical work, but it is not that constructive yet, it's preparatory - studying, translating to-from English, making up the program for that course and preparing lecture slides. A warm up, yet, but it takes time and I have to do it.

You're right (unfortunately or not) that show business is still there, also the pure one - such as Video and Music, but recently the latter is hidden in its own dedicated blog...

Todor