Re: [Fis] INTELLIGENCE INFORMATION (by Y.X.Zhong)
Dear Colleagues, Relating information with intelligence seems to me important for several reasons. I will try to suggest that intelligence might be a good conceptual tool if we want to anchor our understanding of information and knowledge in the natural world. Yixin mentions the problem of three approaches to AI which exist independently, based on the methodological doctrine of divide and conquer. We agree that divide and conquer is just not enough, it is the movement in one direction, and what is needed is the full cycle -bottom up and top down - if we are to understand biological systems. The appropriate model should be generative - it should be able to produce the observed behaviors, such as done by Agent Based Models (ABM) which includes individual agents and their interactions, where the resulting global behavior in its turn affects agents' individual behavior. Unlike static objects that result from a divide and conquer approach, agents in ABM are dynamic. They allow for the influence from bottom up and back circularly. Central for living organisms is the dynamics of the relationships between the parts and the whole. Shannon's theory of communication is very successful in modeling communication between systems, but it is a theory that presupposes that communication exists and that mechanisms of communication are known. On the other hand if we want to answer the question why those systems communicate at all and what made them develop different mechanisms of communication we have to go to a more fundamental level of description where we find information processes and structures in biological systems. Natural computation such as described by Rozenberg and Kari in The many facets of natural computing http://www.csd.uwo.ca/~lila/Natural-Computing-Review.pdf includes information processing in living organisms. Generative models of intelligence may be based on info-computational approach to the evolution of living systems. Three basic steps in this construction are as follows: . The world on its basic level is potential information. (I agree with Guy on his information realism) . Dynamics of the world is computation which in general is information processing (natural computationalism or pancomputationalism) . Intelligence is a potential for (meaningful) action in the world. (I agree with Josph) The minimal claim would be that there is no intelligence without information. For an agent, intelligence is the ability to face the world in a meaningful way and it increases with the number of different ways an agent is able to respond with. (This is a statistical argument: in a dynamical world, ability of an agent to respond to a change in several different ways increases its chances for survival.) Back to the question of Raquel: can a simple organism be ascribed intelligence? - which Pedro suggests to answer in the positive by broadening the concept of intelligence. I agree with this proposed generalization for several reasons. Maturana and Varela conflate life itself with cognition (to be alive is to cognize). Similarly, we can connect the development of life (towards more and more complex organisms) with intelligence (if an organism acts meaningfully in the world, we say it acts intelligently; meaningfulness has degrees and so has intelligence). In that approach intelligence would be the property of an organism which gives it a potential to develop increasingly more complex informational structures and increasingly more complex (meaningful) responses to the environment. One can argue that increasing the repertoire of meaningful responses (interactions with the world) increases agents potential for survival and success. As a consequence this approach makes way for a basic quantitative measure of intelligence as a level of complexity of an organism providing the diversity of its responses.( Of course this measure of intelligence is not in the sense of IQ or specific individual's smartness but of the species increasing capability to flourish.) This view also agrees with the understanding that even in humans there are several different intelligences - linguistic, logical, kinesthetic, naturalist, emotional, interpersonal, intrapersonal, spatial, musical, etc. If the complexity of the information processing structures and diversity of interactions with the environment are the measure, then plants and by the same token even single cells may qualify as intelligent in the sense of naturalist and kinesthetic intelligence. In sum, there are different ways to define intelligence and information dependent on what we want them to do for us. Concepts are tools used by theories. Theories are tools used by people. Many different concepts address different aspects of the world and seem to fill their purpose. From an info-computational approach we may hope to provide a base for the construction of generative explanatory models for the development of
Re: [Fis] INTELLIGENCE INFORMATION (by Y.X.Zhong)
Kumiko Tanaka-Ishii has recently written an excellent little book, The Semiotics of Programming, which may be of interest in connecting semiosis with machine-like processes like that of computation. http://books.google.com/books?id=irizHa1MXJoCsource=gbs_navlinks_s Best, Jacob On 11/13/2010 2:02 PM, Stanley N Salthe wrote: Concerning: The minimal claim would be that there is no intelligence without information. For an agent, intelligence is the ability to face the world in a meaningful way and it increases with the number of different ways an agent is able to respond with. It seems to me that this implies, in any non-mechanistic system, semiosis -- that is to say, a process of interpretation by the agent. Thus, intelligence would be related to the viewpoint of the agent, which would be located by its needs. Semioticians, however, have not been much engaged by this concept. Hoffmeyer claims that it is especially a social skill. STAN On Sat, Nov 13, 2010 at 4:18 PM, Gordana Dodig-Crnkovic gordana.dodig-crnko...@mdh.se mailto:gordana.dodig-crnko...@mdh.se wrote: Dear Colleagues, Relating information with intelligence seems to me important for several reasons. I will try to suggest that intelligence might be a good conceptual tool if we want to anchor our understanding of information and knowledge in the natural world. Yixin mentions the problem of three approaches to AI which exist independently, based on the methodological doctrine of divide and conquer. We agree that divide and conquer is just not enough, it is the movement in one direction, and what is needed is the full cycle -bottom up and top down - if we are to understand biological systems. The appropriate model should be generative - it should be able to produce the observed behaviors, such as done by Agent Based Models (ABM) which includes individual agents and their interactions, where the resulting global behavior in its turn affects agents' individual behavior. Unlike static objects that result from a divide and conquer approach, agents in ABM are dynamic. They allow for the influence from bottom up and back circularly. Central for living organisms is the dynamics of the relationships between the parts and the whole. Shannon's theory of communication is very successful in modeling communication between systems, but it is a theory that presupposes that communication exists and that mechanisms of communication are known. On the other hand if we want to answer the question why those systems communicate at all and what made them develop different mechanisms of communication we have to go to a more fundamental level of description where we find information processes and structures in biological systems. Natural computation such as described by Rozenberg and Kari in The many facets of natural computing http://www.csd.uwo.ca/~lila/Natural-Computing-Review.pdf http://www.csd.uwo.ca/%7Elila/Natural-Computing-Review.pdf includes information processing in living organisms. Generative models of intelligence may be based on info-computational approach to the evolution of living systems. Three basic steps in this construction are as follows: . The world on its basic level is potential information. (I agree with Guy on his information realism) . Dynamics of the world is computation which in general is information processing (natural computationalism or pancomputationalism) . Intelligence is a potential for (meaningful) action in the world. (I agree with Josph) The minimal claim would be that there is no intelligence without information. For an agent, intelligence is the ability to face the world in a meaningful way and it increases with the number of different ways an agent is able to respond with. (This is a statistical argument: in a dynamical world, ability of an agent to respond to a change in several different ways increases its chances for survival.) Back to the question of Raquel: can a simple organism be ascribed intelligence? - which Pedro suggests to answer in the positive by broadening the concept of intelligence. I agree with this proposed generalization for several reasons. Maturana and Varela conflate life itself with cognition (to be alive is to cognize). Similarly, we can connect the development of life (towards more and more complex organisms) with intelligence (if an organism acts meaningfully in the world, we say it acts intelligently; meaningfulness has degrees and so has intelligence). In that approach intelligence would be the property of an organism which gives it a potential to develop increasingly more complex informational structures and increasingly more complex (meaningful) responses to the environment. One can argue that increasing the
Re: [Fis] INTELLIGENCE INFORMATION (by Y.X.Zhong)
Gordana -- Interpretation of information builds more information, which again becomes interpreted. In living systems each generation makes a new interpretation based upon changed conditions of life. But in this case there is not more (genetic) information, but rather recently altered information -- history rewritten according to the latest interpretation of recent conditions. Some might call this process 'intelligence'. This is the (neo)Darwinian interpretation. It does not address your point about increasingly complex patterns of information, which is indeed what appears in the fossil record (as well as in human discourse). To build more requires preservation and interpretation. In the physical world, this image is captured in the asteroid impacts on the moon, with subsequent hits deforming, but not erasing, the original one. Information here increases, but not, I think, intelligence. Intelligence, I think, lies more in reinterpretation than in the building more that may follow upon it. STAN (Pedro -- this is a new week, so this is my first) On Sat, Nov 13, 2010 at 5:28 PM, Gordana Dodig-Crnkovic gordana.dodig-crnko...@mdh.se wrote: I suppose semioticians are interested in an individual human’s sense-making in a context of human society. Or perhaps a social animal’s sense making. What I think about is how life forms organize to produce increasingly complex patterns of information processing. Gordana *From:* fis-boun...@listas.unizar.es [mailto:fis-boun...@listas.unizar.es] *On Behalf Of *Stanley N Salthe *Sent:* den 13 november 2010 23:03 *To:* fis@listas.unizar.es *Subject:* Re: [Fis] INTELLIGENCE INFORMATION (by Y.X.Zhong) Concerning: The minimal claim would be that there is no intelligence without information. For an agent, intelligence is the ability to face the world in a meaningful way and it increases with the number of different ways an agent is able to respond with. It seems to me that this implies, in any non-mechanistic system, semiosis -- that is to say, a process of interpretation by the agent. Thus, intelligence would be related to the viewpoint of the agent, which would be located by its needs. Semioticians, however, have not been much engaged by this concept. Hoffmeyer claims that it is especially a social skill. STAN On Sat, Nov 13, 2010 at 4:18 PM, Gordana Dodig-Crnkovic gordana.dodig-crnko...@mdh.se wrote: Dear Colleagues, Relating information with intelligence seems to me important for several reasons. I will try to suggest that intelligence might be a good conceptual tool if we want to anchor our understanding of information and knowledge in the natural world. Yixin mentions the problem of three approaches to AI which exist independently, based on the methodological doctrine of divide and conquer. We agree that divide and conquer is just not enough, it is the movement in one direction, and what is needed is the full cycle -bottom up and top down - if we are to understand biological systems. The appropriate model should be generative - it should be able to produce the observed behaviors, such as done by Agent Based Models (ABM) which includes individual agents and their interactions, where the resulting global behavior in its turn affects agents' individual behavior. Unlike static objects that result from a divide and conquer approach, agents in ABM are dynamic. They allow for the influence from bottom up and back circularly. Central for living organisms is the dynamics of the relationships between the parts and the whole. Shannon's theory of communication is very successful in modeling communication between systems, but it is a theory that presupposes that communication exists and that mechanisms of communication are known. On the other hand if we want to answer the question why those systems communicate at all and what made them develop different mechanisms of communication we have to go to a more fundamental level of description where we find information processes and structures in biological systems. Natural computation such as described by Rozenberg and Kari in The many facets of natural computing http://www.csd.uwo.ca/~lila/Natural-Computing-Review.pdf includes information processing in living organisms. Generative models of intelligence may be based on info-computational approach to the evolution of living systems. Three basic steps in this construction are as follows: . The world on its basic level is potential information. (I agree with Guy on his information realism) . Dynamics of the world is computation which in general is information processing (natural computationalism or pancomputationalism) . Intelligence is a potential for (meaningful) action in the world. (I agree with Josph) The minimal claim would be that there is no intelligence without information. For an agent, intelligence is the ability to face the world in a meaningful way and it increases
Re: [Fis] fluctuons
Hi, I've been meaning to send a note on Kevin Kirby's brief outline of Conrad's fluction framework, but haven't had the time to compose my thoughts coherently. I realised that I wouldn't really have the time to do so, so I had better send something half-baked along anyway to contribute to the discussion. Kevin concludes his piece with the following remark: quoteOverall, within fluctuon theory the interaction between the manifest organism and its unmanifest vacuum sea image abets the evolution, persistence, and maintenance of this unique complexity [of life]. This is a fascinating and rich notion. What can we unfold from this notion now in 2010?/quote The way I see it, organisms are organisational units, and we tend to view genomic content as informational units. However, genomic identifiers are merely one way of providing information tags. Apart from the presence/absence of sequence, there is also the notion of the multiple/collective (to borrow Alain Badiou's language) -- collections of molecules that bear that signature. It is these collectives that comprise the dynamical state of cells and organisms, and the cardinalities of these sets may often be used as a proxy for snapshots of organismal state. This tells us that organisational units such as tissues may be characterised via such cardinalities -- liver cells and heart cells have different protein number distributions within the same organism yet protein distributions in liver cells are more similar across taxa. Hence the fluctuon concept may be viewed in this concept as the creation and annihilation of molecules following gene expression, or the transition into and out of active or inert molecular state, around the vacuum -- the steady state of an open dynamical network. The response characteristics of this proteomic or messenger RNA cloud and the entropy production (as measured in terms of fluctuating numbers around the steady state) offer dynamical proxies of the organism, extending the static snapshot. This becomes conceptually and mathematically accessible to perturbative ideas from quantum field theory, and the recasting of stochastic processes via Doi's 1976 work (Doi M (1976) Second quantized representation for classical many-particle systems. J Phys A: Math Gen 9: 1465–1477) has been used, for example, in Sasai M, Wolynes PG (2003) Stochastic gene expression as a many-body problem Proc Natl Acad Sci U S A 100(5): 2374–2379 to do that. Moreover, neutral evolution offers a landscape of adjacent vacua in the design space of possible gene expression clouds and their response characteristics. The protein identity matching test pointed out the significance of non-coding, regulatory sequences (King MC, Wilson AC, Evolution at Two Levels in Humans and Chimpanzees, Science1975, 188:107-16) indicating the necessity of moving beyond identifiers as (sole) information carriers and to what is now called evo-devo. The vacua in the fluctuon picture provides a way of characterising the landscape in this metaphorical spatial remapping of a historical process which register the dynamical responses of gene expression clouds of organismal, histological and cytological collectives at multiple-generational evolutionary time scales, with neutrality exploring the adjacent possible of these vacua, via alternative cis-regulatory underpinnings of dynamical states. This has been explored in the popular press by Gerhart and Kirschner in The Plausibility of Life -- they reference Conrad in there, to bring it back to where the discussion started. Cheers, Sri Srinandan Dasmahapatra School of Electronics and Computer Science University of Southampton Southampton SO17 1BJ Phone: +44(0)2380594503 s...@ecs.soton.ac.uk ___ fis mailing list fis@listas.unizar.es https://webmail.unizar.es/cgi-bin/mailman/listinfo/fis