self-reference paradox

Douglas Hofstadter - a computer scientist who authored a few interesting books most notably the Pulitzer prize winning work ‘Godel, Escher, Bach’, explores the phenomenon of strange loops which are observed in nature through a slightly computational lens. Imagine a system programmed with a certain set of rules that allows for useful computation for someone outside the system. The system could be as simple as a set of symbols which enable useful operations (such as 1+1=2) which have meaning for someone outside the system. Let us consider a hypothetical observer within in the system who is executing the operations by selecting and combining the necessary symbols by following a set of programmatic logical rules. The observer within the system is not required to understand the broader meaning of the performed operations in order to execute the said operations. This is pretty much what a computer does where a fixed set of CPU instructions can undergo vast possibilities of permutations and combinations through nested recursive operations by following a programmatic logical syntax to execute almost any complex computational operation which have meaning to an observer outside using the computer but not necessarily to the computer itself. This is however standard knowledge but what is interesting is the possibility of adding an additional set of rules within the system that allow some sort of assignment of broader meaning (similar to what an outside observer is capable of gleaning) to all underlying operations within the system which then allows the earlier hypothetical observer inside the system mechanically executing the operations within the system to now create abstractions and assign broader meaning to all the underlying operations being executed. If we are able to successfully create these additional set of rules that can algorithmically synthesize and assign meaning by different recursive combinations of the existing underlying instructions, the hypothetical observer within the system evolves from a mechanical mindless executor into an intelligent entity equivalent to the outside observer using the computer, now capable of understanding the broader semantics of the programmed syntax. The science of meaning synthesis is fascinating and Douglas Hofstadter uses a basic example to loosely explore how it be could be algorithmized. Considering a simple system of symbols, we know 1 + 1 = 2 is a meaningful operation for addition, if there is a combination of symbols that reads - p - q - - , at first glance it appears to be a meaningless combination of symbols but if we now create an additional rule that equates ‘-’ to ‘1’, ‘p’ to ‘+’, ‘q’ to ‘=’, the earlier meaningless statement becomes equivalent to 1 + 1 = 2 which is meaningful information thereby assigning some level of meaning to the statement ‘- p - q - -’. Meaning synthesis could then be possibly distilled into an art of recursive mapping of existing instructions within the system that result in the emergence of meaning within the system much similar to the meaning that can be gleaned by an intelligent observer outside the system. So, an evolved system capable of internally deducing meaning has two sets of instructions, the first set is typographical instructions which are syntactic rules that can be mechanically executed to serve a purpose without having to understand meaning and the second set of instructions called semantic instructions which help interpret the typographical instructions and assign meaning to it. Douglas Hofstadter uses the example of DNA to further explore this concept. DNA is essentially genetic code i.e. a set of biological instructions which happens to follow the structure of having both typographical instructions as well semantic instructions. DNA as covered earlier in a previous article ‘dissecting genetic code’ is ultimately biological code that is interpreted and executed by biological entities called ribosomes to synthesize proteins which make more copies of the DNA. What is interesting here, is the necessary biological information for creating ribosomes to help interpret the DNA is actually stored in the DNA itself. This makes DNA a very interesting and practical example of a self-referencing loop where meta-information present in the DNA itself helps interpret the primary information of the DNA to create proteins which can help create more copies of the DNA with time creating more sub-loops with each copy of DNA which more or less repeats the same process within each sub-loop essentially creating an emergent system of life. If the architecture of DNA can be applied to a system of primary computational instructions where you add additional set of instructions or meta-instructions to help interpret and assign meaning to the primary instructions, we create an emergent system of intelligence where the system within itself is able to synthesise meaning through different levels of abstraction from the various recursive and syntactic combinations of primary instructions. Since the system is able to synthesize meaning much like an outside observer who uses the system to do computations to serve the observer’s purpose, the system will be able to use itself to run different computations to serve a purpose that gets constructed by the intelligence that emerges from being able to derive meaning within the system. This is essentially what we see in human intelligence where if we apply a reductionist lense to the brain, it is just a set of primary instructions being mechanically carried out by seemingly simplistic and similar neuron firings. A set of meta-instructions encoded in the nervous system would provide a blueprint for the brain to perform high-level interpretation of the different patterns of neuron-firings where the emotion of anger would be a high level interpretation of a certain pattern of neurons firing whereas the emotion of sadness or a distinct past memory remembered would be a different high level interpretation of another pattern of neurons firing. Since, we are able to assign high level meaning to the neurons firing and experience abstract concepts, we can enter into a higher plane of abstraction to synthesize new and more complex concepts by connecting existing concepts which is essentially translated into new computation being computed by the system within the system itself to serve a certain purpose for the system. This is a highly computational view of what we popularly identify as consciousness. Consciousness is a paradoxical strange loop where a computing system by virtue of complex recursive combination of its basic elemental instructions enables higher-level computation allowing it to reference itself as a whole and provide autonomy for the computational system itself to direct its lower-level computation. Douglas Hofstadter highlights the strangeness of the consciousness loop by stating that the underlying complex computation which creates a sense of self can simulate only higher-level phenomenon like thoughts, emotions or memories but oddly cannot simulate more lower-level neuron firing underlying the different higher-level phenomenon.