Basics -- Reasoning
A fact is a relationship between 1) percepts or 2) other facts. A knowledge base is the set of facts that an entity knows about the world (hence the name). Facts can either be acquired from perception (atomic, or axiomatic facts) or inferred from existing facts (derived facts). The process of inferring new facts from existing facts is known as reasoning. I believe reasoning is the root of human conscious experience.
There are various ways of implemeting reasoning systems. One method is known as chaining, which can be of the forwards or backwards variety. Both methods are similar in that they can be used to generate new facts from existing facts, and in my opinion both are important.
For instance, we wish to determine if apples and blood are the same color.
Here is an example of forward chaining: suppose our knowledge base is something like {f1=apples are red, f2 = blood is red, f3=red is a color}. Then with f1 and f2 we can infer {f4=apples and blood are red} and then using f3 and f4 we can infer {f5=apples and blood are the same color}.
Here is an example of backward chaining: starting with {f1, f2, f3, f5} above we note that we can possibly generate f4 if f5 and f3 are true. Going with this choice of f4 (as opposed to “apples and blood are green” or any other color) we then see that f4 is indeed true due to either f1 or f2. Note that the algorithm would possibly cycle through other colors and have to back up because f1 and f2 only apply to red, etc.
You may be sorry you invited me here Rocketdodger.
May I ask what your background is and what is your job? I ask because I'm reading a bias in your views that sounds like you are involved in knowledge or case-based systems engineering.
I don't believe reasoning, at least as you define it, is the basis of consciousness. I would argue you could do everything you discuss without being conscious at all. In fact, AI's first big successes in the '50's were in solving the "hard" problems in creating reasoning engines. Pixi aside, were those systems conscious? It was when AI first tried to tackle the "easy" problem of perception that it fell on its face and still has never fully recovered from the hubris and shame.
I think to find the clues for the basis of consciousness you have to look at evolution, see what the simplest brains do and study evolving functionality. What simple and complex brains really do, what they have to do to survive is not essentially reasoning but modeling and prediction. Modeling and prediction subsumes reasoning with sufficient hierarchical complexity but not vice versa.
Neural networks appear to act as Bayesian modeling systems. Many other statistical modeling systems are also consistent with certain neural network studies so perhaps I should just simplify what I said to "connectionist statistical modeling systems". However, I believe the evidence is accumulating rapidly that the main form utilized by brains from the simple to complex is Bayesian. Bayesian modeling systems can have a wonderful variety of capabilities from the simple to the complex. At the most basic, they serve as pattern recognition systems for our basic senses. No reasoning involved. Alternatively, artificial Bayesian networks are being used to construct real-world application for reasoning (e.g. language recognition and processing) even today.
I think, ironically, that you are making some of the same mistakes Pixi has made arguing with me. You are making something analogous to a reification error by equating abstractions with physical or mental processes and actual computation in the brain (see my last post to Pixi in the "are you conscious>" thread). While the brain may, for example, actually directly compute forward and backward chaining operations computed by NNs somehow, I'm not aware of any evidence for this nor do I think it happens. What the brain does is potentially isomorphic to chaining in function but not form. Massively parrallel NNs statistically model a superposition of states of various
a priori and
a posteriori probabilities and then use a relaxation process to arrive at local or global minima in their "energy/entropy landscape" of possible solutions to find the best or better solutions. The exact nature or natures of the relaxation processes have not been proven in real brains but several workable and successful artifical NN's have been developed that suggests that real brains do something similar. Perhaps one of the best is among the first, i.e., the Boltzmann Machine. I highly recommend you look into it if not already familiar. It will help give you a better connectionist perpective too.
