---- > What does a mind do? > > A mind in an intelligent animal, such as ourselves, does the following: > > 1. Interprets sense-data and symbolically represents the objects, relationships, processes, > and more generally, situations that occur in its environment. > > Extra buzzwords: segmentation, individuation, > "cutting the world with a knife into this and not-this" > (paraphrased from Zen & The Art of Motorcycle Maintenance) > > 2. Creates both specific models of specific situations and their constituents, > and abstracted, generalized models of important classes of situations and situation > constituents, using techniques such as cluster analysis, logical induction and abduction, > bayesian inference (or effectively equivalent processes). > > Extra buzzwords: "structure pump", "concept formation", "episodic memory" > > > 3. Recognizes new situations, objects, relationships, processes as being instances > of already represented specific or generalized situations, objects, relationships, > processes. > > The details of the recognition processes vary across sensory domains, but probably > commonly use things like: matching at multiple levels of abstraction with feedback > between levels, massively parallel matching processes, abstraction lattices. > > Extra buzzwords: patterns, pattern-matching, neural net algorithms, > constraint-logic-programming, associative recall > > > 4. Builds up, through sense-experience, representation, and recognition processes, > over time, an associatively interrelated "library of symbolic+probabilistic models or > micro-theories" about contexts in the environment. > > 5. Holds micro-theories in degrees of belief. That is, in degrees of being considered > a good "simple, corresponding, explanatory, successfully predictive" model of some > aspect of the environment. > > 6. Adjusts degrees of belief through a continual process of theory extension, > hypothesis testing against new observations, incremental theory revision, assessment of > competing extended theories etc. In short, performs a mini, personalized equivalent > of "the history of science forming the evolving set of well-accepted scientific > theories". > > Degree of belief in each micro-theory is influenced by factors such as: > > a. Repeated success of theory at prediction under trial against new observ ations > > b. Internal logical consistency of theory. > > c. Lack of inconsistency with new observations and with other micro-theories of possibly > identical or constituent-sharing contexts. > > d. Generation of large numbers of general and specific propositions which are > deductively derived from the assumptions of the theory, and which are independently > verified as being "corresponding" to observations. > > e. Depth and longevity of embedding of the theory in the knowledge base. i.e. > the extent to which repeated successful reasoning from the theory has resulted in the > theory becoming a "basis theory" or "theory justifying other extended or analogous > theories" in the knowledge base. > > > 7. Creates alternative possible world models (counterfactuals or hypotheticals), > by combining abstracted models with episodic models but with variations generated > through the use of substitution of altered or alternative constituent entities, > sequences of events, etc. > > Extra buzzwords: Counterfactuals, possible worlds, modal logic, dreaming > > 8. Generates, and ranks for likelihood, extensions of episodic models into the future, > using stereotyped abstract situation models with associated probabilities to predict > the next likely sequences of events, given the part of the situation that has > been observed to unfold so far. > > 9. Uses the extended and altered models, (hypotheticals, counterfactuals), as a context > in which to create and pre-evaluate through simulation the likely effectiveness of > plans of action designed to alter the course of future events to the material > advantage of the animal. > > 10. Chooses a plan. Acts on the world according to the plan, either indirectly, > through communication with other motivated intelligent agents, or directly by > controlling its own body and using tools. > > 10a. Communicates with other motivated intelligent agents to assist it in carrying > out plans to affect the environment: > Aspects of the communication process: > - Model (represent and simulate) the knowledge, motivations and reasoning processes of > the other agents, > - Communicate to the other agents, using conventional symbols. > Subjects of communication: > -specific and general knowledge about the environment and its relevant non-optimalities, > -suggested specific courses of action to remedy non-optimality, > with logical justifications, and motivational explanations of likely mutual benefits. > > (Evolutionary side effect of need to model the cognition and action planning process of > assistant agents: The development of ability to reflect on one's own > thinking process. i.e. some of what consciousness is) > > 11. Assesses the effects of planned actions, corrects, reassesses, remodels, > plans again, acts again. > > 12. Prioritizes cognition to meet real-time understanding and planning deadlines, > via techniques such as: > a. Parallel exploration and testing of alternative hypotheses, > b. Parallel associative recall processes > c. Periodic assessment of relative promise of the parallel alternative lines of > reasoning and different directions of associative recall exploration of memory, > and pruning of less promising directions to allow redirection of the finitely > resourced search processes to deeper exploration of more promising directions. > d. Emotion-tagging of episodic memories of "survival-important" aspects of situations, > and carry-over of the emotion-tags to abstracted general rules. These emotion > tags serve as a prioritization on all-directions associative recall and on > all-directions inference during exploration of recalled theories and situations > and current situation. i.e. > - If it could kill you now, recall it fast and don't think about anything else. > - Similarly if it could get you an opportunity to pass on your genes. > - If it could kill you slowly (hunger, lack of shelter), or make your world safer > and more comfortable (food, shelter, friends and helpers, money), > think about it with relatively high priority, and probably quite often. > - If it is new and unknown, it could kill you now, so think a lot about it now until > you can learn enough to assure yourself that that is not the case. > - If it is same old same old, and not one of the above categories, muse on it > if you have the time. You might learn something new of benefit by mixing it with > some old thought to generate an important realization. > > > --------- > > What is a mind? > Something that does all of the above, and undoubtedly more. > > How does it work? F****ed if I know. >Received on Sat Dec 28 2002 - 11:34:14 PST
This archive was generated by hypermail 2.3.0 : Fri Feb 16 2018 - 13:20:07 PST