>JOHN ROSE ============>> I suppose the optimal approach to AGI has to involve some degree of connectionism. But to find isomorphic structures to connectionist graphs that are more efficient. Many things in nature cannot be evolved, for example few if any animals have wheels. Evolved structures go so far until probabilistic limits are hit. Are there structural, algorithmic and mathematic systems that are more optimal than massively dense interconnected graphs waiting to be discovered or, are they actually known but not applied to engineered cognition and consciousness? I say yes. Has activation dynamics been studied enough, - anyone have a literature reference?
Then main reason I take this approach is because of resource constraints. I'm not saying that it's not worth building connectionist prototypes. In fact I'm starting to think that way where I haven't before. ED PORTER ============>> I am not an expert at computational efficiency, but I think graph structures like semantic nets, are probably close to as efficient as possible given the type of connectionism they are representing and the type of computing that is to be done on them, which include, importantly, selective spreading activation. >JOHN ROSE ============>> I agree on your description of consciousness. But it would be nice to have a compact system, a minimized essence, that optimal consciousness engine if one exists. Computation is all that there is. But I often try to imagine something that is not computation. It depends on different things, and goes into subatomic physics, string theory, etc.. I think that there are aspects of computation that we don't understand, and definitely things that I don't understand but are known among well versed individuals. ED PORTER ============>> Although I think my theory of consciousness is as good as any other I have read, it is far from certain, and far from complete, and not necessarily correct in every details. I don't think the richness of human consciousness comes from a "minimized essence", but rather from the complicated full-blown richness of the computation inside our brains. Since our senses can only sense --- and our minds can only think --- in terms of computation (in which I am included representation) --- at least at the moment, I cannot think of what it would mean for something to not be computation, except perhaps nothingness, which we can think of as the absence of computation. -----Original Message----- From: John G. Rose [mailto:[EMAIL PROTECTED] Sent: Tuesday, June 03, 2008 1:13 AM To: [email protected] Subject: RE: [agi] CONSCIOUSNESS AS AN ARCHITECTURE OF COMPUTATION I suppose the optimal approach to AGI has to involve some degree of connectionism. But to find isomorphic structures to connectionist graphs that are more efficient. Many things in nature cannot be evolved, for example few if any animals have wheels. Evolved structures go so far until probabilistic limits are hit. Are there structural, algorithmic and mathematic systems that are more optimal than massively dense interconnected graphs waiting to be discovered or, are they actually known but not applied to engineered cognition and consciousness? I say yes. Has activation dynamics been studied enough, - anyone have a literature reference? Then main reason I take this approach is because of resource constraints. I'm not saying that it's not worth building connectionist prototypes. In fact I'm starting to think that way where I haven't before. I agree on your description of consciousness. But it would be nice to have a compact system, a minimized essence, that optimal consciousness engine if one exists. Computation is all that there is. But I often try to imagine something that is not computation. It depends on different things, and goes into subatomic physics, string theory, etc.. I think that there are aspects of computation that we don't understand, and definitely things that I don't understand but are known among well versed individuals. John _____________________________________________ From: Ed Porter [mailto:[EMAIL PROTECTED] Sent: Monday, June 02, 2008 2:46 PM To: [email protected] Subject: RE: [agi] CONSCIOUSNESS AS AN ARCHITECTURE OF COMPUTATION >JOHN ROSE ============>> So you are saying that consciousness is activations in response to patterns, including activation history. ED PORTER ============>> Yes. But I am also saying the following: -EVERYTHING -- INCLUDING CONSCIOUSNESS --- IS NOTHING BUT COMPUTATION To those who say it is a cop out to say consciousness is computation, I challenge you to describe any aspect of reality, either that in the mind, or that described by current scientific understanding, that is anything other than information and its computation (both of which I am calling computation here). Computation is all there is --- at least it is all we have any evidence for believing there is. Physics is nothing but computation --- and all science follows from the laws of physics, even if many aspects of science relate to higher levels of organization that result from the complexity of the computation of those physical laws. The theory of cosmic inflation seems to imply that our physical universe was created from an infinitesimally small extremely high concentration of mass/energy --- such as that which might exist in a black hole --- that suddenly had a phase transition that allowed a new, more complex from of representation and computation to take place ---- causing it to change from something that was, or approached, a singularity to something approached the number of the particles in the know universe within a brief period of time. Out of such computation the time, space, matter, and energy of our know universe was created. Just as the physical universe is created out of computation, so is consciousness. When you ask people to describe consciousness they tend to do so in computational terms such as awareness, seeing, feeling. I have never heard anybody describe any attribute of consciousness that involves anything other than computing. (Except, perhaps for people who associated religious attributes to consciousness, but all of those religious notions arguably are manifestations of computation within some sort of believed reality.) -THE COMPUTATION OF CONSCIOUSNESS IS MASSIVELY PARALLEL, MASSIVELY RICH, MASSIVELY INTERCONNECTED My model of consciousness is NOT Searle's Chinese room --- instead it is more like a stadium with an audience of tens of billions of agents, each acting somewhat like a Chinese rooms --- i.e., each with its own library of specific information and each with its own input-output function. If one imagines a football stadium with 100,000 seats --- to have a number of input-output units, i.e., Chinese rooms, equal to the number of neurons in the cortex, each of its 100,000 seats would have to have 300,000 neurons. In the stadium of consciousness, the individual Chinese rooms each directly communicate with hundreds or thousands of other rooms, and they respond to collective electrical and chemical communication from tens of millions or billions of neurons, including neuro-chemicals released by the limbic system and EEG brain waves. They all operate in parallel. Yet these billions of Chinese rooms are under a combination of decentralized and centralized control that dynamically help control their activation level, what is being communicated to them, and the preferential broadcast tuning preferences of each of them. This enables those tens of billions of Chinese rooms to behave as a dynamic collective entity, much as a fired up crowd in a large stadium or arena can seem to have a life of its own. -THE COMPLEXITY OF THE BRAIN'S COMPUTATION GIVES RISE TO CONSCIOUSNESS MUCH AS THE COMPLEXITIES OF BIO-CHEMISTRY GIVE RISE TO LIFE Consciousness is to computation, what life is to bio-chemistry. The reductionist attempt to deny that human consciousness could result from computation with the argument of Searle's Chinese room is just as stupid as saying bio-chemistry could not give rise to biological life, because if one looks at any one very small section of space and time what we see of bio-chemistry is nothing but a limited number of chemical reactions --- something none of us would normally call "life." A living organizism is something very different than a few biochemical reactions, it is many billions, trillions, or trillions of trillions of them operating simultaneously in a surprisingly organized manner. Similarly the brain's 30 billion highly interconnected simultaneously operating, highly interactive, Chinese rooms is something very different than Searle's single Chinese room. One of the most important paradigms in science is what the Santa Fe institute calls "complexity". (When I refer to "complexity" herein I mean this conception of complexity --- not the totally unrelated concept of Richard Loosemore's (or RL) complexity, which caries a lot of additional baggage.) The essence of the paradigm of complexity is that as the number of interacting elements in a collection of certain types of elements increases, there is a tendency for higher level and more diverse patterns of organization and behavior to develop. Under the right conditions, as the number of such elements increases, normally the richness and complexity of the new levels of organization themselves also increase. E.g., as a general rule, the larger the city, the more variety of organization and behaviors it contains. This same paradigm is reflected in computing. I have heard it said by many computer geeks that with every increase by a power of ten in computing power there are not only changes in degree but also changes in kind of computation. If this is even partially true, imagine what new levels and types of computation could be achieved by computers that had the representational, computational, and intercommunication capacity of the human mind. My first digital computer was a four bit adder and subtracter I built in the sixth grade from electromagnetic relays I built myself. My first electronic computer was an IMSAI 8080 in 1977 that I originally had to program a word at a time with front panel toggle switches. Now I have a 3Ghz P4 PC. I have seen computers go from merely being able to turn a few flashlight bulbs on and off to now being able to generate awe inspiring multi-dimensional sounds and video rendering of emotionally satisfying DVDs. As the power of computers has grown in the last 50 years their ability to provide our senses with experiences that increasingly occupy our consciousness has grown to the point that --- for many seconds at a time watching a DVD, or entering virtual worlds, can make us feel they have taken us to a different time and place. The computational power of the human brain has been estimated by multiple people as being roughly 100TOpps/sec. This is roughly 100,000 times the power of the current average PC. But the brain is many times more powerful than that in terms of opp to memory bandwidth and opp to opp interconnect. The human brain is powerful enough to dynamically generate --- in response to its sensations and own internal activations --- the informational bandwidth equivalent of roughly one million simultaneous uncompressed 800x600 DVD channels. It is this complex, self-reflecting computations, a million times more informationally rich than an uncompressed DVD image, that provides the sense of awareness, we humans call consciousness. This would be a football stadium of consciousness in which each of its 100,000 seats was interactively generating an information flow as complex as ten uncompressed DVD images. That is one hell of a vibrant, collective computation. -COMPUTATION OF CONSCIOUSNESS IS SELF AWARE The computation of our consciousness is self-aware because its responds to itself. We are aware of the roughly 1 uncompressed DVD bandwidth of info/sec we get from our senses because we dynamically compute roughly another one million times as much dynamic state in response to it, which includes partial activations of a substantial number of all patterns and episodic memories which relates to any portions of that input our consciousness focuses on. This tremendously self aware dynamic activation is consciousness. On could argue that the feedback loop of a simple electronic oscillator is thus self aware. But it does not have the billions or trillions of feedback loops of the brain, many of which can be selectively inhibited or dis-inhibited --- that gives rise --- through the paradigm of complexity --- to the generation of higher levels of organization and behavior which have no parallel in the individual simple feedback loop. Differences in degree this large become massive differences in kind. -CONSCIOUSNESS COMBINES BOTH COLLECTIVE AND MULTI-BILLION CHANNELED COMPUTATIONS People often state they sense a unity of consciousness, even though their consciousness seems to be aware of multiple things at once. This sense of unity results because human conscious appears to result from a selection and broadcast of activation from roughly 30 to 50 concepts a second, where a concept can represent a not just one pattern, but also a set of related pattern activations. The activation of each of these concepts interacts with many if not all the concept and experience patterns recorded as having including or having been associated with it in the past --- and in doing so, changes the activation state of those patterns. As successive concept are consciously activated, the activation state reflects a high percent of all the relevant patterns that directly or, in many cases, indirectly involve subcombinations of such consciously broadcast concepts and their implications. As this composite activation state transitions over time it reflects not only billions of implications from the current activation being broadcast, but also history of billions of implications reflecting the associations between the implications of each of the consciously activated concepts. Thus, the composite activation state of the system over time provides a web of meaningful relatedness between multiple concepts that have been recently broadcasts in the consciousness --- providing a sense of unified awareness of these successive concepts and their relatedness. -ATTENTION FOCUSING J.G. Taylor has hypothesized that we are conscious of a concept when our brain activates to various degrees many --- or all --- of our memories --- both in repeated patterns and episodic memories --- with which that conscious concept is associated in long or short term memory. This makes sense. How better to explain awareness of a concept than to have simultaneously subconscious activation many or all the relevant thing you know about it. This subconscious activation of millions or billions of related patterns or implications enables simultaneous summing of important similarities, important emotions, benefits and dangers, and most importantly, and the relevancies of that concept in the current mental context. Multiple of the more activated of these subconsciously activated patterns may flicker at the edge of consciousness, a few may be sequentially or concurrently selected into full consciousness. But even if these subconscious implications of a conscious concept may not be fully activated by the consciousness of the concept itself, they may have an effect on the activation of concepts in response to subsequent conscious concepts, effecting mood, non-verbalizable intuitive feelings, and which concepts are selected for conscious activation in response to subsequent conscious concepts. -SELECTIVE BROADCASTING Attention focusing and broadcasting are closely related. Broadcasts could be performed by mechanisms such as, selective thalamic dis-inhabitation through the cortico -basil ganglia-thalamic feedback loop, synchronization, substantially cortex wide radial waves of dis-inhibition from the thalamic reticular nucleus, or from massive spreading activation in which the strength of the broadcast is a function of the number of activations that take place over multiple generations of spreading activation from a given node or group of nodes. Some believe people believe that conscious involves a rapid sequence of conscious broadcasts. Many believe that broadcasts of successive conscious concepts can take place at about 30 to 40 concepts a second, with the possibility that the same concept may be broadcast repeatedly, such as to repeatedly broadcast it the activations associate with it change. -COLLECTIVE BROADCASTING This can include EEG waves; communication through changes in neuro-chemical soup in one or more parts of the brain (such as those that can be quickly transmitted by the limbic system); the activation of large patterns of activation in unison (such as by synchronies); and collective activations of all the elements of complex patterns. -COLLECTIVE CALCULATIONS This includes intuitive summations of activations from many sources that result from a given context of activation, for example the summation of various types of emotions that get activated by a situation. It also includes max operations that find which of competing patterns has the highest net activation coming from many sources of activation and inhibition, such as estimates of cost and benefit, for selection in a winner or k-winner take all situation, of which physical, mental, or perceptual, behavior to take. -SUBCONSCIOUS The subconscious is a vital part of consciousness. According to J.G. Taylor it is the massive activation of concepts and emotions in the subconscious relating to a conscious concept, that provides and fleshes out the sense of awareness of the conscious concept. It is in the computation of the subconscious in which massively parallel pattern activation, assembly, and matching can take place. It is from these that collective emotional response can be selectively summed --- contending pattern matching or behavior candidates can be competitively selected --- and patterns and behaviors can be selected for broadcast in the consciousness. -CONSCIOUSNESS COMES IN DIFFERENT DEGREES AND MODES There are multiple modes and degrees of consciousness, such as different degrees of arousal or different types of emotional states. You are conscious during dreams, particularly if you are part awake. In fact, I have read that different parts of your brain can fall asleep at different times, giving rise to some strange states of consciousness. I think the degree of attention given to various concepts of which we conscious vary tremendously, from those we are only slightly conscious of to those that leave an intends and enduring memory. If one has a Novamente-like spreading activation model of intelligence, such an intelligence could have many different degrees of activation associated with a concept, that could correspond to differing levels of consciousness, and it could simultaneously broadcast multiple different conscious concepts. -MASSIVE EXPERIENTIAL MEMORY The human consciousness involves a complex activation state that varies both rapidly and slowly over time, in which the activation patterns are largely created out of learned experiential patterns. It has been estimated by many that the brain probably stores the equivalent of roughly 10TBytes of information. (This not as measures by how many bits it would take a hypothetical turning machine with infinite resources to represent such information in an optimally compressed form, but rather as measured in terms of how much computer memory would be required to store it in an operationally efficient manner.) I think this estimate is within one or two order of magnitude of being sufficient for computers to have human level intelligence. 10TBytes is enough information to store 10K bytes per second for a 31 year life time. Of course the amount of memory stored per second varies tremendously depending on importance and recency. Since much of this information would be in the form of learned repeated patterns in hierarchical memory, such a representation would allow very efficient representation. It would because there would be a vocabulary of tens of millions of hierarchical patterns, each of which could represent complex invariant patterns memories, which when remembered in a given context would tend to dynamically assume the hierarchical activation most appropriate for that context. >JOHN ROSE============>> Do you have a non-connectionist view of consciousness? Just curious. ED PORTER ============>> I consider myself to be a connectionist in roughly the same way Ben considers Novamente to be connectionist. This is to say my theories have many connectionist aspects, but they not are limited by some of the restrictions associated with more narrow interpretations of connectionism. For example, any system that uses natural language words would have symbols. Some people think a system with symbols is not connectionist. But the same people consider the brain to be connectionist, despite the fact it uses words, which are symbols. I think you can make a system that mixes traditional programmer taught symbols, and automatically learned patterns by the experiential knowledge learned about the relationships between these two types of knowledge. In the type of artificial brains I image there would be many automatically learned patterns associated with what I call sub-concepts, i.e., concepts that relate to things which we humans would usually only be conscious of as part of the activation of larger patterns. I favor an architecture in which the vast majority of the patterns in the system are learned automatically, and a symbol's meaning is defined by what it is connected to by links. I tend to think in terms of semantic nets having a very small number of pre-defined link-types (such as say less than 10) --- something that is not very dis-similar from Novamente's hypergraphs. -----Original Message----- From: John G. Rose [mailto:[EMAIL PROTECTED] Sent: Sunday, June 01, 2008 12:36 PM To: [email protected] Subject: RE: [agi] CONSCIOUSNESS AS AN ARCHITECTURE OF COMPUTATION Hi Ed, So you are saying that consciousness is activations in response to patterns, including activation history. Do you have a non-connectionist view of consciousness? Just curious. John _____________________________________________ From: Ed Porter [mailto:[EMAIL PROTECTED] Sent: Thursday, May 29, 2008 9:54 PM To: [email protected] Subject: RE: [agi] CONSCIOUSNESS AS AN ARCHITECTURE OF COMPUTATION Matt makes some good points below. Qualia are a key aspect of consciousness. But consciousness probably comes in many different forms and degrees. It is not clear episodic memory is essential to all consciousness, but it plays a large part in ours. I do think consciousness require a state space with a temporal dimension. It would be very different than our own if it did not have a memory of temporal patterns. I CONCEIVE OF CONSCIOUSNESS AS THE SENSE OF AWARENESS OF SENSATIONS, EMOTIONS, MEMORIES, AND THOUGHTS MY MIND EXPERIENCES --- and that other humans also claim to experience. I THINK CONSCIOUSNESS RESULTS FROM THE NATURE --- AND COMPLEXITY OF --- COMPUTATION WITHIN THE HUMAN MIND --- THAT IT IS NOT AN INCONSEQUENTIAL SIDE EFFECT --- BUT RATHER A KEY FUNCTIONAL ATTRIBUTE OF COMPUTATION WITHIN THE BRAIN. I BELIEVE CONSCIOUSNESS RESULTS FROM A MENTAL ARCHITECTURE SOMETHING LIKE B. J. BARRS' THEORY OF THE THEATER OF CONSCIOUSNESS. The analogy to a theater is only partial --- but valuable. IT SUGGEST THE MIND IS LIKE A THEATER FULL OF AGENTS --- EACH WITH INTERNAL STATE AND DESIGNED TO RESPOND TO CERTAIN PATTERNS RECEIVED BY IT IN CERTAIN WAYS. THE THEATER HAS A MECHANISM FOR SELECTING WHICH OF THEM GET TO BROADCAST AT ANY GIVEN MOMENT BY SOMETHING ANALOGOUS TO SELECTIVE SPOT LIGHTS --- OR SELECTIVE LINKS BETWEEN DIFFERENT COMBINATIONS OF MICROPHONES AND SPEAKERS. TO MAKE THE ANALOGY BETTER, THE AGENTS SHOULD ALSO BE CONNECTED WITH SOMETHING LIKE INSTANT MESSAGING. The analogy assumes human consciousness results from the behavior of a theater of millions or billions of agents that transmit and respond to broadcast patterns, including temporal patterns, spread through significant portions of the brain. The agents have state information that reflect not only what is currently being broadcast to them, but also what has been broadcast to them before, and more private communications they are receiving, or have received, reflecting the state of directly connected agents. There is machinery --- such as the cortical - basil ganglia - thalamic feedback loop --- for selecting which patterns get broadcast and when, based on measures such as degree of match, importance, fit or importance within context, or surprise. Some of the patterns activated in response to broadcasts can themselves be selected for successive conscious broadcast in a stream of consciousness manner. The broadcasts vary in the number of elements they reach, but we humans are probably only directly "conscious of" broadcasts that reach a relatively large number of other elements, either directly or indirectly. This broadcast normally result in mullti-hop spreading activations. The spreading activation and the activation of responsive patterns are modulated by the context of co-occuring and prior activation patterns --- and by attention allocating mechanisms. CONSCIOUSNESS IS NOT JUST WHAT IS BROADCAST --- BUT ALSO THE TOTALITY OF ACTIVATION IN RESPONSE TO IT --- as sensed by intuitive qualia-like and emotional experiential responses --- and as sensed by individual subconscious activations that pre-dispose the later conscious or subconscious activation of related patterns or transmission. OUR INTUITIVE EMOTIONAL RESPONSES ARE SOMEWHAT LIKE CROWD RESPONSES IN A THEATER. They tend to sum the emotional associations of many different agents to what is being, and has been, broadcast into one composite emotional sensation stream --- even though that composite can include activations of differing emotion patterns, corresponding to the boos and cheers of a crowd. Listen to radio of a crowd during an exciting situation in an important playoff, or world series, game. You will get a feeling for the sense of the stadium's (a type of theater's) crowd as a dynamic, living, collective spirit. You can hear the difference between a pop ball and a ground drive --- between a home run, a pop fowl, and a pop out --- and you can hear group chants --- that often start in the unconscious of the crowd and then build up so until they are heard in its consciousness --- or that are driven by the stadiums organ. You can hear all this from the crowd's collective voice. You can sense its collective mind. The mind's theater is probably something like a caucus meeting, because apparently it can be divided up into differing sub-theaters. This is indicated by the fact that synchronies --- which appear to be associated with at least certain kinds of conscious experience --- can be dynamically wired between different neural sub-populations in the mind depending on the particular mental function being performed. There is also evidence that other types of broadcast spread selected information throughout much of the cortex, such as by the waves of decreased inhibition that radiate out from selected points to almost all of the thalamic reticular nucleus, which controls inhibition from most of the thalamus to the cortex. Sensory inputs would be inputs into sub-populations of the theater, which would process them through a succession of hierarchically and laterally connected sub-populations by relatively local communication, and activations in such populations would only enter consciousness if activations in response to them within some one or more levels of these sub-populations were selected for wider broadcast. THE SENSE OF AWARENESS WE ASSOCIATE WITH CONSCIOUSNESS RESULTS FROM THE HUNDREDS OF BILLIONS OF SYNAPSES THAT FIRE IN RESPONSE TO EACH OF THE ROUGHLY 40 HERTZ GAMMA-WAVE ACTIVATION CYCLES OFTEN ASSOCIATED WITH CONSCIOUS RELATED BROADCASTS --- FIRINGS THAT ACTIVATE MILLIONS OR BILLIONS OF PATTERNS TO VARIOUS DEGREES. THIS MEANS THE SYSTEM IS WATCHING --- I.E., RECURSIVELY RESPONDING TO --- ITS OWN COMPUTATION, with an informational bandwidth equal to tens or hundreds of thousands of simultaneous uncompressed HDTV channels --- bandwidth dynamically generated in response to dynamically selected portions of itself by dynamically select portions of the equivalent of roughly hundreds of trillions of computations a second --- computations that respond to massive, instantaneous broadcasts and collective summations --- while at the same time responding to billions of instantaneous, individual, highly non-linear distinctions. THIS SELF-REFLECTIVE COMPUTATION --- THAT HAS BOTH MASSIVE UNITIES AND DISTINCTIONS --- AND STORES A DYNAMIC HISTORY OF ACTIVATIONS --- REPLACES THE HOMUNCULUS AS THE SOURCE OF OUR PERCEPTION OF A UNIFIED AWARENESS OF MULTIPLE EXPERIENCES, JUST AS THERE IS NO EVIDENCE PHYSICAL REALITY IS ANYTHING OTHER THAN COMPUTATION --- SIMILARLY --- THERE IS NO EVIDENCE CONSCIOUSNESS IS ANYTHING OTHER THAN COMPUTATION. JUST AS THE NATURE OF PHYSICAL REALITY IS DETERMINED BY THE ARCHITECTURE OF ITS COMPUTATION (THE LAWS OF PHYSICS) --- SIMILARLY --- THE NATURE OF CONSCIOUSNESS IS DETERMINED BY THE ARCHITECTURE OF ITS COMPUTATION. There is no reason to believe consciousness does not come in many different forms and degrees. We humans sense different degrees and types of consciousness, depending on whether or not we are asleep, awake, dreaming, meditating, concentrating, or how aroused we are, or what chemicals we may have taken into our bodies. There is no reason to believe that many lower animals --- such as primates, pigs, dogs, perhaps birds, and perhaps even insects --- are without any form and degree of consciousness. In fact, it is arguable that consciousness as we normally use the word to describe our own sense of self awareness is just a special kind and degree of the computation of physical reality itself --- and that all of reality is conscious to some degree. The rate and complexity of quantum mechanical and chemical computations of the atoms and molecules within a lowly worm are probably many billions of times more complex than the logical computer computations within the first human level AGIs. But presumably whatever "consciousness" pervades most of physical reality --- including our lowly worm --- lacks many of attributes of human consciousness that we value most --- which define the type of consciousness we normally mean when we use the word. I think within a few decades many AGI's will have computations with many of the characteristics we humans attribute to our own consciousnesses --- and that those characteristics will be very useful to those AGIs. I DON'T KNOW IF CONSCIOUSNESS IS NECESSARY FOR HUMAN LEVEL AGI --- BUT I THINK SOMETHING WITH MANY OF THE ATTRIBUTES OF HUMAN CONSCIOUSNESS IS ESSENTIAL FOR ANY AGI TO HAVE AN ARCHITECTURE OF THOUGHT THAT FUNCTIONS LIKE OUR OWN. Ed Porter -----Original Message----- From: Matt Mahoney [mailto:[EMAIL PROTECTED] Sent: Thursday, May 29, 2008 2:58 PM To: [email protected] Subject: RE: [agi] Consciousness vs. Intelligence What many people call consciousness is qualia, that which distinguishes you from a philosophical zombie, http://en.wikipedia.org/wiki/P-zombie There is no test for consciousness in this sense, but humans universally believe that they are conscious, and this belief is testable. Just ask someone. Do you really feel pain, or do you just behave as if you feel it? The belief in experiencing qualia is what I call recursive episodic memory. Episodic memory is the ability to recall a time sequence of events in the correct order. These events could include earlier acts of recall. For example, earlier today I recalled how yesterday a tune was playing in my head that I heard the day before (and so on). You probably do not remember any events that happened before you were 3 years old. You were clearly learning then, but it was not in episodic memory. A person without a hippocampus lacks episodic memory. He could learn new skills but wouldn't remember the lessons. Episodic memory has been demonstrated in birds, but we do not know if it is recursive. I don't know if recursive episodic memory is necessary for intelligence. When I need to come up with an algorithm when writing software, it is useful to go through the steps in my head and then be able to recall my thought process. It is also useful for databases to log read-only transactions. It is useful for computers to copy recently read data to cache. However, recursive episodic memory could also be an artifact of the brain's memory management system. Long term memory is written at a constant rate (about 2 bits per second, according to Landauer). During quiet times, it has to write something. -- Matt Mahoney, [EMAIL PROTECTED] ------------------------------------------- agi Archives: http://www.listbox.com/member/archive/303/=now RSS Feed: http://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: http://www.listbox.com/member/?&; Powered by Listbox: http://www.listbox.com ------------------------------------------- agi Archives: http://www.listbox.com/member/archive/303/=now RSS Feed: http://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: http://www.listbox.com/member/?&; Powered by Listbox: http://www.listbox.com ------------------------------------------- agi Archives: http://www.listbox.com/member/archive/303/=now RSS Feed: http://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: http://www.listbox.com/member/?&; Powered by Listbox: http://www.listbox.com ------------------------------------------- agi Archives: http://www.listbox.com/member/archive/303/=now RSS Feed: http://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: http://www.listbox.com/member/?member_id=8660244&id_secret=103754539-40ed26 Powered by Listbox: http://www.listbox.com
