Chuck Rosenberg

Rambling Thoughts...

These are all pretty old and outdated...

Head Tilting Behavior A pretty simple connection. People and many animals seem to tilt their head when they are puzzled by something. My idea here is that this helps overcome certain aspects of object detection and recognition processing in the visual system. The two mechanisms are built in priors that toward upright objects and certain mechanisms that only operate on objects which are upright. Other possible explanations besides the one that I am advocating are that it allows the individual to re-orient their ears for better sound localization or that it provides some additional parallax for improved depth perception. -- 3/17/02 Dark Matter This is probably a pretty silly idea, but here it is. This is an idea I came up with about two years ago but never jotted down. The whole concept of it will probably just show how naive I am about modern physics. I know that currently it is estimated that 90% or more of the mass of the universe must exist in some form of dark matter. My idea is this, what proof do we have that the "Big Bang" only created these three dimensions of matter. The way to think of this is by the standard balloon analogy of the universe. That is, in a 2D analog of the universe all of the galaxies are sitting on the surface of the balloon (curved space time) and are moving away from each other. My idea is this, is there evidence that such a balloon is empty? What if there was matter in the center of the balloon and the only way that matter could interact with our 3D universe was via gravity? Could that explain the presence of a large amount of matter that is invisible? Is it at all plausible given the tenets of modern physics? That I do not know, but would love to find out. -- 11/5/97 Programs with Free Will I came up with this idea about nine months ago. I wonder if the current history of computer science has predisposed us to building systems without free will. (By free will here I am not thinking of the deep metaphysical problem, but the issue of whether a system has to do exactly what it is are told or if it can make some decisions for yourself.) If you take a look at almost every application in use today, a computer simply follows rote orders. We expect our word processors and spread sheets to be obedient servants. Right now agents with free will don't make sense in the way we expect to use them. In certain ways a program with free will seems ridiculous - should we allow it to go off to school? What if it won't co-operate? Can we convince it somehow? I guess I'm not sure if we would accept such a program. I also believe that the tools for building programs were not designed with this purpose in mind. As we build more and more intelligent agents the definition of free will change. Obviously a program which decides to buy stock has some sort of free will. And a robot which plans its path through a cluttered office has some. But as these agents become more intelligent, how much control will they be given - will a human have to OK every step? And as humans OK less and less of the steps, what use will we have? -- 10/5/97 What is a Program? I came up with this idea about a year or two ago. The issue is what is a program? How would we explain it to someone from a time before computers? If I had to describe my job to someone, would I tell them that all I do all day is punch little keys? And to what end? How do I describe information processing and virtual machines? Could I describe it as if I was a manager of a very dumb work force that speaks a very specialized language and I have to tell them in great detail what to do? That does make computers the equivalent of very dumb animals you have to train? But the strange thing is that programming is the construction of machines with possibly no connection to the physical world. All a program does is shunt little electrons around. This is the whole issue of what makes software patents so difficult. Of course a program is some form of virtual machine which does some sort of work. A physical machine which did the same thing could be patented. But since we can duplicate a program at basically no cost, does it have any intrinsic value? Obviously it cost something to build it, but it costs basically nothing to manufacture it. Programs are strange beasts. -- 10/5/97 Out of Tune This is a very old idea, one I had probably more than 10 years ago, but I don't think I have ever written down before. The question is why don't we hear an accent when someone sings? Is it because they don't sing with an accent? Specifically, is there interference between the parts of the brain that create the "tune" of a language and the "tune" of a song?. Or is it some masking effect where we can't hear an accent in speech while someone is singing? Specifically, is there an effect on the listener where the tone of a singing voice and an accent cannot simultaneously be perceived. -- 9/24/97 Breaking the Speed Limit Does quantum computation say something fundamentally about the nature of the universe? It seems like your are getting something for nothing, like it must violate some basic principle. It almost seems like there should be some infinite mass or energy involved. I guess I am asking if there is any fundamental theory about computation, like it must take at least one photon to compute one bit. Does it say something about P=NP? What if mathematics could prove that P != NP, but quantum computation would allow you to solve NP problems in P time? -- 9/4/97 The Ultimate Purpose of the Universe An interesting link... Ever wonder what's the purpose of it all? Author Douglas Adams suggests that the Earth was built as a giant computer whose purpose was to compute the ultimate answer (actually the question) to life, the universe, and everything. Well, what if the purpose of our universe was a little less grandiose. Ever wonder why astrophysicists cannot decide whether the universe is closed or open? What if the purpose of our universe was to compute a single bit - the value of which is determined by whether the universe is open or closed. -- 1/20/97 Stealing Cycles A variation on "the ultimate purpose of the universe". Recently I was reading an article about web host sites being used as supercomputing centers. The web hosts could steal cycles from the computers browsing them by the use of applets performing a distributed computation. Playboy could be in the supercomputer business. This led me to an idea, an old one in Sci-Fi, but maybe one which needs more consideration. If the universe is a computer which computes its physical laws, is there a way to steal cycles from it to get the computations that we need done? At the basest level, any computer or even the human mind is doing that. But, is there a deeper way to exploit that computational power? Would a quantum computer somehow do a better job of that? Or is it a simple matter of how many particles of the universe you are using for computation? What if you used half the particles or energy of the universe for a computation? Would you seal off a large area of space, place the particles in some initial configuration, and then wait for the results? Then you get to the questions of multiple universes. Would some set of physical laws be better suited to computation than others? What if a universe could be created just for computational purposes? You could get one bit out at the end - either the universe expands forever or it contracts on itself. What if some group searched from universe to universe looking to use them for computation? Or could someone steal cycles from the machinery of the universe? If things slowed down would we notice? This would assume that there is some machinery that is behind the physical laws we observe. These ideas crystallized out of a dream I had where a dragon-like creature (obviously from another universe or dimension) flies out of a swimming pool, and makes a comment about how "low res" everything is. I took it to imply that he came from some universe / simulation where there were more CPU cycles in the universe to generate a higher resolution reality. -- 1/20/97 Bio-Mimetic Computer Software I had the thought that some of the trends in materials engineering might give some direction to current software trends. One trend in materials engineering is bio-mimetic materials. In nature strong materials are often flexible and have rounded shapes - there are no square corners in nature. Bones are light and rounded and many strong natural structures have some flexibility. Whereas many engineered structures are made of hard brittle materials. Could a similar concept be applied to software engineering? Are our current programs made of strong, rigidly structured logic with sharp corners and therefore necessarily brittle? Are biological software systems like biological materials? Are these systems strong, lightweight, yet flexible? I think we have reason to believe that nature would carry these "design principles" forward from materials to software and neural circuitry design. It might be useful conceptually, to look at these analogies between human engineered and the equivalent (functionally) biologically engineered materials. Are there any useful concepts in the whole concept that typical human engineered materials are bulk machined instead of built molecule by molecule. Or is it more of a design issue, we humans just have the wrong way of thinking about building things - top down, the same processes used over and over again. If we followed more of nature's lead (GA's with molecular manufacturing) would there be a big win. Of course this whole software concept applies to hardware. Look at how biological system build computer systems - flexible, adaptable hardware - vs. our rigid hardware system with typically fixed wiring. -- 10/17/96 What is Consciousness? I thought that the Discover Magazine article in the November 1996 issue, pg. 89-98 had some very interesting points to make. It talked about experiments with the mark test. This is when a red dot is put in the animal's face when it is anesthetized. It is then placed in front of a mirror to see if it recognizes that the animal in the mirror is itself. It seems as though, in primates, only the great apes have passed the test: chimpanzees, orangutans, and one gorilla (Koko). However, other experiments, show that these great apes have other deficiencies. They fail other tests, like when presented with two over turned cups, one with food, and one without. When a human experimenter points at the "winning" cups, the do no better than pure chance. In another experiment the experimenter's gaze is substituted for pointing, the great apes still fail this test , where human babies understand that the experimenter is pointing to the winner. The article brings up the issues of self-awareness and mental models. It suggests that while the great apes have a mental model of themselves, they do not have an model of other's intentions. But, by age 4 or 5, humans have this model. It is suggested that other animals like monkeys, don't even have a model of their own intentions, they just do. This sounds very much like Rod's stimulus response model. I think one way to understand this is the point about humor. A monkey enjoys spitting water at someone, but doesn't have a model of the fact that spitting water at someone embarrasses them. It is more of a stimulus response thing, see someone, spit water, be happy. It is an automatic kind of response there is not planning or model building, it is just done. Like the described swinging from branch to branch in the trees - it is all an automatic response - the eyes see and the muscles react. To me the points in the article really bring the issues of consciousness into focus, they start to create a framework for hanging things on. Is consciousness like the A-B brain stuff Minsky talked about or is it just simple planning and model building capability? The three levels presented in this article are interesting: 1) automatic stimulus / response - no conscious planning of actions 2) model of oneself - planning of actions based on that model 3) model of oneself and others - planning of actions based on those models -- 10/17/96 A Reproductive Advantage? A interesting link... Some recent work on auto-immune diseases has suggested that people who eat certain proteins can gain relief from their diseases. This is based on research which has shown that there is a mechanism by which the body suppresses immune reactions to ingested proteins. This allows us to eat meat without having an immune reaction to the proteins in the meats. On a related note, I wonder if any of the problems with infertility have to do with the lifetime of sperm. I would assume that one of the mechanisms in a woman's body which kills sperm is an immune reaction. Do I need to go further? This would suggest a possible reproductive advantage. If my premises are correct, then this could be verified through research. -- 8/27/96 Understanding the Brain I don't think that people have thought enough about the software / hardware issue when observing the brain using electrodes or FMRI. I guess they are lucky that the brain distributes both its memory and processing among neurons and that those neurons seem to encode pretty much the same thing all the time. The problem is that in my mind, I think what if someone was given a standard computer without a video card. If you knew nothing about it and it was running, what could you figure out. Would you be able to decode the software? (Would compiled code be better or worse?) The data buses would always have data representing different things on them. It just seems like it would be very difficult to peer into the brain of a regular computer and see what is going on? Maybe looking at the address bus would tell you what routine it was in. But you'd also have to distinguish between data and code fetches? I just think similar problems are there when trying to figure out what the brain is doing. Maybe people who do reverse engineering would be better than those who don't. -- 1/4/96 Building a Better Cockroach This is a more detailed continuation of a previous insight. This one has to do with how complex spiders and other insects are. The fact is that we still have not built anything as complex as a cockroach. Why? I think part of it is that sensors and actuators are free in biotechnology and brains are expensive. This may mean that the Rod's ideal of embodied intelligence is closer to the truth, it is the complexity of the sensors and actuators that make a bug such a complex creature. So is this the major limitation to making a bug, that we can't make that complex a sensor / actuator system? Or is it that we measure the computing capability of our computers incorrectly? (This is an old idea of mine.) Even though are computers are theoretically faster and more complex than the brain of a bug, could we actually simulate that brain with sufficient detail to make it work in real time. Or are these observations only true for biological creations and not true for silicon ones, are we being lead astray by our need for an existence proof and example of what we are building in the biological world... -- 11/19/95 Environment and Genome One interesting idea that has been around for a while is that the environment of a creature affects the complexity of its behaviors. But I think that this can be taken further in two ways. In the first way, if a creature modifies his environment to remind himself of something, then does that part of the environment, in effect, become part of the creature. I think this carries over to sticky notes to remind ourselves of things - here this object becomes a part of our memory. I also applies to tools and clothing and shelter - some of these things become such an integral part of us that we can not survive without them. In my mental picture, these artifacts are a part of who we are, because they affect our actions, extend our memory. The second way is via learning. I am especially talking about things like the vision system organizing itself early in life and then freezing. Here the organism has not explicitly coded part of the brain structure in the creature's genome, but in its environment. Obviously this makes the creature able to evolve more quickly and makes it somewhat more adaptable. It is just such an interesting thought that part of the creature's effective genome is encoded in its environment. -- 11/19/95 Consciousness Do I have any good thoughts here? I guess it seems like the question to answer here is what purpose does consciousness serve? Why is it there? I wonder if it is akin to some of Minsky's A-brain B-brain stuff where two brains observe each other to check each other. I think something like that happens when people learn sports. They consciously go through the motions, then another part of the brain observes it and learns it - coding up a macro for it the next time it needs to be done. I think this says something interesting about brain structure. There may not be explicit communication between modules, but rather indirect communication through the environment, or learning by doing. And there may just be some automatic processes which run and help codify, make short cuts, etc., in the background and these probably run at all levels of cognition from low level motor skills to high level problem solving. Could consciousness be just some fiction created on top of the mind? Possibly, but I am not sure what purpose it would serve then. If it was some isolated thing that was around to help set a direction, that might be a reasonable possibility. I think that a thing that is missing is a better definition of what is considered a conscious process in a human. Is it only when we talk to ourselves in our brain? I mean we are so disconnected from what we do and why we do it, that I'm not sure what is conscious and what is not. When I am typing this I don't think about the words I am going to say or how to type it, I often don't even think of the concepts I am going to present, it just pours forth - is it just that everything is so tightly integrated that I don't think about it - and that's OK, low level functions should be automatic, or is it that consciousness is just this little ephemeral thing sitting on top somewhere that has very little say about how things are done and just has some final say on what is done. How can we exist as creatures and have so little knowledge of what goes on below? How does a chess master know that he can trust his memory? Why do I believe that I will know how to read tomorrow - I don't know how the information or processes are organized? How can I trust something I don't understand or didn't explicitly build? What is the little part that is consciousness? I guess the big difference is how I interface with and outside machine - but maybe there isn't so much difference, I don't have to know how my computer works in every detail to use it, and just like I back up my computer, I may back up my memory on paper. But I don't back up my ability to read and write, there are many questions there.... It may be possible to resolve the two viewpoints of consciousness as all controlling or consciousness as an artifact by the following analogy. What if we picture consciousness as the CEO of a big company. From they outside it may seem that the CEO is all powerful - but from another viewpoint the CEO is just a man. All the decisions he makes are based on information gathered by the marketing department and are executed by production. The CEO, like consciousness, is the final decision maker, but can look in detail at something if necessary. He may not how to do certain low level functions at the company. The interesting thing about this analogy is whether, more of the corporate structure analogy also applies, are there submanagers (middle level managers) that also have decision making ability, but are some how distinct from the CEO - are there other consciousness which the main consciousness is not directly aware of, are those the agencies communicated with? Is hypnosis when an outside force can take control away from the CEO? Is it part of the key to consciousness? Your have to admire the human brain - all the fail safes - the fact that people can be hypnotized and then get back control - the organization of data so old stuff gets thrown away and new stuff gets inserted. Has anyone ever looked at the failure rates among humans? Is it 1% or worse? How many people have cognitive or neurological disorders - I bet the number is quite high.. -- 11/19/95 Dealing with Another's Subconscious All people have many facets and aspects to their personality, some of which they are aware of, some of which they are not. However, all of these aspects affect their actions. A person may not be aware, himself, that a certain aspect of his personality is affecting his actions. However, this aspect may be observable to other people. It is also possible to understand the motivations of this aspect and deal with it directly. In this way you can affect another's actions without them knowing it and you may be able to control parts of their personality and actions which they themselves cannot control. -- 10/22/95 Dumb Luck? One theory I have abut luck relates to the previous idea about the subconscious. I believe that there are forces at work determining a person's actions which they themselves are not aware of or control of. In the case of "luck", I think that in certain cases, people who are lucky are actually planning or making the lucky things happen on a subconscious level. To them it just seems like a present, good things are always happening, but at a subconscious level some agency is hard at work planning things. -- 10/22/95 Corporate Accounting One of the important things for a company to do is to get its accounting straight and done in the proper way early on, no matter how much of a pain in the ass it may seem. -- 10/22/95 Management Skill Is managing something that can really be taught in the classroom? Or is it something innate, or is it something that can only be learned through experience? -- 10/22/95 Corporate Databases A good database is a very important part of the company. Everyone needs access to the centralized information. -- 10/22/95 Watching a Strange Bug In bio-technology actuators are free and compute power is expensive. Joe and I found the bug at the IS Robotics office. It had very long back legs which it normally didn't use, but it would sacrifice them in one big jump if it needed to escape. The thought was reinforced when Joe and I visited the National Museum of Natural History in D.C. We saw a larger than life model of a spider. It had hundreds of sensors on each leg alone... -- 9/22/95 Comparing Audio Compression to Image Compression If the visual world was like sound, everything would be transparent and have a light texture on it. -- 9/22/95

Return Home