valich said:
...the world doen't necessarily have to be a "simulated world," with or without GFW. In what sense are you suggesting that it is simulated?
Since you asked, I post the paper to explain. I wish I could make it an attacment as it is long but have wasted more than one hour trying to attach. Sorry about the length.
Genuine Free Will is Possible
Before the advent of Quantum Mechanics, the future appeared to LaPlace to be exactly determined by the past state of the universe, even if it was clearly unpredictable. Chaos theory and measurement errors plus ignorance about small asteroid orbits, rupture stresses in tectonic faults or vascular systems, etc. makes LaPlace’s future unpredictable, perhaps fatally so in only a few seconds for some individuals. Quantum Mechanics destroyed LaPlace’s deterministic world. Thus, thanks to QM, a “probabilistic will” is at least possible. I.e. we can have the illusion of making “choices” that are actually made by the chance results of QM; however, Genuine Free Will, GFW, i.e. real choices made by one’s self, still appears to be impossible without some violation the physical laws that govern molecular interactions in our complex neuro-physiological processes.
If GFW does not exist, it is perhaps the most universal of all human illusions. This article will show that GFW is physically possible, even probable, without any violation of physics if one is willing to drastically revise the usual concept of one’s self. Furthermore, it argues that the required revision is a natural consequence of a better understanding of how the human visual system functions and the fact that we are highly visual creatures. The possibility that GFW is only an illusion is not excluded, but is made less probable, by the arguments presented in “Reality, Perception, and Simulation: A Plausible Theory”1. This text extracts from that article some aspects related to the existence of GFW.
That article focused almost entirely on the human visual system. How the brain uses a 2 dimensional array of information (neural activity present in the retina) to form a 3 dimensional perception of the environment is the central mystery of vision. The most accepted general concept is that this 2D data array is “computationally transformed” in successive stages neural processing until the 3D perception “emerges.” This processing begins in the retina itself where data is compressed by almost 100 fold. (Retinal photo detectors greatly outnumber optical nerve fibers.) The first cortical processing area, V1, extracts some “features,” (mainly line orientations and intensities) that are present in the visual field. Color, motion, and other primary features are extracted later in entirely separate regions of the brain. These separated features are processed further in other regions of the brain, but no one knows how they are reassembled into the unified 3D perception we experience. I concur with this feature extraction and segregation model, but do not think the 3D perception we experience is the “emerging result” of automatic computational transforms of the retinal data array as the standard theory suggests.
I contend that the visual features extracted in separate regions of the brain are never “reassembled” and do not “emerge” to form our unified visual experience. Instead, I believe that currently available sensorial information and one’s memory are used to construct, probably in the parietal region of the brain, a real-time simulation of the visual world. We experience that simulated world, not the physical world. Evolutionary selection has forced this simulation to be a nearly perfect model of our immediate physical world. (Excepting electromagnetic waves and other features for which we lack neural sensors.) Thus, continuous detailed guidance is required from the senses, but hallucinations and illusions can be, and occasionally are, created in the simulation that conflict with the physical world. These “errors” together with dreams and visual images formed with eyes closed are difficult for the conventional view of vision to explain in terms of automatic transformations of retinal data. Hallucinations, visual dreams, etc. are easily understood with the concept that what we experience is an internal simulation of the world, not an emerging transform of the retinal data.
Now I reproduce three of the several arguments I presented in above cited article to support this simulation concept and to demonstration that the standard concept of 3D perception as the emerging end result of automatic neural computational transformations of retinal data is surely wrong:
1) Our visual experience is uniformly rich in details over a wide area. That is, we see / experience the environment in front of us everywhere with high resolution, but the optical system of the eyes has very low resolution, except for one extremely small (solid angle slightly more than one degree) part, the fovea. How can high-resolution perceptual experiences, spanning a large part of one hemisphere, emerge from such low resolution input data? Clearly what we experience is derived from some inter construction, not the computational transform of retinal data. I am referring to our visual experience, not our ability to perceive fine details far from the point of fixation. Our perception of fine details is limited to that part of the image falling on the fovea. If our visual experience emerged from successive neural transforms of the retina data, it would be like looking through a lightly frosted sheet of glass, which had only one small spot completely clear (without frosting).
2) One’s perceptual experience when viewing a movie can also prove that the conventional view of visual processes is simply wrong. Imagine that a motion picture camera, held by someone seated in an extreme left seat of a theater, is filming actors on the stage and that doors on the left and right sides of the stage are equally large. Now suppose that this film is projected in a movie theater from a centrally located projection booth. The image projected on the screen will have doors of different size. Assume the left one is 25% larger. If you are seated on the right side of the movie theater in a location that is exactly symmetric to the location of the camera that filmed the movie, then the two doors will form equally large images on your retinas, but the standard theory’s “automatic computational transforms” will compensate for the 25% greater distance to the screen image of the left door and you will correctly perceive that the left door image is 25% larger. (This is correction for distance called “perceptual size constancy”) 2 Likewise if a movie character, who is 80% as tall as the real stage doors, should enter right door, and exit the left one, his height should be perceived as changing by 25% as he walks from one door to the other, according to the convention theory. Regardless of where they are seated, people never perceive an actor’s height as changing as his movie image moves from one side of the screen to the other. If one of the stage actors had walked from the extreme right rear of the stage to the extreme front edge, directly towards the camera making the movie film, the size of his film image could easily have increased by 50%. When you see this walk in the movie, his image on the screen will grow larger by 50%, but its distance from you remains constant so that the automatic computational transforms applied to this retinal image also remain constant. (The convergence angles of your eyes, etc. do not change during his walk.) Consequently, if the standard theory were correct, his perceived size should increase by 50% as he walks, but his perceived size is constant regardless of where he walked on the stage. Clearly the conventional view of how visual perceptions are produced (emerging end result of automatic computational processes) is simply wrong and needs to be replaced. My JHU/APL article gives several other reasons why I believe we experience the results of an internal simulation of our environment, not automatic transforms of our retinal images. It also explains, at the neural level, how we segregate objects (parse them) from the continuous visual field and how we then identify these parsed objects, but these processes are not discussed here because this article concerns only the existence, or not, of GFW.
3) The primary task of living organisms is to stay alive, at least long enough to reproduce. Neural computations require time. The world we would experience, if our experiences were the emergent results of many successive stages of neural transformations would be delayed by a significant fraction of a second. During our evolutionary history nothing truly discontinuous ever happened in our visual environment. (The discontinuous changes in movie and TV scenes did not exist.) None the less, it was essential for our ancestors to have a real-time understanding of their surroundings despite nature’s temporal continuity and our neural delays. - Try ducking a rock thrown towards your head if your only visual experience of it is a display projected into the eyes (electronic goggles) that delay the image by 0.1 seconds! A real-time simulation of the environment, can be achieved in a neural simulation by slightly projecting ahead the sensory information to compensate for neural processing delays.
A real-time simulation would have great survival value. Perhaps the Neanderthals still experienced slightly delayed “emerging transforms” of retinal data when our smaller brained and weaker ancestors perfected a real-time simulation of their environment. (Ecological pressure from the larger and stronger Neanderthals would have accelerated the rate of evolution in our ancestors.) Likewise, the “Out of Africa” mystery, (Why one branch of hominoids, expanded and dominated all others approximately 50,000 years ago.), which is often assumed to be related to the acquisition of “autonomous language” (no gestures required - hands free and education facilitated), might better be explained by the development of the real-time simulation of the environment.
Furthermore, I think everything we perceive as being “real” in our environment, including our physical bodies, is a part of this same simulation, not an emerging result of neural transformations of sensorial data from any of our neural transducers. That is, all of the senses only guide the simulation, feature by feature, to keep it highly faithful to the current external reality. When an abrupt external event unexpectedly occurs (hidden firecracker exploding, etc.), it significantly conflicts with the events projected in our simulation for that moment. We are startled and the simulation must be quickly revised to conform to the unanticipated external reality. This revision requires approximately 0.3 seconds. I think it probable that the simulation is paused while the revision is in progress, but we do not notice as we are also “paused” during this brief interval, just as we are not aware of hours passing while we sleep. I think the unusual electrical activity in the brain associated with the re-initiation of the simulation produces the EEG signal commonly called “P300,” or the “startle spike.” P300 is strongest over the parietal region.3
Why the continuous natural environment should be dissected into “features” and separately processed as a means of achieving a unified perception of the world is a great and unexplained mystery for most cognitive scientists, but easily understood if a simulation of the world is constructed by the brain. The physically sensed world is dissected into “features” for the same reason that a pilot uses a checklist before takeoff. Dividing a complex task into its component details and separately checking each, item by item, feature by feature, improves task performance accuracy. Thus, both the real-time simulation and the dissection of the visual field into features have significant survival value and consequently are probable natural developments in the evolution of creatures as complex as man.
In order to compare the features derived from retinal data with those derived from the simulation, they must be brought to the same neural tissue. Clearly it would be advantages to make this comparison as early as possible in the sequential stages of “computational transforms” of the retinal information. If the simulation is constructed in the parietal region of the brain, then one would expect that the number of neural fiber leaving the parietal cortex and returning to the visual cortex would at least equal those coming there, via the LGN, from the eyes. In fact they are somewhat more numerous. They are called “retrograde fibers” and no plausible reason for their existence has been suggested. Some of the comparison may be made even earlier in the LGN, which is usually considered to be mainly a “relay station” between the eyes and visual cortex. (Both areas have large projections into the parietal cortex, so it can easily “know” when, where and what difference has been detected.) The quantity of retrograde fibers from the visual striate cortex to the LGN slightly outnumbers the number of fibers coming there from the eyes. About this second set of retrograde fibers, DeValois4 states: “It is by no means obvious what function is subserved by this feedback.” (from V1 to LGN) About the retrograde set from the parietal to V1, they state: “Even less is understood (if that is possible) about these feedback connections...” They also note that both sets are “strictly retinotopic,” which is the neuro-physiologist’s way to compactly state that each small part of the visual field is mapped in one-to-one correspondence with neural tissue. That is, the retrograde fibers return to the same small area of processing cells that the prograde fibers enter and these cells are concerned with only a small part of the image on the retina. This approximately equal number of retinotopic retrograde fibers entering the visual cortex, is not only explained by the theory I am suggesting; they are required for the simulation to rapidly correct for unpredictable external events!
If a buzzer sounds while one is watching the steady predictable movement of a small light spot and one is asked: “Where was the light spot when the buzzer first began to sound?” the location indicated is later than the true location. Thanks to the predictive simulation, the subject is continuously aware of the true location of the light in real-time but he only becomes aware of the sound later after the simulation has been revised to include the sound of the buzzer and he associates it with that later location of the light. Retrograde fibers project back to early sensory processing stages for all of the senses to make correction of the simulation as rapid as possible but perception of new events is still delayed enough to be easily demonstrated in this type of psychological test. - For example, a reasonable competent computer programmer can program his computer to move a light pixel across a stationary grid displayed on the monitor and to randomly make a brief sound. With fine pointer, he quickly points to the light spot location where the light was when the sound started. A few seconds later, the computer displays where the light pixel actual was when the sound started. Note how quickly he moves the pointer (his reaction time) does not matter. The delay measured is the time required to revise the simulation to include the new sound. This small revision will not produce a “P300” EEG signal because the simulation is not paused while it is made. Only major environmental discontinuities, usually sudden unexpected loud noises, pause our existence (startle us).
Thus the only reality we directly experience is this simulation and we are part of it. That is, we are an informational process in a simulation, not a physical body. When we are in deep dreamless sleep the simulation is paused and we do not exist - only our physical bodies exist. Our bodies are at all times completely governed by physical laws, like any other physical object; but if we are only an informational process in a slightly imperfect simulation of the physical world, then we need not be deterministic (or quantum mechanical) beings. That is, we may not exactly follow physical laws just as the creatures modeled in modern computers making movies, pixel by pixel, without actors or optical cameras do not exactly follow the physical laws. The meaning of symbols manipulated in a computer does not depend upon the physical construction or deterministic details of the computer. The human brain is a parallel processing computer, much more advanced than any man has yet conceived, and is fully capable of making a real-time simulation of the world we experience.
Other humans, some of the more advanced animals, and ourselves are modeled in this simulation as having wishes and making choices, not as bio-mechanical creatures governed by physics. That is, the simulation in which we live and exist assumes GFW exists for some of the more advanced creatures. Thus, GFW does exist in the only world we exist in and directly experience. Neither we, nor GFW exists in the physical world. From our direct experiences in the world we exist in, the simulation, we infer (I think correctly) that the physical world does exist, but as Bishop George Berkeley noted, the existence of a physical world may be only an erroneous belief, commonly deduced from our direct experiences. That is, the directly experienced GFW has a stronger claim to “reality” than the inferred physical world!
Summary: This definition of one’s self as an informational process in a simulation, not a physical body, permits you to have GFW and make other violations of physical laws, especially in your dreams, when sensory guidance of the simulation is weak or absent. For example, some people sincerely report “out of body” experiences etc. These physically impossible experiences and GFW are directly experienced and thus have a strong claim to being “real.” In contrast, the existence of a physical world is only inferred from these direct experiences. Bishop Berkeley argued consistently that it may not exist, but he required a God to give him his experiences. My view is similar to his in some aspects (I do not exist as a physical object in the material world.) but it makes no reference to God. Instead, a brain-based simulation is creating both my experiences and me. Being non-physical is the price one must pay for GFW if one rejects miracles that violate physics.
Question: Are you a complex bio-mechanical machine without GFW or only an informational process that has GFW in a simulated world? If you believe the former and that belief is correct, you must. I.e. you can make no real choices without postulating a “soul” or other miracles that violate physics, but if the latter is correct, I can chose to believe it (or not) and still be consistent with physics and logic. Some who believe they have free will and yet reject the second alternative of the question may find in this dilemma a strong argument for the existence of God and miracles, but if they do their “free will” is not GFW. Instead it is the potentially capricious and reversible gift of a greater being, whose postulated existence is not supported by any physical evidence. In contrast, there is a large body of physical evidence (some given above) supporting the simulated world in which I postulate we exist with GFW. See the first reference for more of this evidence. Consider also how many of the strange aspects of human psychology easily fit within the framework of a simulated world and being (phantom limbs, multiple personalities, false memories, sincere denial, déjà vu, hallucinations, etc.).
References and Notes:
1) For reprint, contact the Johns Hopkins University / Applied Physic Laboratory (helen.worth@jhuapl.edu). "Reality, Perception, and Simulation: A Plausible Theory" appeared in the JHU/APL Technical Journal, volume 15, number 2 (1994) pages 154 - 163. The last two pages (Philosophical Implications and Speculations) give the above solution to the freewill vs. determinism problem.
2) For example, if a father is standing three times farther from you than his half grown son, his image on your retina is smaller than that of his son, yet you perceive their relative sizes correctly. Standard theory suggests that we automatically correct retinal image sizes to compensate for distance. “Perceptual size constancy” is usually reasonably accurate. The most notable natural exception is the moon illusion. The near horizon moon appears to be larger than the overhead moon because humans conceive of the “sky dome”, on which the moon and stars appear to move, as more distant near the horizon than at the zenith. Why this is so, is partially caused by the slowing of the angular rate of movement of clouds, birds, etc. we watch as they move towards the horizon.
3) There are many other reasons to suspect the simulation takes palace in parietal cortex, but I will only briefly mention two. First is the geometric efficiency of the brain’s structure for a parietal simulation. The simulation requires four main inputs. Tactical sensory cortex contacts the anterior parietal; Visual cortex contacts the posterior, Auditory input contacts it laterally and the primary tissue associated with memory is directly below the parietal cortex. This minimizes neural conduction delays and “white tissue” (nerve fibers) brain volume requirements. Even stronger support is found in the sequela of parietal strokes, which result in “unilateral neglect.” Victims of these strokes do not recognize the existence of the contra lateral half of the physical world. They eat only the food on one side of their plate, etc. Their visual system can be shown to continue functioning perfectly. For example, if one briefly flashes a small light in that part of the world that does not exists (for them), and then demands that they guess whether this non-existent light was red or green, they perform far above chance, while complaining that it is silly to name the color of something that did not exist. This proves their visual system is functioning well, at least through the stage where small color features are extracted. I explain unilateral neglect sequela by postulating that the undamaged side of the parietal brain is continuing to make a simulation, but only of its half of the world. Because their personality is not drastically changed, I believe frontal cortex is utilized to construct much the “psychological self” included in the simulation but their physical body image is a parietal construct. - If they happen to turn their head and see their leg, whose existence is no longer represented in the simulation as part of their body, they may try to throw this “foreign leg” away – it is disgusting close to them.
4) Page 101 of Spatial Vision, first edition, Oxford Psychology Series No. 14, by R.L. & K.K. DeValois Oxford University Press (ISBN 0-19-505019-3)
