from http://www.newscientist.com/newsletter/features.jsp?id=ns22962
Tripping on hallucinogenic drugs reveals more about our inner selves than the hippies ever bargained for, says Dana Mackenzie
IN A DORM ROOM dimly lit by a lava lamp, a freshman awaits the beginning of his first LSD trip. Slowly, the walls come alive and begin to dance with colour. And then he sees whirling spirals of stars that disappear into the distance. A network of cobwebs that grows across the room. An infinite subway tube, surrounded by fluorescent lights...
Across campus, his science teachers experience their own psychedelic visions--but without resorting to illegal mind-altering substances. Jack Cowan, a mathematician and neuroscientist at the University of Chicago, has built a neural network so powerful it can trip out. His computer's hallucinations match with almost spooky accuracy the visions of acid trippers, shamans and seers--visions that have always been interpreted as revelations from a transcendental consciousness.
Now, after more than two decades, Cowan and his team think they have found where hallucinations really come from. And there's nothing transcendental about it. An LSD trip is really a journey into the brain, says Cowan. "It's just the innate tendency of the brain to make patterns when it goes unstable."
Cowan's goal is to find out how the brain makes sense of the visible world--not when we're tripping, but under ordinary circumstances. In the process, he may learn how it breaks down in other extraordinary conditions, such as migraine headaches. Hallucinations could even offer a route to the more profound depths of the mind, to emotions and conscious thought.
Hallucinations seem to come in an endless variety, as individual as dreams. So it seems improbable that they can even be categorised, never mind calculated by a computer. But in the 1920s, Heinrich Klüver, a neuroscientist at the University of Chicago, discovered they did indeed fall into a number of distinct categories. Klüver interviewed dozens of people who had taken the drug mescaline, and even took it himself. Keeping a commendably straight head, Klüver eventually saw patterns in the patterns.
In the earliest stages of a trip, most subjects reported seeing abstract, geometrical images. Other writers have noted the same thing. "The typical mescaline or lysergic acid experiment begins with perceptions of coloured, moving, living geometrical forms," wrote Aldous Huxley in 1954 in Heaven and hell. "In time, the pure geometry becomes concrete, and the visionary perceives, not patterns, but patterned things, such as carpets, coverings, mosaics." Klüver classified these patterns into four types or "form-constants": tunnels, spirals, cobwebs and honeycombs.
Unlike Huxley and Klüver, Cowan has never sampled the drugs he studies. "I feel bad about it," he says. "I have to rely on all these reports in the literature." He also hears plenty of personal accounts from students and others who attend his lectures. "Some people see these illusions when they're going to sleep or waking up," Cowan says. "People have seen them after taking anaesthetics. People claim to see them when they meditate, or have so-called near-death experiences." Cowan believes that the "tunnel of light" illusion commonly reported in near-death experiences is simply the first of Klüver's four form-constants.
Cowan was turned on to the study of hallucinations from an unexpected direction. In 1977 he was working on pattern formation with graduate student Bard Ermentrout when he stumbled across illustrations of Klüver's patterns. "We saw immediately that the hallucination patterns were similar to convection patterns," says Cowan.
The convection of hot water involves a delicate interplay of forces. When a pan of water is heated from below, the hot water at the bottom is more buoyant than the water above, and tries to rise. If the temperature difference is not too great, the lower layer sheds its heat by diffusion before it can rise very far, so the water remains stable. But at a certain critical temperature, diffusion is not enough to cool off the lower layer, so plumes of hot water start to rise. Between each pair of rising plumes, cold water descends, so a pattern spontaneously emerges: rolling tubes of water that form parallel stripes, or square or hexagonal cells. Cowan guessed that hallucinations must also be spontaneous patterns of activity produced by two competing forces--this time in the brain. One, like the water's buoyancy, tends to excite neurons while the other, like the diffusion of heat, tends to calm them down. He speculated that this could happen in the primary visual cortex, sometimes called V1. This is a layer of tissue two to three millimetres thick at the back of the brain which serves as the first layer of processing for images gathered by the retina.
To test their idea, Ermentrout and Cowan developed a mathematical model of V1 and gave it a dose of virtual LSD. Their model reflects the fact that each neuron tends to excite its neighbours and inhibit those a little farther away. Then when the eye sees a large, featureless object, like a big red blob of paint, every neuron in the middle of the image will be excited by nearby neurons and inhibited by those farther away. So it receives no net input from other neurons. It's the brain's way of saying, "There's nothing interesting happening here."
LSD upsets this balance. One of the effects of the drug is to allow neurons to fire when there is nothing in the visual field. Ordinarily, a neuron won't start firing unless the input from the retina and from neighbours exceeds a critical threshold. This ensures that if a neuron fires by mistake, it won't convince its neighbours to fire and the activity dies out. But drugs can lower the threshold--LSD does it by making the brainstem secrete less of the inhibitory chemical serotonin. If the threshold is lowered far enough, then excitation starts to beat inhibition, and spontaneous waves of activity form in the brain. It's like turning up the heat under the pan of water. The first patterns that form will be the same ones that are seen in the water: parallel stripes, checkerboards and hexagons.
So why don't LSD users see parallel stripes across their visual field? Because these patterns are in the cortex, not the retina, Cowan reasoned. A lot of cortical real estate is devoted to objects close to the centre of the field of vision, where our sight is sharp, while relatively little is used for peripheral vision. Mapped onto the cortex, an ordinary scene is grossly distorted: objects near the centre loom large, taking up most of the brain area. When you run this distortion backwards, evenly spaced parallel lines in the cortex appear sucked together into the centre of the visual field, creating the visual impression of either a spiral or a tunnel. The regular checkerboard and hexagon patterns turn into spiralling squares or hexagons.
So more than half a century after Klüver set out his form-constants, two of them were finally explained. LSD users see spirals and tunnels because those are the real-world objects that fit the patterns of neural firing in their cortex. Timothy Leary, the guru of "tune in, turn on, drop out" fame, speculated in The Psychedelic Experience, "These visions might be described as pure sensations of cellular and sub-cellular processes." So just as Leary guessed, the spaced-out brain is tuning into its own architecture.
But what about the other two form-constants, the cobweb and honeycomb illusions? These are both lacy, filigree patterns, while water boils in fat rolls, so it's obvious the convection analogy won't work here. Cowan was confidant that his theory would provide the framework to understand these hallucinations, too.
In the 1980s, it became clear that the neurons in V1 are not sensitive simply to the position of an image on the retina. Most of them are sensitive to edges, firing if they sense an edge passing through a particular point in the visual field but remain silent if that point is similar to its surroundings. These cells are arrayed in little patches called hypercolumns that represent a particular part of space (see Diagram, left). Within the hypercolumn, each neuron responds to an edge at a slightly different orientation.
Instead of signalling to their neighbours in the same hypercolumn, these neurons contact their counterparts in different columns, which represent similar orientations in slightly different parts of space. Then, if there really is an edge, neurons with the right orientation excite each other, so the brain is more likely to detect it.
These long-range connections seemed essential to understanding the last two hallucination types, but they added a new level of complexity to Cowan's mathematical model of the cortex. Hot water was no longer a good analogy, because the forces at work there--buoyancy and viscosity--are all short range. Now equations were needed to describe something long range and direction-sensitive. The maths turned out to be like those of a hot gas in a magnetic field.
Cowan and his graduate student Matthew Wiener programmed in these equations, and found many possible waveforms could result. But they couldn't tell which of these patterns would be the first to appear spontaneously. They needed someone who could combine an expert's understanding of quantum mechanics and neuroscience, and in 1998, Cowan found just the person. Paul Bressloff of Loughborough University in Leicestershire had trained as a specialist in quantum gravity, then taken a detour into neural networks. In a few months of intense work at Chicago, he helped Cowan and Marty Golubitsky of the University of Houston work out the waves of activity that should emerge spontaneously among orientation-sensitive cells. The results appeared earlier this year in Philosophical Transactions of the Royal Society (vol 356, p 1).
The winning patterns were those in which the edges naturally close up into small square or hexagonal cells. Cowan's theory precisely reproduces Klüver's two missing form-constants. When the fine-edged squares and hexagons on the cortex are filtered back through the retinal map, they look like lacy cobwebs and honeycombs.
So far so good. But has Cowan done any more than confirm a wiring pattern for the brain that neuroscientists had already worked out? He points out that to understand how the brain works, we need more than wiring: we have to know how these circuits actually behave.
In fact, Cowan's model does hint at this. One unexpected outcome is that subtle changes in the wiring of the model brain can cause significant changes to its preferred hallucination patterns. For example, if the long-range connections in the model always run between edge neurons that represent identical orientations, would generate hallucinations resembling herringbone twill. Clearly our brains are not wired this way; if they were, who knows what effect psychedelic visions of tweed blazers might have had on 1960s fashion. To produce cobwebs and hexagons, we actually need the connections to be a little more slapdash. Perhaps the human edge- detection system is wired this way because it helps us spot small, closed contours.
On the other hand, the herringbone patterns may emerge if the chemical stimulation is changed. Perhaps the theory can explain other kinds of visual disturbances that were thought to be unrelated to LSD hallucinations, such as the auras and zigzag patterns seen by people suffering a migraine attack. If so, it could tell us what changes in the brain cause migraines, and perhaps set us on course for a cure.
Lurking in the background is the much bigger issue of where the mind comes from. To what extent is the mind, and all the rich variety of inner experiences that gives us a sense of self, simply a product of physiological processes in the brain? Hallucinations could be a perfect place to start answering this question.
The apostles of the psychedelic sixties scorned the scientific approach to understanding an LSD trip. "Bobbing around in this brilliant, symphonic sea of imagery is the remnant of the conceptual mind," Leary wrote. "On the endless watery turbulence of the Pacific Ocean bobs a tiny open mouth, shouting (between saline mouthfuls), 'Order! System! Explain all this!'" To appreciate a hallucination, Leary said, you have to let go of the urge to rationalise it.
Tom Wolfe pitched in with The Electric Kool-Aid Acid Test. "The White Smocks liked to put it into words, like hallucination and dissociative phenomena. They could understand the visual skyrockets. Give them a good case of an ashtray turning into a Venus flytrap or eyelid movies of crystal cathedrals, and they could groove on that... That was swell. But don't you see?--the visual stuff was just the décor with LSD... The whole thing was ... the experience ... this certain indescribable feeling ... The experience of the barrier between the subjective and the objective, the personal and the impersonal, the I and the not-I disappearing ... that feeling!"
Cowan makes no apologies for being one of the White Smocks. He thinks that the "visual skyrockets" and that "certain indescribable feeling" are part and parcel of the same experience. As the drug penetrates to deeper and deeper areas of the brain--visual layers, cognitive layers, emotional layers and, finally, whatever part of the brain gives us our sense of self-awareness--our subjective experience becomes enormously more complicated and richer. And yet what's going on at the cellular level may not be so different at each layer.
"Does that mean that everything can be observed and described?" Cowan asks. "I happen to believe the answer is yes. I don't think there's anything in the brain that science can't ultimately deal with." But the answers aren't going to come along tomorrow. "There are a hundred vision chips, a hundred sound chips. We now understand a bit more about one of the vision chips," he says. Cowan is already planning to look at other aspects of visual hallucinations, such as texture and size perception.
Journeying deeper still into the mind might not be much harder. The neocortex, the layer of the brain that includes V1, is the part that evolved most recently. It is also the part that supposedly makes humans so intelligent. Because it hasn't been around long, its cells are all structurally quite similar, even if their functions are quite different. "The reason this is a note for optimism," says Gary Blasdel of Harvard University, "is that when you really understand the operations that go on in a particular cortical area, it will generalise to other areas." Cowan's computerised visions might just be the beginning of a really cool trip.
Further reading:
The Doors of Perception by Aldous Huxley, Flamingo (1994)
"Hallucinations" by R. K. Siegel, Scientific American, vol 237, p 132 (1977)
Dana Mackenzie is a mathematics and science writer based in Santa Cruz, California
Tripping on hallucinogenic drugs reveals more about our inner selves than the hippies ever bargained for, says Dana Mackenzie
IN A DORM ROOM dimly lit by a lava lamp, a freshman awaits the beginning of his first LSD trip. Slowly, the walls come alive and begin to dance with colour. And then he sees whirling spirals of stars that disappear into the distance. A network of cobwebs that grows across the room. An infinite subway tube, surrounded by fluorescent lights...
Across campus, his science teachers experience their own psychedelic visions--but without resorting to illegal mind-altering substances. Jack Cowan, a mathematician and neuroscientist at the University of Chicago, has built a neural network so powerful it can trip out. His computer's hallucinations match with almost spooky accuracy the visions of acid trippers, shamans and seers--visions that have always been interpreted as revelations from a transcendental consciousness.
Now, after more than two decades, Cowan and his team think they have found where hallucinations really come from. And there's nothing transcendental about it. An LSD trip is really a journey into the brain, says Cowan. "It's just the innate tendency of the brain to make patterns when it goes unstable."
Cowan's goal is to find out how the brain makes sense of the visible world--not when we're tripping, but under ordinary circumstances. In the process, he may learn how it breaks down in other extraordinary conditions, such as migraine headaches. Hallucinations could even offer a route to the more profound depths of the mind, to emotions and conscious thought.
Hallucinations seem to come in an endless variety, as individual as dreams. So it seems improbable that they can even be categorised, never mind calculated by a computer. But in the 1920s, Heinrich Klüver, a neuroscientist at the University of Chicago, discovered they did indeed fall into a number of distinct categories. Klüver interviewed dozens of people who had taken the drug mescaline, and even took it himself. Keeping a commendably straight head, Klüver eventually saw patterns in the patterns.
In the earliest stages of a trip, most subjects reported seeing abstract, geometrical images. Other writers have noted the same thing. "The typical mescaline or lysergic acid experiment begins with perceptions of coloured, moving, living geometrical forms," wrote Aldous Huxley in 1954 in Heaven and hell. "In time, the pure geometry becomes concrete, and the visionary perceives, not patterns, but patterned things, such as carpets, coverings, mosaics." Klüver classified these patterns into four types or "form-constants": tunnels, spirals, cobwebs and honeycombs.
Unlike Huxley and Klüver, Cowan has never sampled the drugs he studies. "I feel bad about it," he says. "I have to rely on all these reports in the literature." He also hears plenty of personal accounts from students and others who attend his lectures. "Some people see these illusions when they're going to sleep or waking up," Cowan says. "People have seen them after taking anaesthetics. People claim to see them when they meditate, or have so-called near-death experiences." Cowan believes that the "tunnel of light" illusion commonly reported in near-death experiences is simply the first of Klüver's four form-constants.
Cowan was turned on to the study of hallucinations from an unexpected direction. In 1977 he was working on pattern formation with graduate student Bard Ermentrout when he stumbled across illustrations of Klüver's patterns. "We saw immediately that the hallucination patterns were similar to convection patterns," says Cowan.
The convection of hot water involves a delicate interplay of forces. When a pan of water is heated from below, the hot water at the bottom is more buoyant than the water above, and tries to rise. If the temperature difference is not too great, the lower layer sheds its heat by diffusion before it can rise very far, so the water remains stable. But at a certain critical temperature, diffusion is not enough to cool off the lower layer, so plumes of hot water start to rise. Between each pair of rising plumes, cold water descends, so a pattern spontaneously emerges: rolling tubes of water that form parallel stripes, or square or hexagonal cells. Cowan guessed that hallucinations must also be spontaneous patterns of activity produced by two competing forces--this time in the brain. One, like the water's buoyancy, tends to excite neurons while the other, like the diffusion of heat, tends to calm them down. He speculated that this could happen in the primary visual cortex, sometimes called V1. This is a layer of tissue two to three millimetres thick at the back of the brain which serves as the first layer of processing for images gathered by the retina.
To test their idea, Ermentrout and Cowan developed a mathematical model of V1 and gave it a dose of virtual LSD. Their model reflects the fact that each neuron tends to excite its neighbours and inhibit those a little farther away. Then when the eye sees a large, featureless object, like a big red blob of paint, every neuron in the middle of the image will be excited by nearby neurons and inhibited by those farther away. So it receives no net input from other neurons. It's the brain's way of saying, "There's nothing interesting happening here."
LSD upsets this balance. One of the effects of the drug is to allow neurons to fire when there is nothing in the visual field. Ordinarily, a neuron won't start firing unless the input from the retina and from neighbours exceeds a critical threshold. This ensures that if a neuron fires by mistake, it won't convince its neighbours to fire and the activity dies out. But drugs can lower the threshold--LSD does it by making the brainstem secrete less of the inhibitory chemical serotonin. If the threshold is lowered far enough, then excitation starts to beat inhibition, and spontaneous waves of activity form in the brain. It's like turning up the heat under the pan of water. The first patterns that form will be the same ones that are seen in the water: parallel stripes, checkerboards and hexagons.
So why don't LSD users see parallel stripes across their visual field? Because these patterns are in the cortex, not the retina, Cowan reasoned. A lot of cortical real estate is devoted to objects close to the centre of the field of vision, where our sight is sharp, while relatively little is used for peripheral vision. Mapped onto the cortex, an ordinary scene is grossly distorted: objects near the centre loom large, taking up most of the brain area. When you run this distortion backwards, evenly spaced parallel lines in the cortex appear sucked together into the centre of the visual field, creating the visual impression of either a spiral or a tunnel. The regular checkerboard and hexagon patterns turn into spiralling squares or hexagons.
So more than half a century after Klüver set out his form-constants, two of them were finally explained. LSD users see spirals and tunnels because those are the real-world objects that fit the patterns of neural firing in their cortex. Timothy Leary, the guru of "tune in, turn on, drop out" fame, speculated in The Psychedelic Experience, "These visions might be described as pure sensations of cellular and sub-cellular processes." So just as Leary guessed, the spaced-out brain is tuning into its own architecture.
But what about the other two form-constants, the cobweb and honeycomb illusions? These are both lacy, filigree patterns, while water boils in fat rolls, so it's obvious the convection analogy won't work here. Cowan was confidant that his theory would provide the framework to understand these hallucinations, too.
In the 1980s, it became clear that the neurons in V1 are not sensitive simply to the position of an image on the retina. Most of them are sensitive to edges, firing if they sense an edge passing through a particular point in the visual field but remain silent if that point is similar to its surroundings. These cells are arrayed in little patches called hypercolumns that represent a particular part of space (see Diagram, left). Within the hypercolumn, each neuron responds to an edge at a slightly different orientation.
Instead of signalling to their neighbours in the same hypercolumn, these neurons contact their counterparts in different columns, which represent similar orientations in slightly different parts of space. Then, if there really is an edge, neurons with the right orientation excite each other, so the brain is more likely to detect it.
These long-range connections seemed essential to understanding the last two hallucination types, but they added a new level of complexity to Cowan's mathematical model of the cortex. Hot water was no longer a good analogy, because the forces at work there--buoyancy and viscosity--are all short range. Now equations were needed to describe something long range and direction-sensitive. The maths turned out to be like those of a hot gas in a magnetic field.
Cowan and his graduate student Matthew Wiener programmed in these equations, and found many possible waveforms could result. But they couldn't tell which of these patterns would be the first to appear spontaneously. They needed someone who could combine an expert's understanding of quantum mechanics and neuroscience, and in 1998, Cowan found just the person. Paul Bressloff of Loughborough University in Leicestershire had trained as a specialist in quantum gravity, then taken a detour into neural networks. In a few months of intense work at Chicago, he helped Cowan and Marty Golubitsky of the University of Houston work out the waves of activity that should emerge spontaneously among orientation-sensitive cells. The results appeared earlier this year in Philosophical Transactions of the Royal Society (vol 356, p 1).
The winning patterns were those in which the edges naturally close up into small square or hexagonal cells. Cowan's theory precisely reproduces Klüver's two missing form-constants. When the fine-edged squares and hexagons on the cortex are filtered back through the retinal map, they look like lacy cobwebs and honeycombs.
So far so good. But has Cowan done any more than confirm a wiring pattern for the brain that neuroscientists had already worked out? He points out that to understand how the brain works, we need more than wiring: we have to know how these circuits actually behave.
In fact, Cowan's model does hint at this. One unexpected outcome is that subtle changes in the wiring of the model brain can cause significant changes to its preferred hallucination patterns. For example, if the long-range connections in the model always run between edge neurons that represent identical orientations, would generate hallucinations resembling herringbone twill. Clearly our brains are not wired this way; if they were, who knows what effect psychedelic visions of tweed blazers might have had on 1960s fashion. To produce cobwebs and hexagons, we actually need the connections to be a little more slapdash. Perhaps the human edge- detection system is wired this way because it helps us spot small, closed contours.
On the other hand, the herringbone patterns may emerge if the chemical stimulation is changed. Perhaps the theory can explain other kinds of visual disturbances that were thought to be unrelated to LSD hallucinations, such as the auras and zigzag patterns seen by people suffering a migraine attack. If so, it could tell us what changes in the brain cause migraines, and perhaps set us on course for a cure.
Lurking in the background is the much bigger issue of where the mind comes from. To what extent is the mind, and all the rich variety of inner experiences that gives us a sense of self, simply a product of physiological processes in the brain? Hallucinations could be a perfect place to start answering this question.
The apostles of the psychedelic sixties scorned the scientific approach to understanding an LSD trip. "Bobbing around in this brilliant, symphonic sea of imagery is the remnant of the conceptual mind," Leary wrote. "On the endless watery turbulence of the Pacific Ocean bobs a tiny open mouth, shouting (between saline mouthfuls), 'Order! System! Explain all this!'" To appreciate a hallucination, Leary said, you have to let go of the urge to rationalise it.
Tom Wolfe pitched in with The Electric Kool-Aid Acid Test. "The White Smocks liked to put it into words, like hallucination and dissociative phenomena. They could understand the visual skyrockets. Give them a good case of an ashtray turning into a Venus flytrap or eyelid movies of crystal cathedrals, and they could groove on that... That was swell. But don't you see?--the visual stuff was just the décor with LSD... The whole thing was ... the experience ... this certain indescribable feeling ... The experience of the barrier between the subjective and the objective, the personal and the impersonal, the I and the not-I disappearing ... that feeling!"
Cowan makes no apologies for being one of the White Smocks. He thinks that the "visual skyrockets" and that "certain indescribable feeling" are part and parcel of the same experience. As the drug penetrates to deeper and deeper areas of the brain--visual layers, cognitive layers, emotional layers and, finally, whatever part of the brain gives us our sense of self-awareness--our subjective experience becomes enormously more complicated and richer. And yet what's going on at the cellular level may not be so different at each layer.
"Does that mean that everything can be observed and described?" Cowan asks. "I happen to believe the answer is yes. I don't think there's anything in the brain that science can't ultimately deal with." But the answers aren't going to come along tomorrow. "There are a hundred vision chips, a hundred sound chips. We now understand a bit more about one of the vision chips," he says. Cowan is already planning to look at other aspects of visual hallucinations, such as texture and size perception.
Journeying deeper still into the mind might not be much harder. The neocortex, the layer of the brain that includes V1, is the part that evolved most recently. It is also the part that supposedly makes humans so intelligent. Because it hasn't been around long, its cells are all structurally quite similar, even if their functions are quite different. "The reason this is a note for optimism," says Gary Blasdel of Harvard University, "is that when you really understand the operations that go on in a particular cortical area, it will generalise to other areas." Cowan's computerised visions might just be the beginning of a really cool trip.
Further reading:
The Doors of Perception by Aldous Huxley, Flamingo (1994)
"Hallucinations" by R. K. Siegel, Scientific American, vol 237, p 132 (1977)
Dana Mackenzie is a mathematics and science writer based in Santa Cruz, California
![:\](https://bluelight.org/xf/BL_Images/Orig_Emoji/wrygrin.gif "wry grin :\")