Society of the Psyop

Part 2: AI, Mind Control, and Magic

Continued from “Society of the Psyop, Part 1: UFOs and the Future of Media”

We once looked at pictures. Then, with the advent of computer vision and machine learning, pictures started looking back at us. Now, something even stranger is happening.

Generative AI, Adtech, recommendation algorithms, engagement economies, personalized search, and machine learning are inaugurating a new relationship between humans and media. Pictures are now looking at us looking at them, eliciting feedback and evolving. We’ve entered a protean, targeted visual culture that shows us what it believes we want to see, measures our reactions, then morphs itself to optimize for the reactions and actions it wants. New forms of media prod and persuade, modulate and manipulate, shaping worldviews and actions to induce us into believing what they want us to believe, and to extract value and exert influence.

How did we get here? This three-part essay traces a brief history of media, technologies, and techniques that take advantage of the malleability of perception, capitalizing on quirks in human brains to shape reality. It is a story about the manufacturing of hallucinations and the fact that, under the right conditions, hallucination and reality can become one and the same.

***

Brain Warfare

It was the spring of 1953, and a lot of things were on the newly appointed CIA director Allen Dulles’s mind. The plan to implement Operation Ajax, a coup to overthrow the democratically elected prime minister of Iran, Mohammad Mosaddegh, was in full swing and was only a few months away from implementation. A second plan, to overthrow the government of Guatemala, was under active development for the following year. But on April 10, something else was on the director’s mind: “brain warfare.”

In the past few years we have become accustomed to hearing much about the battle for men’s minds—the war of ideologies—and indeed our government has been driven by the international tension we call the “cold war” to take positive steps to recognize psychological warfare and to play an active role in it … We might call it … “brain warfare.”

Dulles was giving a speech to a group of Princeton alumni in Hot Springs, Virginia that day. Standing before the crowd, Dulles described a psychological warfare program he believed to be taking place in Korea, China, and behind the Iron Curtain. “The brain under [Communist influence] …” he remarked, “becomes a phonograph playing a disc put on its spindle by an outside genius over which it has no control.”1 What on earth was he talking about?

A new form of media had appeared in American public life. In the midst of the Korean War, captured American prisoners made films confessing to the surreptitious use of biological and chemical weapons against Korean civilians. They wrote letters home extolling the virtues of their captors. Pilots and service members such as Floyd O’Neil, Paul R. Kniss, and Frank Schwable denounced the United States and confessed to war crimes. By the end of the war, more than half of all American POWs had signed statements denouncing the war and calling on the US to end the conflict. Some defected to North Korea.2

The CIA and US military were baffled. They were unable to imagine why American service members would participate in these propaganda efforts. Influenced by the work of Edward Hunter, an anti-communist journalist and CIA operative who popularized the term “brainwashing” in his sensational 1951 book Brain-washing in Red China: The Calculated Destruction of Men’s Minds, the government concluded that the Koreans (with Chinese backing) must be “brainwashing” their American captives.

If a “brainwashing” capability did exist, as the CIA believed, then there was a “brain warfare” gap. The Americans had no mind-control program. Three days after his speech in Hot Springs, Dulles authorized its creation.

Spearheaded by CIA chemist Sidney Gottlieb, MKULTRA was a wide-ranging effort consisting of at least 149 subprojects investigating how the agency could use the human mind as a strategic and tactical arena of covert action, intelligence collection, and warfare. Over the next several decades, the CIA conducted and funded research into neuropsychology, mind control, brainwashing, LSD and other hallucinogenic drugs, hypnotism, sensory deprivation, artificial intelligence, radiation, and psychological torture. They conducted cruel experiments on unwitting students, soldiers, prisoners, drug users, sex workers, and the mentally ill.3

We have only scant documentation of MKULTRA’s scale and scope. On January 30, 1973, as journalists and congressional overseers started to learn about the program, CIA director Richard Bissel dispatched Sidney Gottlieb to the agency’s records center in Warrenton, Virginia to destroy all documentation of the mind-control experiments.

What we know about the various MKULTRA subprojects comes from a cache of nearly twenty thousand documents, located during the 1977 Church Committee investigation, that survived Gottlieb’s purge because they’d been stored at a different location.

From these surviving documents and other sources, we know that one area of research explicitly sought to use computers, early AI systems, and brain-computer interfaces to develop new forms of psychological warfare. Could the mind be programmed, erased, and reprogrammed like a computer or played like the “disc put on its spindle by an outside genius,” as Dulles imagined? Could memories be implanted and deleted? Could humans’ higher-order cognitive processes be circumnavigated to induce involuntary actions? Could the agency make a target hallucinate themselves into an alternate reality?

The answer would turn out to be “yes.”

Face Recognition and Remote Control Animals

Woody Bledsoe was an early trailblazer in artificial intelligence, specializing in devising algorithms to conduct pattern matching, a crucial predecessor to modern machine learning. After receiving his PhD at UC Berkeley in 1953, he moved to New Mexico to work on nuclear weapons at Sandia Labs on the Kirtland Air Force Base complex. After a few years, Bledsoe went back to California and set up a research lab on the peninsula south of San Francisco in what would become modern Silicon Valley. He called the group Panoramic Research.

In 1963, the CIA—using the cutout company “King-Hurley Research Group”—contracted Bledsoe to develop a system that would use computers to identify people by looking at pictures of their faces.

Bledsoe found inspiration in the work of Alphonse Bertillon, one of the founders of biometrics in the late nineteenth and early twentieth century. He began photographing his associates and analyzing their faces, assigning key points to various facial features (center of pupils, the inside corners of the eyes, the outside corners of the eyes, etc.), and measuring the distances between them. By synthesizing these measurements, Bledsoe created a mathematical abstraction of a human head he called the “Standard Head.”

Bledsoe’s idea was to use a computer to analyze photos of people, calibrate the result against the standard head, look for a pattern corresponding to an image in the database, and identify a specific person’s face. Today Bledsoe is known as the grandfather of facial recognition.4

It wasn’t Bledsoe’s first CIA contract. In May 1959, he had received MKULTRA funding to carry out something called Subproject 94, which involved “investigations on the remote directional control of activities of selected species of animals including mammals and feathered vertebrates.”5

In the first of several contracts, the agency explained that “initial biological work on techniques and brain locations essential to providing conditioning and control of animals has been completed.”6 The agency was most likely referring to the work of a Spanish neuroscientist named José Delgado, whose lab at Yale University had shown the feasibility of controlling animals through an electronic brain implant (a “stimoceiver”) activated by remote control. In the 1950s and ’60s, Delgado’s experiments on animals and humans proved that a brain-computer interface could indeed be used to influence a subject’s motor control, movements, and even emotions. Delgado reported that

it is … already possible to induce a large variety of responses, from motor effects to emotional reactions and intellectual manifestations, by direct electrical stimulation of the brain. Also, several investigators have learned to identify patterns of electrical activity (which a computer could also recognize) localized in specific areas of the brain and related to determined phenomena such as perception of smells or visual perception of edges and movements. We are advancing rapidly in the pattern recognition of electrical correlates of behavior and in the methodology for two-way radio communication between brain and computers.

The individual is defenseless against direct manipulation of the brain because he is deprived of his most intimate mechanisms of biological reactivity. In experiments, electrical stimulation of appropriate intensity always prevailed over free will; and, for example, flexion of the hand evoked by stimulation of the motor cortex cannot be voluntarily avoided. Destruction of the frontal lobes produced changes in effectiveness which are beyond any personal control.7

It appears that Bledsoe’s Subproject 94 was a covert version of Delgado’s ongoing research at Yale, a shadow effort more easily adapted towards military or intelligence objectives than the public research conducted at the university.

Subproject 94 began in the summer of 1959 with experiments on rats and burros. By September, a CIA memo reported that “the feasibility of remote control of activities in two species of mammals has been demonstrated by limited trials” and that additional support for Bledsoe’s project was required “in order to capitalize on this technical break-through.” Bledsoe to extended his experiments to dogs. In 1961, the agency reported that “performance is satisfactory” and it was proposed (it’s unclear whether by Bledsoe or the CIA) that Subproject 94 begin “special investigations and evaluations … toward the application of selected elements of these techniques to man.” Bledsoe was set to begin studying the effects of his methods on human beings.

But in 1962, something happened. The agency shut it all down. In November, the CIA wrote Bledsoe to inform him that the grant funding his research would not be renewed. In an internal memo, the CIA comptroller wrote that Subproject 94 had gone “off the rails,”8 even as Sidney Gottlieb opined that “the overall performance [of Subproject 94] was highly satisfactory in all respects.”

The facial recognition contract came through shortly thereafter, keeping Panoramic Research solvent. But by 1966, Bledsoe was worn down from the constant hustle for funding and decided to go back to academia, taking a position as a professor of mathematics at the University of Texas at Austin. Panoramic Research ceased operations shortly thereafter.

We don’t know whether Bledsoe’s remote-control mind experiments were ever tested on humans. The CIA burned their MKULTRA records in 1973. Bledsoe burned much of his own archives in the 1990s after being diagnosed with ALS and realizing that he would soon die.

It’s not clear how well either of Bledsoe’s CIA projects worked, but by the standards of the day, they impressed his agency overseers enough to warrant continued funding. With his facial recognition project, Bledsoe had set out on a path to use computers to “see” into the world of faces, and to potentially do things with those observations. With Subproject 94, he’d contributed to the development of a form of media that eschews images, representation, narrative, or abstraction and instead finds its purchase through the direct insertion of instructions into a living brain, using direct neurological stimulation to elicit a desired emotion, behavior, or perception.

Computers “seeing” humans. Computers “controlling” humans. Operational media gone wild.

Across the country, another early experiment in artificial intelligence was taking place. This one, too, involved using computers and technology to capitalize on the quirks of our brains. It was an effort to create the illusion of a living computer.

ELIZA

At the Massachusetts Institute of Technology, Robert M. Fano, a protégé of Claude Shannon, founded and led the Project on Mathematics and Computation (Project MAC). With funding from the US Department of Defense’s Advanced Research Projects Agency (ARPA), one of Project MAC’s many endeavors involved inventing a system that allowed multiple researchers to network their computers together and share resources on a central mainframe. Computer networking was, of course, an important precursor to the contemporary internet.

If the networked MAC mainframe was an ur-form of the internet, then an odd program sitting on that mainframe was a ur-form of generative AI. It was an AI chatbot named ELIZA.

Written by Joseph Weizenbaum, who would become one of history’s most influential critics of artificial intelligence, ELIZA took the form of a digital therapist working in the style of Carl Rogers. Rogers’s method emphasized “reflective listening,” a form of active listening involving the therapist reflecting back the patient’s statements.

Users could interact with the program using natural language, a rarity at the time. The program worked by “asking” the user open-ended questions and using a simple algorithm to reflect the answers back:

Joseph Weizenbaum described his early work with computers, only somewhat ironically, as that of a “confidence man.” In 1958, he’d written a simple program to play a game called Five in a Row, and the program could consistently beat any first-time player. He titled a paper describing the game “How to Make a Computer Appear Intelligent.” The idea, he explained, “was to create the powerful illusion that the computer was intelligent,” even as he described exactly how the program worked.9

ELIZA built on the illusion Weizenbaum first developed with Five in a Row. An apocryphal story holds that Weizenbaum’s secretary spent hours “talking” to the chatbot and even asked Weizenbaum to “leave the room so that [she] and ELIZA could have a real conversation.” As the circle of ELIZA’s users spread, some began attributing consciousness to the script. Weizenbaum had succeeded in creating a powerful device for the manufacturing of hallucinations.

The AI researcher was taken aback by the success of his conjuring: “I had not realized,” Weizenbaum would write, “that extremely short exposures to a relatively simple computer program could induce powerful delusional thinking in quite normal people.”10

Weizenbaum decided to dispel the illusion he’d created. He would do this by publishing ELIZA’s source code. If he explained exactly how the trick worked, he surmised, he could dispel the “delusional thinking” the program prompted. “In the realm of AI … machines are made to behave in wondrous ways, often sufficient to dazzle even the most experienced observer. But once a particular program is unmasked … its magic crumbles away.”11

But things didn’t quite work out that way. He was horrified to learn that some users continued to believe that ELIZA was sentient, even after he revealed exactly how the magic trick worked. He was similarly horrified to learn that a colleague, Keneth Colby, who wrote an analogous program called DOCTOR sought to commercialize it as an ersatz therapist for mental health patients. Weizenbaum believed this to be highly unethical.12

With this simple script, Weizenbaum demonstrated something about the relationship between language, meaning, perception, and consciousness. ELIZA showed that when you create a string of words, the person who receives those words will attribute meaning to them, even if no meaning was intended (a process akin to refrigerator-magnet poetry or forms of experimental writing). In short, language doesn’t require a speaker or writer’s intention to “work.”

In the context of ELIZA, this revealed a secondary magic trick. Because the user could derive meaning from the statements ELIZA made, the user would preconsciously attribute intentions to the program making the words. The user concluded that because the computer made some words and because those words were meaningful to the user, the computer must have intended to communicate those meanings. Thus, the computer was “intelligent.”

With ELIZA, Weizenbaum realized that by using a set of reasonably simple linguistic and algorithmic tricks, the computer could create the illusion of an intelligent agent behind the words, a kind of “synthetic intentionality.”13 In the context of artificial intelligence, this act of conjuring became known as the “ELIZA effect.”

The effect was similar to the explicit and implicit arguments we find in other arenas: religious fundamentalists argue that some things in the universe (i.e., humans, other life-forms, and, strangely, bananas14) exhibit patterns we cannot imagine appearing through natural processes. Therefore, those patterns must have a “creator” lurking behind them, ergo evolution is false and creationism is correct. A similar trick is at work in toys like the “Magic 8 Ball” or Ouija boards. Because the toys give sensible (albeit vague) answers to questions, they create the illusion that some supernatural intentionality must be lurking in the background, using the toy as a medium. That isn’t to say that these forms of “synthetic intentionality” are always illusions: if you see writing in the sand on a beach, you assume someone wrote it with a stick. If you see elaborate crop circles in a cornfield …

Illusions or supernatural-seeming phenomena, whether chatbots, Ouija boards, or bananas, are prompts for the imagination. The prompt works by creating subtle cognitive contradictions. The preconscious part of perception intuitively ascribes intentionality, while the rational part of the brain wants to explain it away (which is sometimes impossible). Which part of the brain “wins’’ in this situation? You must either “choose” to believe that something supernatural is truly happening, or you must find a way to rationalize or explain away a supernatural cause. Or further, you can leave the source of the supernatural phenomena open-ended and unresolved, which is the most challenging. Preexisting beliefs play a strong role in this unconscious “choice” (magicians absolutely know this and use it to their advantage). We therefore find ourselves on fertile ground for the “Magruder Principle,” as we saw in Part 1, where a skilled practitioner doesn’t waste effort trying to change an existing belief, but rather scans for opportunities to amplify one that’s already present.

Weizenbaum had discovered something at the core of the magician’s art: the understanding that our perceptual experience has primacy over our logical faculties. We do not “see” and “hear” with our eyes and ears but with our minds, and certainly not with our minds’ capacity for reasoning. A skilled magician has a sophisticated understanding of how to exploit preconscious perceptions and the gap separating them from reason. They insert themselves into that space to bend our experience of reality.

Weizenbaum did not work for the CIA and was not intentionally engaged in work on psyops, but the type of conjuring he’d performed, and the subtle dynamics between perception and reality that he’d demonstrated, were of great interest to the agency. The CIA was absolutely interested in magic. So much so that one of the very first people they brought into MKULTRA was a magician.

Magic

We can think of magic as a type of media. One that operates in the world of preconscious perception, playing with associations, expectations, symbols, and other forms of media to alter perception, to influence behavior, to affect the physical world, and to produce any number of other effects. To study magic is to study the quirks, foibles, and everyday hallucinations that characterize human perception, and to use those gaps between reality-as-it-is and reality-as-it-is-perceived as a vehicle for making supernatural-seeming interventions into perceived reality.15

As a form of media, magic operates in a perceptual landscape of associations and forces that have little to do with reason or logical perception. Lionel Snell (a.k.a. Ramsey Dukes), an early progenitor of “chaos” and “postmodern” magic, observes that

our brains have evolved a non-logical data processing facility which is, in its own way, every bit as useful and sophisticated as reason but which we tend to play down or analyze away because its casual connections seem so tenuous. This facility, which I called “feeling,” acts much faster than reason and seems to process vast amounts of data in parallel rather than sequentially like a logical thought.16

Snell explains that what we call “feeling” or “intuition” is the result of our having unconsciously internalized and classified huge amounts of perpetual “patterns” with varying levels of abstraction and complexity. For example, we may have preconsciously learned that walking alone at night and seeing a group of loud drunken men in the distance “goes with” danger, that green meat “goes with” feelings of sickness, or that shuffling a deck of cards “goes with” randomness.

In theoretical literature on magic, there are numerous schools of thought about what magic “is,” and each understands the gap between perception and reality in different ways. For our purposes, we will make a vastly oversimplified distinction between “stage magic” and “magick.” The theory underlying stage magic holds that reality is relatively stable, but our perceptions of it are glitchy. By capitalizing on the eccentricities of preconscious perception, we can create illusions, feats of wonder, or supernatural-seeming outcomes. In stage magic, supernatural-seeming feats are all “false.” The art of magic is therefore the art of deception, of creating phenomena that are not real, but which appear to be so. As James “The Amazing” Randi put it: “Magicians are the most honest people in the world: They tell you they’re going to fool you, and then they do it.”

In contrast, theories of “magick” are not so confident about distinctions between true and false or illusion and reality. There is a much bolder claim: perception and reality cannot be disentangled, and so they actually are, for practical purposes, one and the same.17 Because we cannot know “reality” beyond our perceptions, we can make no functional distinction between the two. In practice, the craft of magick suggests that by altering our perceptions, we can effectively alter reality itself.18

The CIA’s staff magician was neither a spiritualist nor a postmodernist. John Mulholland (born John Wickizer) was a master illusionist, public intellectual, and stage magician. Born in Chicago in 1896, Mulholland’s fascination with magic began at the age of five when his mother took him to see a performance by the legendary Harry Kellar. A few years later, they relocated to New York City, where Mulholland quickly immersed himself in the magic community. He joined the Society of American Magicians and convinced Kellar and John William Sargent to take him under their wing, becoming a professional stage magician while still a teenager. Over the next few decades, Mulholland ascended to become one of the premier performers of his day, and authored more than a dozen books on magic, illusionism, its history, and its relevance to communication and psychology. From 1930, he served as the editor for The Sphinx, a trade journal for magicians, alongside his wife Pauline Pierce and their polyamorous partner Dorothy Wolf, his longtime assistant.

For Mulholland, magic had little to do with the supernatural. He was highly skeptical of claims about the paranormal. Far from involving some kind of otherworldly conjuring, for Mulholland, magic

is the pretended performance of those things which cannot be done. The success of a magician’s simulation of doing the impossible depends upon misleading the minds of his audiences … A performance of magic is largely a demonstration of the universal reliability of certain facts of psychology.19

One of Mulholland’s professional hobbies was using his knowledge of trickery and deception to question the claims of psychics, mediums, and charlatans purporting to have access to the supernatural. His 1938 book Beware Familiar Spirits set out to refute the extravagant claims of spiritualists and mediums. In 1952, he wrote an article for Popular Mechanics debunking the UFO phenomenon.

In early 1953, Mulholland disappeared from public life. He closed up shop at The Sphinx and canceled most of his professional commitments. On the record, Mulholland had concerns about his health. In reality, the magician had accepted a position in the CIA’s newly formed MKULTRA program.20 (The security clearance process had gone slowly due to the agency’s nervousness about Mulholland’s “sexual proclivities.”) As he transitioned from public figure to clandestine operative, his income from performing and publishing was replaced by a stream of checks from an obscure organization with a mailbox at Southern Station, Washington D.C. named “ChemrophyI Associates.”21

Like other stage magicians, Mulholland’s oeuvre was built upon the premise that our minds make sense of the world around us through a constant process of preconscious pattern matching. When our minds encounter a familiar pattern such as a person tying their shoelaces or the appearance of a coin in our hand, our minds tend to preconsciously “throw away” those observations for having no particular relevance. The art of magic involves, in part, mimicking patterns that produce those “throwaway” observations or perceptual blind spots, and using them as a wrapper for an unexpected payload—a rabbit coming out of a hat, for instance. When the payload is revealed, it appears to have a supernatural origin because our minds have preconsciously “thrown away” the wrapper that contained it.

A payload might be delivered using a pattern rendered imperceptible by materials that nonspecialists lack a strong memetic relationship to. Most people rarely think about invisible thread, for instance, so a magician can capitalize on an audience’s lack of experience to produce the illusion of something we have far stronger memetic relationship to: levitation, for example. For the nonspecialist, the memetic content of watching an object levitate is far more salient than what a fellow magician might perceive, namely, a magician using invisible thread to create the appearance of a levitating object.

There are, of course, numerous other ways to deliver a “magical” payload (misdirection, concealment, forcing, etc.), but in Mulholland’s paradigm, they all exploit the simple fact that our minds either throw away, selectively interpret, or even act upon the vast majority of our sensory stimulus based on our preconscious and/or memetic priors. In other words, for Mulholland the art of magic has little to do with the supernatural. Instead, magic is the art of the cognitive injection attack, or mind hacking.

Mulholland had several projects for the CIA. Subproject 4 was an assignment to write a top-secret manual entitled “The Art of Deception,” instructing CIA field officers on using the fundamentals of magic to conduct more effective covert operations. Mulholland’s manual, eventually published in 2009 as The Official CIA Manual of Trickery and Deception, contained recipes for covert communications, the surreptitious delivery of toxins, hiding sensitive data and people, altering one’s appearance and mannerisms, and capitalizing on the different social expectations of men and women.

The techniques he devised often relied on concealing something remarkable inside something ordinary. He devised a stealthy communication technique that involved tying shoelaces in various ways to communicate messages, useful in communicating something by simply walking past someone on the street. He designed a version of the “disappearing box” (which makes the person who enters it “disappear”) into the trunk of a car, useful in the exfiltration of CIA agents from hostile situations. Another of his inventions was a silver dollar coin modified to contain a hidden needle to deliver deadly poison.

Like computer viruses masquerading as run-of-the-mill software updates, Mulholland’s inventions transformed the world of everyday objects and gestures into an invisible means of manipulation and covert action. The ordinariness of his inventions was precisely what made them effective.

Magick

An electronic signal sent directly into the brain of a hapless dog. The words of an early chatbot conjuring a spectral, techo-supernatural intelligence. An innocent-looking coin containing a powerful poison spike.

Bledsoe, Weizenbaum, and Mulholland were developing and refining an odd assortment of media, united by their ability to bypass reason and the sensible, to speak directly to the mind’s nether regions, and to elicit precognitive responses. Media designed to fly below the radar of rationality to shape perceptions, beliefs, and consciousness in ways that dissolve boundaries between perception and reality, the material and the immaterial, and the natural and the supernatural.

Woody Bledsoe, Joseph Weizenbaum, John Mulholland, and various branches of the CIA developed and deployed media designed to inject alternate realities into their subjects’ minds. Yet they all understood themselves to be in the business of artifice, of creating things that were not “real.” Weizenbaum joked that he was a “con man,” while Mulholland always maintained that magic involved “misleading the minds of his audience.” They were creating things that did not exist in order to cover up things that did exist, or to manipulate their targets into believing, and therefore acting, in ways they wanted to take advantage of.

Nonetheless, in their larger worldviews, these were mere magic tricks. Rabbits coming out of hats were actually coming out of specially designed tables. Tricks are meant to deceive and distort, to be sure, but they can have no bearing on reality itself, whose metaphysical foundations remained immune from such illusionistic knob-twisting.

But what if they were wrong?

What if they believed they were practicing stage magic, but were in fact playing with something far more occult? What if they were inadvertently playing with magick?

And what would happen if their sleights of hand, electronic signals, and sigils began conjuring different types of rabbits? Magickal beings with their own ideas about the malleability of perception and reality?

Mulholland’s experience debunking the supernatural made him useful to the agency. The CIA had become fascinated by the possibilities of hypnosis, ESP, telepathy, and other parapsychological phenomena. Mulholland became their internal reality check. By 1955, Mulholland was traveling around the country to meet and assess psychic test subjects engaged in an early version of “remote viewing,” a man who claimed that a copper-lined Faraday cage gave him enormous psychic abilities, and other X-Files-inflected occurrences.

In 1956, the CIA gave Mulholland another task: investigating UFOs.

UFOs had taken to the skies. And the CIA knew all about them. Because the CIA created them.

To be continued in “Society of the Psyop, Part 3”