|
iPlant
|
Fiction These are the first few chapters of a novel I started writing while working on iPlant.eu. Actually the second part of the first chapter was written over a year before I got started with the site but the rest is about the development of iPlants. One day I hope to have enough time and dopamine to write the whole thing. In the back-story of the novel conditional rewarding brain stimulation is first introduced as a last-resort procedure to help morbidly obese patients exercise. The move is motivated by a worsening epidemic of cardiovascular and neurodegenerative disease, stemming from obesity and an aging population. Following success in a number of hospitals, the procedure is rapidly applied to a variety of patient groups to motivate a range of behaviours. As the efficacy and safety of the implants improve a growing number of clinically healthy individuals choose to obtain implants. Private clinics specializing in the procedure become increasingly lucrative. Ike, Meg, Lucy and the narrator Chris are reserchers in such a clinic. Their work is necessarily self-experimental. The novel begins with the public launch of the first "iPlant" - an advanced deep brain stimulation implant for inducing conditional rewarding brain stimulation, aimed at a general market. Unlike previous conditional rewarding brain stimulation implants the iPlant targets dopamine and serotonin nuclei directly. -- Christopher Harris Chapter 1 - Program yourself Chapter 2 - iPlant-driven research Chapter 3 - New job Chapter 4 - Make your needs known 
Chapter 1,
Chapter 2, Chapter 3
Additional material: Breasts, JealousyChapter 1 - Program yourself (written 2005-2007) This chapter was reprinted on the Institute of Ethics and Emerging Technologies website. Special thanks to Laura Kilarski for inspiration, discussion and proof-reading. - "Most people don't realize how common brain implants have become in the last couple of years. Every month thousands of patients all over the world have electronics surgically implanted into their heads to treat problems with hearing, movement and pain, and more recently with epilepsy, vision, paralysis, depression, compulsive behaviour and loss of consciousness (Schwalb and Hamani, 2008; Perlmutter and Mink, 2006; Lebedev and Licolelis, 2006; Kringelbach et al., 2007). The iPlant is just another implant, aimed at new regions in the brain." Ike was addressing the press conference. Dark shirt, two deep blue wires running down the back of his head towards a console strapped to his chest. Two days ago he snapped Lucy's right index finger in three places when she tried to disconnect his dopaminergic nuclei. She's having problems not blaming him. - "It's eight implants really; each controlling a group of cells that supply the brain with dopamine or serotonin (the VTA, SNc, dorsal- and medial raphe nuclei, bilaterally). Electrical activity in these cells is at the core of our mental lives. Dopamine is the goal of behaviour and in large quantities it drives activity and growth in brain tissue. When a thought grabs you so you can't think of anything else it's because the neurons that make up that thought are being fed more than enough dopamine to keep on firing and connect to other neurons. When you're comparing two options, unable to decide, it's because they drum up the same amount of dopamine. Dopamine is motivation, its fluctuations describe where you direct your attention, what's important to you and what you remember (Cools & Robbins, 2004; Schultz, 2007). This is a gross oversimplification but it works - our implants control the flow of dopamine in the brain." What began as a problem of mapping and reducing noise in the high power spectra of Ike's dopaminergic implants quickly became a problem of security, of preventing over-use. We knew we would have to 'tune' the implants, find the combination of electrodes and currents with the steadiest and most selective effect, with nothing but Ike's own words to guide us, but we didn't expect him to lie, and his eyes didn't reveal a thing. Then Lucy noticed his white knuckles, moved to disconnect and it took me and two of the software engineers to hold him down. - "The point is, there's no groundbreaking science here. This could have been done many years ago, probably was done in military labs somewhere, in Russia, the US or China, anywhere really; the regions in the brain where motivation and learning can be generated electrically were mapped in animals and humans more than half a century ago (Olds and Milner, 1954; Heath, 1963) and people immediately started thinking about realistic forms of mind control (Delgado, 1971). Only a few years ago the US army developed implants that allow them to remote control rats via normal laptops, 'for search-and-rescue missions' (Talwar et al., 2002)." Ike could not have full access to the implants, so much was clear. Maybe it's a general rule that no one can control their own dopamine but in his case the problem had been exasperated - there was no way of knowing for sure how far he'd go to repeat that overflow. The security we put in place meant the computers had to be unlocked and operated from three points simultaneously: two in opposite corners of the lab, one in the library on the other side of campus. Whosever implants were being programmed could still direct the work, but all uploading of compiled programs had to be cleared and confirmed by the others. - "The first programs we developed for the iPlant were exercise programs, originally for treating obesity. Take running: the soles of these trainers here contain sensors that punch out a bolt of dopamine with each step. Nothing like orgasm, but enough to make you want it like those few lucky people who really enjoy running want it, with a strict time limit of course. Implant-driven exercise programs have been available for rats for decades (Burgess et al., 1991; Gardner et al., 1991), now we've got them too. Learning programs work on a similar principle, dopamine for correct answers. The current version of the iPlant has about six thousand hours of training in French, German, Mandarin, Japanese and maths, but you can expect a lot more with each update. There are a few purely clinical programs as well, like DeTox and DePhobe. Dopamine-driven software lets you want what you want to want." The rest of that day would have been a lot easier if he'd been genuinely affected by Lucy's finger, but the way he saw it him and the electrode arrays had malfunctioned, like an allergic reaction, not his fault. And who knows, who knows what it feels like to have pure dopamine filling up your frontal lobes with nothing to hold it back but... what? Only Ike. Rats given full access to their own dopamine self-stimulate until they collapse from starvation or lack of sleep. - "All eight implants also regulate baseline or so called 'tonic' electrical activity in their target cells. For dopamine we call this program Focus. People perform best when their dopamine concentrations are at an optimum level for the task at hand and most of us experience daily problems maintaining sufficient dopamine levels; maintaining concentration. Focus works similarly to the stimulants children are prescribed for ADHD, but it's cleaner and more flexible and can be turned off, which means fewer side effects. The analogous program for serotonin we call AntiDep because it works like SSRI antidepressants. If you're depressed or chronically anxious, chances are you were born with an underdeveloped serotonin system or its growth was stunted by stress (Jans et al., 2007). Like antidepressants, AntiDep prevents some of the more vicious long-term effects of not having enough serotonin, like overproduction of stress hormones, inhibited growth in the hippocampus, and, at least for some people, social isolation and despair (Dranovsky and Hen, 2006)." I met Ike at UCL in 2008. In med school he'd done enough ecstasy to mentally cripple a rhesus monkey, but the damage MDMA does to humans is still not known. Ike wanted me to find out: to inject him with radioactive serotonin transporter ligands and PET scan his brain, then repeat the test yearly to check for regeneration. It didn't look good. The serotonergic branches perfusing his forebrain were down to 60% of normal thickness, and a mesh of regrowth had formed a fine, essentially useless cloud of axons around the serotonergic cell bodies back in the brainstem. This was ecstasy damage, same as in monkeys (Hatzidimitriou et al., 1999), and if the ligand had been approved for use in humans we might have been able to publish the scans. How much of the damage had been there before the pills and the powders - as the serotonergic vulnerability that surfaces as anxiety and compulsions - was not something we could know, and it didn't matter to Ike. I decided to introduce him to the team and to our research. - "All changes to your iPlant have to be uploaded and tuned here at our facilities using special surgical equipment: this means that no government satellites can turn you into mindless drones; no one can hack your brain and make you desperate to give them your money. No, the main challenge to the iPlant are those segments of society that will react to it with fear and try to demonize it. These people are often the ones who would benefit the least from an iPlant: they don't suffer the never-ending frustrations of a short attention span, excessive fatigue, addiction, depression or intense social anxiety; they are the ones whose dopaminergic and serotonergic cell groups are working fine, and they'll say that you using an implant to reach your optimum is somehow wrong, as if you were cheating in some petty game. No, this is a 21st century class struggle - those born rich in dopamine and serotonin protecting their 'hard earned' privileges."leges." 
• • • We pumped all the money generated by the iPlant into the development of new electronics and signal processing software for iPlant 2, a wireless comb of high-density electrode arrays for the corpus callosum and frontal lobes. Like the original BrainGate it was to enable direct neural control of a computer (Hochberg et al., 2006), but at a much much higher bandwidth than the mouse-and-keyboard interface. After five years we were at a stage where Meg could volunteer. Some of us had thought she'd break the moment she connected. That her flow would rupture, her synchrony leaking out like in that book where a brain locked up in a lab aquarium gets hold of the wire of a security camera and almost manages to 'escape' (Jersild, 1988). The idea that the mind is ethereal is old, extremely old, but I don't think she doubted the hold of the body for a second, not really, I don't think she could. The look on her face, it was the same smile Ike had had when he came back from his first three hours of running. She described the sensation as 'inward touch', as something like a blind person seeing shapes by touching and stroking. The link was largely one-way, brain to computer, the team relying on screens to structure her output, but from the start she had wanted 'primers' - faint feedback to the electrodes involved in any successfully executed command. It was no more than two hours before she turned the screens off and adjusted the music in the lab using nothing but the direct link, flipping between tracks with a wide smile, holding back the laughter. She could write, of course, but we never understood how she learned to read, not with the screens off. It seemed impossible. Her control of the computers was at the coding-level of mouse and keyboard and there was no way of re-routing the screen-output to the implant. We thought maybe she'd found a way to locate text and somehow 'check' each letter or object against output of her own, using some difference in the primer-feedback for correct and incorrect trials to determine a letter sequence. But the software for that kind of checking, however simple, wasn't installed. She said it was a matter of 'jumping', said it felt like skipping stone. She'd stopped smiling. She wasn't there in the morning. Her body I mean, she wasn't there at all. We'd all stayed at the compound overnight, in case of emergency, sleeping right there along the walls. She sat by the desk with the software-people for hours asking questions, more about computer science it seemed then about the implant or the decoding software. Then she went to sit on the operating table, drawing beautiful flowers in Photoshop, with edges that looked smooth and soft and razor-sharp at once, and had layers of coloured shadows hanging under them. Why didn't I make her tell me more about it? She explained 'jumping' in some detail, like it was a matter of getting through a series of commands so quickly the computer had to catch up, which gave you a half-second glimpse of your own serial trace at deeper levels of coding. Then you took that output and repeated it with minor changes and eventually you learned how to get behind the surface of programs. She said the internet was beautiful. She wrote her parents a 32 page letter, all hand, which they won't let us read. Her dad quit his job and moved to France within a week. We still haven't heard from her. I think it'll take us between three and five months to assemble another implant. Ike and I flipped a coin and I'm next. I've agreed to keep the doors locked this time. Chapter 2 - iPlant-driven research (written in 2008) Science is about money. Scientific facts are not so much discovered by intellectual curiosity as they are constructed by cash investments. Science is objective but also directed. Research equipment has a cost, as do scientists. Our initial large source of funding, the one that carried the first one hundred iPlants through clinical trials demonstrating their capacity to treat morbid obesity, was unexpected. We arranged to meet Aubrey de Grey in a pub in Cambridge in 2009. He surprised me by bringing chips to the table - I'd thought he would at the very least avoid saturated fats. - "And you are confident patients undergoing this new form of deep brain stimulation therapy would willingly volunteer some of their free time to become conditioned to conduct biomedical research?" Aubrey asked. - "Maybe initially they would be people from the obesity trials." Lucy said, "Initially. But we're convinced a lot of people with no clinical problems will want iPlants once we get them past clinical trials. And if we can show in concrete terms, in research hours and actual findings, just how much a single iPlant can contribute to medical science... It would be exactly what we need to generate public support and get a foot in the door on government funding. We considered a climate change oriented approach - iPlant-driven research into renewable energy - but that kind of science is unstructured and abstract; it's engineering, not lab work, takes years of training. SENS is different; you've worked out in detail what needs to be done, even for cancer (de Grey, 2005; de Grey et al., 2004). What you need is an army of motivated scientists to do it. This is that army." - "Bunch of fat people in lab coats addicted to exercise and arbitrary research protocols" Ike said and sipped his pint. - "Nooo..." Lucy said, "Let's say four hours of iPlant-driven research per week per participant, one hundred participants in the stage II trials and one year of actual trials. That's over twenty thousand research hours; something like £250.000 in saved research funding at a typical UK pay rate. And that's from the clinical trials alone. You've estimated the full cost of SENS at $100 million per year. That's roughly twenty thousand volunteers working four hours a week. That's a lot of people, but say we advertise it as a Manhattan Project to cure cancer - you really think finding volunteers would be a serious problem? If it's safe and the initial clinical trials show how effective it is? If it requires no effort whatsoever?" - "Mind control on a mass scale, ain't nothing like it" Ike said. - "Ike! Fuck! What's wrong with you?" Meg punched Ike's shoulder. - "Just sparing Dr. de Grey the trouble of listing reasons why he wouldn't want his reputable foundation associated with an existential bomb-shell like this one." - "I'm not sure I accept your calculations." Aubrey said. "But apart from that, how would you organize tens of thousands of unskilled researchers? How exactly would you train them? What about laboratories and equipment and logistics?" - "We're not sure" Lucy said. "We have a detailed plan for how to organize up to two hundred researchers across four universities. Basic molecular biology mostly, building on old protocols. It's tempting to go further and try and to model the Human Genome Project but we'd like to avoid having the Chinese play the role of Celera in all this. You know what their demographics look like. We also feel there's an event horizon shortly after a few hundred." - "Things won't exactly stay the same after Hu gets his hands on this implant." Ike said. Aubrey coughed. He was looking out the window. "How far away from clinical trials are you?" - "That depends on how much you want to invest."  • • • Friday. Me, Ike and Meg are in the new proteomics lab in Reading, tuning the reinforcement system. At least that's what we keep telling ourselves. Why we're really doing it... well that's the big question. Ike is still in the protein factory. Meg finished purification a few hours ago and said she'd give microarray imprinting another try even though it looks too complex for a protocol. I'm doing quality control. Proteomics is the ever-expanding region between cell biology and genetics. The science of proteins. The Reading facility is a high-throughput unit developing antibodies towards short, specific segments of protein (compare Nilsson et al., 2005; Uhlen et al., 2005). The segments are isolated in the protein factory and sent to a farm in Cardiff where they're used to immunize rabbits. The serum is returned to us and the antibodies are extracted by washing the diluted yellowish fluid through columns containing the target protein segments. The end product is a transparent liquid of pure, polyclonal antibody. I'm not sure how many thousand pounds they get for each millilitre. Quality control means taking a sample of the finished product, pipetting it onto glass microarrays containing twenty different sets of protein fragments, incubating and washing the slides, running them through a scanner and going through the resulting images to make sure the antibody sticks to the right fragment and the right fragment only. Maybe it sounds difficult but it's not. It's about two hundred separate behaviours, none of which require knowledge of what a protein or an antibody is. It's surprising how much of science consists of these monotonous construction lines; so called 'protocols', and how much money is spent getting people to do them. The reinforcement system I've developed this week adds rewarding dopamine pulses at thirty-two points along the protocol. Another round. One more. Then I'm probably done. It's not gonna get much better than this. Picking up the vials from the freezers and carrying them over to the bench. You can tell which ones are Ike's from the labelling. OK. Four at a time. I plug in and carefully read the labels, entering the letters slowly on the computer as I go along. One. Yea, just right. Makes you wonder how you just did that. Makes you think about the motion of your arm, the fingers on the keyboard, the muscles in your fingers. Mindfulness meditation comes to mind but of course it's nothing like it. Two. What is it this experience, what is it? It's a pulse of dopamine of course, mostly prefrontal cortex, and a weak tail of serotonin, which, if the scans are any good, seems to be better... work better... if it's focused along the ventral branches. But that's not it, there's something to this. Three. Summers from when I was twelve years old or so visualize quite a lot these days. I was happy in a very different way back then, unrestrained. Plastic? Remembering learning to do somersaults in a field or a garden. Four. Ouch. Ooouch. Not right. Way too hard. Ike's right: you can't do more than three in a row if they're identical, even though you want to. I tell the computer to randomize the dopamine outwards on the last pulse next time; to push it more towards the edges. Next time... hah... This segment might be a bit too good, despite that last one. I make a note of it, put the music back on and start diluting PBS for the first microarray. Chapter 3 - New job (written in 2010) - "Every time, every time we meet she's late." Lucy and Ike stood waiting outside D's assistant's office. The glass door was shut and the room inside in semi-darkness. The assistant was almost twenty minutes late. - "Dopamine deficiency, swear to God" Lucy said. - "Which one?" Ike asked. - "Midbrain insufficiency. Plain cell numbers, not enough dopamine." - "Not a receptor problem? D1, D2? Unresponsive adrenal glands?" - "Adrenal problems don't cause chronic lateness, she relies on her adrenal stress response to get her ass off the couch in the morning, but that doesn't kick in until she's critically late and then there's terrible traffic or some shit and we end up loosing our fucking morning staring at her door.." - "Never heard you curse before" Ike said. - "Period Lucy said, holding her elbow and stepping restlessly on the spot. - "D1 then? D2? Receptors not growing the way they should?" - "Midbrain insufficiency" - "How do you know? How would you know without scanning her? You got her scans??" - "No. I just know." - "Bullshit" - "Fuck you" They glared at each other for a brief moment. - "It's not D1" Lucy said "because she's not inattentive and certainly not impulsive. And it's not D2 because she's a vicious learner, not just memoranda but procedure as well, that's why D hired her." - "When was that?" - "Same as me, for a while I thought we'd work together - her background is biotech and biomedical patents - but she's all administration now and real close to D." - "Balanced D1 and D2 deficiencies then? Hyperactive dopamine transporter?" - "Knew you'd say that" - "Well?" - "You're pretty obvious Ike, as a person" - "You're stalling" - "Midbrain insufficiency. Not enough dopamine neurons." - "How could you possibly know that without looking at her scans?!" - "Calm down" Lucy said and lowered her voice as a small group of people emerged from the elevator at the end of the corridor and disappeared around a corner. "Look, first of all Meg and I agree on it and that's rare and I trust her judgement when it comes to guessing phenotypes and so should you. Second, Marlena is unstable, not in a way that really impacts her work but she's selectively anhedonic, severely - sometimes she truly doesn't see the point or doesn't care unless she's told and I bet you she's a lot less polished at home. But it's a partial problem and if you don't look for it you might not notice: and that's the point, receptor deficiencies have a smooth psychological profile, transporter deficiencies doubly so. Marlena's problem comes in patches. Third, look at her forehead! I'm not saying she's got a small midbrain, I'm saying her forebrain is oversized and sometimes she doesn't have the dopamine to keep all her units running, especially in the morning, which is why we're standing here wasting time." Ike watched her speak. - "Don't you ever tell anyone what I just said" Lucy said and suddenly looked nervous. - "Course not" - "I shouldn't have said all that" - "I'm not telling, why would I tell?" - "Shouldn't have said that" - "Look, we're not gonna work well together if you keep distrusting me" - "'Keep'?" Ike paused, unsure. - "Meg" he said finally "You're trying to protect her from me." Lucy burst into a high laugh. - "Don't wanna sound cliché or theatrical but I'm more concerned about what she'll do to you, if you step on her feet. Do whatever you want, just don't fuck up so you can't work together." - "How about some pointers then? What do you mean 'what she'll do' to me? She 'unstable' too?" - "Oh no, I've gossiped more than enough." They were silent for a while.  - "How about you tell me something?" Lucy said after a while. "How about you tell me where you got that scar?" She stroked the left side of her jaw, indicating a thick scar on his. . - "Stepped on someone's feet" Ike mumbled. Marlena suddenly appeared from the elevator at the end of the hallway and hurried towards them. - "I'm so sorry I'm late" she said. "Traffic's terrible and it's raining. Please come along." Lucy and Ike looked with surprise at each other as they followed Marlena down the corridor towards D's office, which she opened with a metal key. - "We need Ike fully installed by the end of the day" Marlena said. She sat down behind D's large black desk, roused his computer with a quick mouse shake, typed out a long string of characters and hit enter unnecessarily hard. The computer logged in, showing the company logo against a dark background. She gestured for them to sit down in two easy chairs opposite the desk. "Water?" She fished up a bottle of sparkling mineral water from a drawer. - "No thanks" they both said, attentive. Marlena paused and looked at them. - "You've been hired" she finally said to Ike. "I thought you knew." Ike beamed. - "D wants you installed and ready to work by the end of the day, you're going with the others to Brussels tomorrow." - "That's excellent!" Ike exclaimed. - "You need to bring your scans, we need you to discuss them with some people from the council." There was a pause. - "The brain scans" Marlena continued, looking steadily at him. "You and Chris used his new ligands to make scans of your brain. You imaged your serotonin system a week ago in the PET scanner. You found substantial changes." - "Chris told you?" Ike said. "He told D? He said I'd get him fired if I told anyone. So did you" he turned to Lucy. - "I don't know anything about this" Lucy said and held up her hands "don't particularly want to know." - "You're going to Brussels to determine the council's true limits on experimental and commercial deep brain stimulation" Marlena said. "You want to work here because, one day, you want to undergo such surgery yourself - sooner rather than later I understand." - "I want an iPlant" Ike said. - "The iPlant is a theoretical construct as far as human application is concerned" Marlena said. "We need a battery of permissions and suspicions cleared before we can proceed with surgery. You're going to Brussels to test the European medical law authorities on that particular point. You'll pursue the argument that our implants and surgical procedure are worth the risks of surgery to customers who have never been hospitalized but who might nevertheless consider themselves neurologically handicapped. Your scans will provide a vivid example of such a case." - "Do we really want to use one unauthorized procedure to get permission for another?" Lucy asked. - "You're not going there to get permisison, just to test the waters.." Marlena began. - "I know" Lucy said "And to be frank we'll probably proceed with the iPlant either way. I'm asking whether these scans might do more harm than good." - "D tells me they are quite convincing" Marlena said and looked at Ike. The three fell silent for a moment. - "Jeez Ike" Lucy finally said "how much ecstasy did you have?" - "It makes sense" Ike said. "I'm in." - "Brilliant" Marlena said and began filling out a form she'd pulled up on the screen. - "How much did you have?" Lucy asked again. - "Enough" Ike said. - "If you'll come with me over here.." Marlena said and led Ike over to an eye and fingerscanner beside a small safe in the far corner of the room, next to a large liquor cabinet. "Just place your fingertips here please.. and look at the white dot there.. and again.. excellent." Chapter 4 - Make your needs known (written in 2010) - "I am still your body, you're just a brain!! You have no right..!" - "A brain with electronic motivation", Meg whispered. - "Fuck you!" her body shouted, in its own way, and ramped up lactic acid synthesis. Meg pressed on. It hurt, but the pain was part of her running, and she wanted to run. She'd been running non-stop for almost two hours on the big treadmill in the lab. The brain network maintaining her posture, driving her legs, arms and lungs, was lean, mean and optimized. Pressure-sensors in her shoes triggered dopaminergic electrodes in her brain with every step; honing, shaping, supporting and reinforcing the neural network and its muscle contractions. Interruption was impossible. Tiredness irrelevant. Even pain was part of the purpose, part of her strength, her will, her artificial motivation. Lucy stood at the door, watching her. - "It's time" she called. - "Noo.." Meg groaned. - "It's time", Lucy said again. - "Fuck!" Meg shouted and punched the stop button on the treadmill. The treadmill started slowing down, and for a moment Meg found herself trying to keep it going, pressing her hands against the railing and pushing her feet against the rubber sheet, harder and harder, the neural network in her refusing to disassemble, even though the link was lost, the sensors in her shoes inactive. - "Wow" she gasped, between strained gulps of air, and stopped. "Wow." She released the railing, jumped off the edge of the treadmill and stomped both feet hard against the floor, putting all her weight and strength into it. But the sensors were silent. She groaned again and collapsed on the floor. Lucy watched her from the door. - "You ok?" she asked? Meg rolled over on her back, spread-eagle, and lay panting, staring at the ceiling. "Wow."  • • • - “It’s a bit like a movie” Ike said one morning “living with these reinforcement schedules. A good movie, but with advertising... only the adverts are the minutes spent jacked out and you just want back to the real thing. Like television before on-demand.” - “You know what that means” I said. - “Yeah” Ike said and stretched his back. “Time to stop. I could use a break. I just wish we knew how to deal with this when I get back, I’ll need something to work with.” We leaned against the steel railing of the balcony and observed the narrow alley below. - “You should take Meg to the conference” I said, spontaneously, after a brief silence. - “She doesn’t want to go to the conference with me, Christopher, you know that.” - “Yea there’s a reason for that.” Ike grumbled. - “Her software is almost ready” I said. Ike looked at me. - “It will work” I said, and turned to watch a stream of water creeping down the dusty brick wall of the building in front of us. Steam was rising from the alley below. The heat was thick. Like Singapore. - “What are you afraid of?” I asked. - “Her amygdala and her computer” Ike murmured. “What? What is it?? First Lucy, now you. These vague warnings and innuendos. The fuck is up with you people?” - “Ever read Burning Chrome?” I asked. - “Burning who??” - “Forget it” I said “Look, Ike, don’t worry about it.” - “She gonna leave?” Ike asked, sounding distinctly worried “Now that her software's working, she off?” - “Meg doesn’t leave. Don’t worry about it. We just want you calm.” - “I am calm.” Ike said. We paused to let an old couple walk by on the narrow street. “But you guys creep me out” he continued. “The way you’re all real proper idiosyncratic people before and after, but every work-day you synchronize into a single fucking mind. It’s creepy.” - “We don’t become a mind…” I began. - “You know what I mean” Ike said. - “We work on a single reinforcement schedule” I said. “That’s all. We want you on it, but you gotta get this conference over with. And Meg...” - “What?” - “Meg wants your brain.” There was a brief silence, then we both cracked up laughing. - “Look” I said “Honestly. Meg’s running half the show here. And she needs your brain, the scans and… for test-runs, you know, all the time like.” - “Speak English” Ike snapped. - “She’s not an option for you. Not now. Risk-reward analysis, you know the drill.” - “And you know how much dopamine I need to maintain that attitude. My supplies don’t change just because...” - “See that?” I said “Right there. Right there is why Meg won’t go to the conference with you.” - “What?” - “She’s not a complete workaholic you know” I said. - “I know that.” - “She’s human.” - “Right.” - “She’s female.” Ike was silent. - “She’s single” I continued. - “Thought you said she wasn’t an option” Ike said. - “Not now, no, certainly not, so don’t worry about it.” - “I am calm” Ike said again. “Tell me about her new software again.”  We sat down opposite each other, leaning against the brown and orange walls of the small balcony. It was early morning. We’d spent the previous hour trying to beat our brains into shape with coffee and dB, but our body-clocks were 9 hours behind. We had arrived in Seoul the day before. Our hotel was small and cheap, a hostel really, but central. The heat was intense and the sky cloudy. Last night the area had been all neon, buzzing. Rain fell briefly during the early morning and was now evaporating as buildings woke and people prepared for the new day. - “The human brain contains over 100 billion neurons” I began. “Neurons are either sensory neurons, interneurons or motor neurons. We’re concerned with motor neurons.” - “And interneurons one or two synapses upstream.” Ike chimed in. - “Right. Now, a motor neuron is a neuron that controls one or more muscle cells. Bicep, tricep, calf, diaphragm, etc. - these are muscles. Muscles are aggregates of many million muscle cells, each controlled by a specific motor neuron. There are 640 muscles in the human body, each controlled by a specific group of motor neurons that cluster together anatomically in the motor cortex, brain stem and spinal cord. The neurons group together and also fire together, since all cells in a muscle have to contract more or less simultaneously.” - “This is key” Ike said. - “Yes” I continued “because it means that the brain’s intention or signal to contract a specific muscle can be detected as a unique electrical potential in various scalp recordings on and around the motor cortex (e.g. Phon-Amnuaisuk, 2008; see the BCI competition website for more references). Electrophysiologists call these chunks of brain activity ‘motor programs’. Each motor program is a burst of brain activity that drives drives the contraction of a specific muscle. Walking, jogging, typing, speaking, chewing, swimming, crawling, making coffee...” - “...having sex” Ike said. - “...having sex, playing tennis, tying shoes, doing sit-ups; all human behaviors that involve repetitive muscle contractions can in principle be identified in a scalp recordings as recurring combinations of specific motor programs (see chewing in figure A). Like lifting weights is alternating activation of biceps and triceps: biceps, triceps, biceps, triceps; each contraction driven by activation of a specific group of motor neurons. So if you plot the activity of motor neurons driving a repetitive behaviour against eachother you get a cycle as the brain goes back and forth between the different muscle contractions it needs: a cyclic attractor, often quite beautiful and smooth since every motor program differs slightly on each revolution around the cycle, even in highly repetitive behaviours like chewing (Horn et al., 2004). - "Hurrah for free will!" (Brembs, 2010) Ike exclaimed and poked his grumbling stomach. - “Let’s get going” I said and stood up. “Bottom line is: almost every human behaviour can be detected in scalp recordings because they all involve unique patterns of brain activity. Meg’s software does this, checks whether the behaviour needs to be reinforced, and sends the resulting command straight to the patient’s iPlant. It’s a closed loop. She’s removing the need for human supervision. No need to have exercise equipment or training software monitoring your behaviour. No need to plug yourself in.” We walked inside. The hallway was empty. A screen buzzed nervously behind a half-open door further down the hall. We got our bags and went out into the streets, which were already crowded, in search of food. 
- “Right now” I called “Right now, what would recordings from our motor neurons look like? What kind of cycles are we in?” Ike looked glum. “Walking?” he ventured. “It doesn‟t matter. She's worked it out, the sequences of motor programs or whatever that identify behaviours we need to reinforce.” - “The rowing machine for sure” I said. “All the upper body stuff, not sure whether she can get to the leg muscles, the neurons are all tucked inside the folds…” - “Behavioural fingerprints” Ike said. “Sure. Now what does she do with them exactly?” - “Stay on it” I said “We need to know how she does it first.” - “What do you want? Brainstorming?” Ike stopped walking. “Why don’t we go and talk to her? Right now. Instead of guessing.” - “Yea?” I asked skeptically. Meg and Ike had not spoken since his 'accident' with Lucy, three weeks earlier. - “Yes” Ike replied. - “Let’s do it" I said. We got sandwiches and coffee, crossed a few streets and hiked up into Namsan park. It was quiet and very green and we kept walking for almost an hour, up roads and old stone stairs, not saying much. Eventually we sat down on a bench and called on our tablets. I decided to try the display in my new glasses. Lucy answered.  - “What‟s up?” she greeted. “You haven't got keys to the engineering workshop do you? They're closed over the weekend.” We both shook our heads. - “Lucy, where's Meg? We need to have a word with her.” - “She’s asleep.” The picture of Lucy on the display in my glasses seemed to merge with the green bushes and the people in the background in front of me. I sat very still, unfamiliar with the new display. - “That’s a shame. Lucy, can you wake her up for us? It’s rude but we’ve got a few questions about her new software and it's urgent. Please?“ - “Always Meg, you never ask for my advice...” Lucy pouted. - “No, we want your input too, of course, we...” I began. - “Relax” Lucy said and smiled. “Personally I’m beginning to find it hard to spend less than half my days somehow interacting with her. You guys must be starving.” - “There are quite a few questions, yes” I said. “Please?” - “Bore” Lucy disappeared from the display, picked up the tablet, and its camera took us on a shaky ride down the hallway. - “Meeg!” Lucy shouted. “Darling! The boys want a word with you.” Ike gave me a strange look, probably indicating this was one of the moments where we ‘creeped him out’. - “Why is she like that?” he mouthed. “She’s never like this.” I just held up my hands, urging calm. He seemed to get it. - “Maybe... let me do the talking...?” I said. Lucy was shaking a pile of duvets. It was dark in the room, except for a single, very big candle. Lucy leaned the tablet against the candle and shot us a quick smile. - “What the...” we heard Meg’s muffled voice from under the pillows. - “The boys dear, are so desperate for information, they’re making me wake you up.” - “Morons…” - “Actually, we want a conversation, Lucy” Ike said, loudly. I put my chin in the palm of my hand and gave him an exasperated look, but he just smiled. - “Ike...?” The pile was suddenly still. - “Meg we’d like to talk to you about the new algorithms.” I began. “And whatever you, the two of you want to talk about, of course. Brainstorming. Yea?” - “I get to ask anything I want?” Meg asked, now clear from under the feathers. - “Meg, tell us about the EEG-to-iPlant loop” I said. The line went silent. In the distance, a family was purchasing balloons for their infant son. - “You guys,” Meg said finally “You guys bore me.”  - “Generating, evaluating and maintaining good, adaptive behaviours by stitching together motor program sequences is an almost constant challenge for the brain of Homo sapiens. Yea?” Lucy was talking. - “Like other mammalian nervous systems, ours has adapted to this challenge by developing a highly sophisticated mesocorticolimbic neural network dedicated to dopamine-mediated reward-learning, which is supposed to ensure our evolutionary fitness. Yea?” - “Come on Lucy...” Ike said. - “Now,” Lucy continued “given a few large electrodes in the middle of this neural network, it’s possible to study the dynamics of a person's reward system in real-time (Cohen et al., 2008; Münte et al., 2008). Add hardware and software for identifying ongoing behaviours on the basis of scalp recordings and you’ve got a reward value associated with everything a person does. You can correlate that value with how well or how often the behaviour is performed. Optimal dopamine concentrations are already known for a range of behaviours in mice and rats (Cools & Robbins, 2004) but now we’re talking about the human brain, almost all its behaviours, and a completely personalized reward-performance analysis.” - “Yea we need a name for that” Meg called from her spot under the covers. Only Lucy was watching us on the display. The room was very dark, and the big candle seemed to shine a weird white light. - “Is that snow?” Ike asked absentmindedly. - “Huh?” I said. - “There’s snow on their window, the one in the ceiling. That’s why the light’s so spooky.” - “This is boring” Meg called. “My head hurts! Why’d you fucking wake me up!?” - “Meg” I said, slowly. “Tell us about the loop, please.” Lucy looked at us, then over at Meg. There was silence for a while. Then Meg shot up, grabbed the tablet and stared at us, hissing “Where is he? Where is this goddamn imbecile?!” We both recoiled in our seats, on the bench, almost 10.000 kilometers away from the threat. Like clockwork. - “Hi Meg” Ike greeted, sounding less than happy. - “You!” Meg shouted “When I get to you I’m gonna drive your head through a fucking wall!!” - “Meg, please” I said. “Dialogue. We want to talk. About your software.” - “I didn’t mean to...” Ike mumbled and hung his head. - “Oh no no no no nooo” Meg hissed and actually gritted her teeth at us. She looked quite terrifying in the candle light, against the blue sky of Seoul, and the green shrubs. “Chris, you are not letting him get away with this bullshit.” - “Meg!” I shouted, suddenly tired. “We’re not talking about this now. We will talk about this another time. We’re here to talk about the new software.” - “It’s OK Meg” Lucy’s voice said from a distance. Meg glared at us for several, long seconds. It was very quiet. Then, like liquid to crystal, her entire aura shifted and she beamed at us brightly. - “So how are you guys!?” she asked, smiling. Lucy came back into view with a laptop. She dialled in and appeared on our displays. There was a long silence. - “Well?” Meg chirped, still smiling, showing all her teeth. - “Well” Ike said. “We've been thinking.” - “Woow” Meg contemplated this. - “...about the EEG-to-iPlant loop” - “You did?” - “Cut the crap” Ike said. - “You know, Christopher” Meg said “I'm really really starting to wonder why you hired this guy.” - “Me??” I exclaimed, genuinely stunned. - “You. Not D. You two planned this.” - “You know that's not what happened” I said. - “Do I?” Meg looked at me. We fell silent. 
- “My first question” Ike said after a while. “Is this. You’re bound to get the scalp recordings wrong some of the time. Not completely wrong; I get it, you can discriminate behaviours, but you've also got some sort of ideal brain state template in there, for adjusting the reward. You record EEG for a while, you've got the patient pumping away in the rowing machine or whatever, you extract your motor programs, identify the behaviour, check what the reward-system is doing, check what it should be doing given the identified behaviour and correct the difference through the iPlant. Fine. But you're not letting the brain figure out for itself how to get its reward; you're reinforcing brain activity directly, based on this ideal brain state template. But if you get that template even slightly wrong you could drive up a twisted ankle or some weird muscle tension and you wouldn't even know it because the patient won't tell you cause he'll be having too much fun.” - “Will he now?” Meg said. - “You tell me” Ike finished. - “‘Too much fun’” Meg repeated. “Most of these people are still running for their lives. Not everyone can spend all their time ‘optimizing’ the current parameters of their iPlant, you know?” Ike was silent, defiant. Another silence followed. - “Look,” Meg said finally, making a gesture of resignation with her hand “you underestimate the learning and training this system requires. Even when supervised training is complete, the target attractor – what you call a template - remains very flexible and can change from cycle to cycle if it has to.” - “‘Has to?’” I asked. Lucy was off on the net somewhere, her face vacant. - “Has to. The system is trying to provide optimal reward for a given behaviour, right? It does this primarily on the basis of known optima we've established by recording from our four implants and from patients who don’t have a problem with whatever behaviour we're trying to reinforce. But, the software is also able to be creative and try variations on the target attractor - within limits - to find the permutations of the behaviour that the patient is most... eh... able to produce. Presumably the most comfortable attractor is the one that causes the least long-term damage. The real problem is getting rid of the noise from all the movemet, but...” - “No, hang on” Ike said. “People break their backs using rowing machines the wrong way all the time.” - “I said within limits” Meg replied sourly. “We’ve got the back muscles, remember? They're part of the target attractor. We set it up during training.” Everyone was silent for a while. - “See what I have to deal with back here while you guys are on ‘conference’?” Lucy said sarcastically, smiling. - “Look, here's how to think about it” Meg continued. “Reward, dopamine in particular, increases signal-to-noise in the brain. Right? This is not even argued about anymore, but what it means is that when dopamine increases – when reward increases – strongly active brain networks get strengthened and continue to function while weakly coupled ones break apart (Gruber et al., 2006; Djurfeldt, 2001). Dopaminergic networks survive. We’ve been over this endlessly. But now apply that signal-to-noise principle to decoding neural activity. In our case the signal is a repeating sequence of motor-related potentials in EEG data from the patient's brain.” - “We talked about this, this morning” I said. “I think we understand the basics. We're just not sure how the actual... eh... match happens. A recorded pattern of electrical activity is sufficiently similar to a template – a target attractor – you've somehow built up...” - “Not somehow, I told you...” Meg began. - “...and that triggers...” I continued. - “Adjusts” Meg corrected. - “...and that adjusts the stimulation parameters of the DBS electrodes in the patient’s reward system. The result is really smooth artificial motivation towards the template attractor, right?” - “And it obviates the need for human supervision.” Lucy chimed in. “You realize what that means for PR, right?” - “Maybe,” I said “but this template... we want you to explain it to us. It's always some kind of cyclic attractor, right?” - “I was explaining. You interrupted to say that you know what motor neurons do, it was very helpful” Meg said sarcastically. We looked at each other for a while. - “You wanted Ike hired. You did those scans. You talked D into doing the interview. Why didn't you ask us first? What else have you got planned?” she said eventually. - “Can we finish the previous conversation first?” I replied. - “But you said I could ask anything I want” she said with something resembling mock pouting. Ike was right; something was different, about both of them. - “You can” I said “but first we had our question. You were explaining about the signal-to-noise principle. I assume the signal for the EEG-to-iPlant loop is some kind of above threshold match between EEG data and a template activity pattern that we know drives a behavior that needs reinforcement.” - “Well...” Meg said “that's how it started.” - “The way she explained it to me” Lucy said. “You really have to get to know every motor network really well before you figure out how to see it and reward it properly. They're pretty complex, and highly variable. She's been working on this non-stop since we got back.” She stopped talking. - “That was three weeks ago” I said. ”What do you mean non-stop? Meg, how much have you been working?” - “Eh...” Meg stalled. - “Hang on” Ike said. “You've been jacked in. While coding, you've been jacked in the entire time!” - “What?” I said. - “Not the entire time” Meg said. - “It's OK" Lucy said. "She's been low-balling the current and I make sure she gets at least five hours of sleep every night. - “That's not enough” I said. “How did you create the schedule? What kind of schedule is this??” I was getting angry. - “I haven't slept more than five hours in a night since I was eight” Meg said defensively. A tense silence followed. I was trying to decide which issue to deal with first: their security breach, or the fact that Meg appeared to have been running an essentially permanent sequence of reinforcement schedules on her iPlant for three weeks; far in excess of anything we had ever planned or received approval for. The latter. Pull rank. - “Lucy” I said. “While I'm away you're senior resident in our group, I mean lab. That makes Meg your personal, medical, legal responsibility...” - “Chris” Lucy said, holding up her hands in the sign of calm. “I've got her entire physical. We've been making it together actually: blood composition, all major hormones and second messengers, cross-matching with urine and saliva. We even got a bloody biopsy at her latest operation; we've been breaking it down in the wet-lab for over a week actually... anyway she's fine.” I had to suppress a smile. - “Look,” she continued “what are you in this for, really? What's your motive?” - “Fear of death, I think” I replied. “iPlant-driven research is the best way I can imagine for developing really powerful medical technology.” - “Well, me too” she said. “And I'm telling you Meg is fine. She's cool. She's... really something.” She looked over at Meg. Meg sat very still, with a small smile on her face. I imagined, or tried to imagine, the two of them deciding, at some point, to go in and break the lock on the computers. That lock required at least three of us to be present on campus to run untested reinforcement schedules on our iPlants. Granted, the lock had been installed just after the accident with Ike and Lucy, almost a symbolical gesture since we all knew Meg could probably break it if she wanted to. Well, she had. They had. - “Meg” Ike said all of a sudden. “Do you... do you wanna go to the thing, the conference, in Austria? The optogenetics thing. When I did five days in a row you know, well it really kind of crept up on me. It's good to take a break, while it's still good. Conference would give you a reason to get away.” Everyone was silent and for a long time nothing happened. Eventually, Meg started to laugh; a low, tapping chuckle. For a long time, there was only the sound of Meg quietly laughing. She wasn't looking anywhere in particular. 
|