2.4% of College Students Misused Stimulants in 90 Days. NEUROPUMP Would Replace Forced Focus With an Opioid Reward Loop.

Two point four percent sounds small until you translate it into a campus. In a national study of 224,469 undergraduates at 395 U.S. universities, 2.4% reported misusing a prescription stimulant in the previous three months. That is roughly one student in every 42 during a single academic quarter. At a university with 30,000 undergraduates, the implied count is more than 700 students using someone else's medication, taking more than prescribed, or otherwise using stimulants outside medical direction.

The same study found a supply effect hiding inside the culture. For every one-percentage-point increase in a university's ADHD-medication treatment rate, prescription-stimulant misuse was 7% higher on a relative basis after adjustment. This is not evidence that legitimate ADHD treatment causes abuse, and the authors did not claim that. It is evidence that the informal market expands where pills are more available. The campus library during finals is not merely a place of learning. It is a distribution network shaped by deadlines.

We call this the Adderall crisis because Adderall is the recognizable brand, but the phrase obscures the deeper problem. Students are not generally seeking the pleasure of doing calculus at 2 a.m. They are buying the ability to override fatigue, boredom, anxiety, and the impulse to do almost anything else. The drug forces arousal. It does not make the theorem, case brief, or blank document intrinsically rewarding. The market exists because modern institutions demand hours of cognitive labor from a nervous system whose reward machinery was not designed around ten-page papers.

The Device That Changes the Question

NEUROPUMP is a defensive prior-art concept, not a product. Its proposed architecture combines a wearable micro-pump, a camera, and a vision-language model. The user defines a target—read constitutional law, write Rust, solve plasma-physics equations. The model watches the work, verifies that the activity matches the goal, scores sustained engagement, and authorizes a tiny transdermal dose of mitragynine, the main psychoactive alkaloid in kratom. Reward follows verified effort. Repeat the loop enough times and the desk, book, or code editor should become a conditioned cue for anticipation rather than dread.

That is a different theory of enhancement from Adderall. Stimulants increase the system's capacity to stay awake and suppress distraction. NEUROPUMP would attempt to rewrite the system's preferences. It does not ask, “How can we make a person work longer?” It asks, “Can we make the person want the work?”

Pavlov did not make dogs intellectually appreciate metronomes. He made a neutral signal predict a biological reward. NEUROPUMP applies the same logic to deep work: genuine engagement becomes the signal; an opioid-receptor effect becomes the reward. The ambitious social claim is that, after enough pairing, the work itself could inherit some of the positive valence. The unsettling possibility is the reverse: remove the chemical reward and ordinary learning feels flatter than it did before.

The Camera Is the Real Invention

The pump attracts attention because it touches the body. The VLM is what makes the concept historically new. A timer can reward twenty-five minutes in a chair. A keystroke counter can reward output. Neither knows whether a student is deriving an answer, copying one, reading a paper, or hiding a video behind the document. A multimodal model can inspect visible text, application context, posture, gaze direction, typing cadence, and the sequence of actions. It can plausibly tell the difference between reading a case and scrolling a feed.

But plausibility is not verification. A camera can observe behavioral proxies for thought; it cannot see comprehension. It may reward the appearance of concentration, privilege bodies whose gaze and movement match its training data, and penalize neurodivergent work styles. A student who paces while thinking may score lower than one who sits still while copying. Once a drug dose depends on a model's classification, a false positive is not a bad recommendation. It is an unauthorized pharmacological event.

The incentives also produce a new version of Goodhart's law: when engagement becomes the dosing metric, users will learn to perform engagement for the model. The system will respond with more cameras, biometric signals, screen capture, artifact inspection, and anti-spoofing. The endpoint is not simply personalized education. It is a proctor with permission to alter your neurochemistry. On-device processing and zero video retention would reduce the privacy risk, but they would not erase the power relationship between the person being scored and the model controlling the reward.

Kratom Is Not a Safety Loophole

Mitragynine is interesting precisely because it does not behave like a conventional full opioid agonist. It acts as a partial agonist at the mu-opioid receptor and is slowly converted in the body into the more potent 7-hydroxymitragynine. In animal research, the conversion rate appears to saturate, producing a built-in ceiling on mitragynine-induced respiratory depression. That is the scientific kernel behind NEUROPUMP's proposed safety advantage.

The words “animal research” are doing essential work. A ceiling effect observed preclinically is not a guarantee that a wearable device can safely deliver repeated rewards to humans. It does not eliminate tolerance, withdrawal, substance-use disorder, liver injury, interactions with other drugs, manufacturing impurities, or variability in metabolism. “Plant-based” describes origin, not risk. Belladonna is plant-based. So is nicotine.

The FDA says there are no legally marketed prescription or over-the-counter drugs containing kratom or its alkaloids in the United States and warns of risks including liver toxicity, seizures, and substance-use disorder. NIDA's summary is more nuanced: serious outcomes are rare, and deaths linked to kratom products have nearly always involved other substances or contaminants. Those positions can both be true. Kratom may be less lethal than classical opioids and still be far too uncertain for automated productivity dosing.

What the September 2026 Trial Will—and Will Not—Tell Us

For the first time, purified mitragynine itself is about to enter a controlled human trial. In June, NIH announced that its Investigational New Drug application had taken effect, clearing a Phase I study of oral MG001. The trial is estimated to begin in September 2026 and enroll 32 healthy adults. Four cohorts will receive a single 25, 50, 75, or 100 milligram oral dose or placebo under randomized, blinded conditions.

This is significant and narrower than it sounds. The trial is designed to measure safety, tolerability, and pharmacokinetics after a single oral dose. It is not testing transdermal delivery, repeated micro-dosing, cognitive enhancement, reward conditioning, or whether mitragynine can make work feel good. NIH's intended therapeutic target is opioid-use disorder, not exam preparation. If MG001 clears Phase I, NEUROPUMP moves from “compound not studied alone in humans” to “compound with first-in-human safety data.” It does not become a plausible consumer device overnight.

When Learning Feels Good, Who Owns the Feeling?

Suppose the engineering and pharmacology eventually work. The first-order benefit is enormous. Students who struggle with task initiation could build durable study habits. Adults could retrain into difficult fields without spending every evening fighting the reward asymmetry between a textbook and a feed optimized by thousands of engineers. Society already permits caffeine, nicotine, prescription stimulants, antidepressants, and algorithmic gamification to shape performance. A closed-loop reward device differs in degree, timing, and precision, not in the basic human desire to alter motivation.

The second-order effects are where the concept becomes political. If chemical conditioning raises output, elite schools and employers will create pressure without issuing a mandate. “Optional” enhancement becomes functionally compulsory when the enhanced cohort produces twice as much. Insurers could subsidize it for students with diagnosed attention disorders while denying it to everyone else. Wealthy families could condition children around mathematics years before public schools can afford the hardware. A productivity divide becomes a preference divide: one class is trained to experience difficult work as rewarding while another is told to develop discipline.

There is also the question of autonomy. A person can reasonably choose to condition their own habits. But who sets the goal domain? A student might choose constitutional law. An employer might choose claims processing. A military might choose target identification. The same loop that makes self-directed learning pleasurable can make institutionally approved labor pleasurable. The dystopia is not a boss forcing a drug into workers. It is a compensation package that rewards consenting workers for letting software decide which mental states deserve relief.

The Regulatory Maze Is the Point

NEUROPUMP would not inherit kratom's current retail status. A pump that meters a pharmacologically active compound in response to software would likely be regulated as a drug-device combination product. Its sponsor would need to establish the identity and purity of the active ingredient, dosing accuracy, transdermal pharmacokinetics, software reliability, human-factors safety, cybersecurity, and clinical benefit for a defined indication. A productivity claim is not an indication the FDA has an established pathway to approve.

Meanwhile, the legal status of botanical kratom remains fragmented. It is not federally scheduled, but states and localities impose their own bans and consumer-protection rules. FDA also distinguishes ordinary leaf material from concentrated 7-OH products, which have far greater mu-opioid potency and have drawn enforcement action. A NEUROPUMP using purified mitragynine would sit in the least forgiving part of this landscape: too drug-like to be a wellness gadget, too novel to borrow an existing approval, and too behaviorally consequential to evaluate on pharmacology alone.

The Strongest Counterargument

The strongest case against NEUROPUMP is not that it sounds strange. It is that the premise may be wrong. Deep intellectual work can already become rewarding through mastery, curiosity, social meaning, and flow. The problem may be institutional design: overloaded students, poor teaching, sleep deprivation, economic pressure, and coursework disconnected from purpose. If a class requires a drug-triggered reward to tolerate it, the humane intervention may be to redesign the class.

A closed-loop chemical reward risks medicalizing a conflict that is partly social. Universities compress learning into high-stakes deadlines, employers colonize attention, and platforms monetize distraction. NEUROPUMP could help individuals survive that system while making the system harder to challenge. Adderall misuse already performs this function: it lets students adapt chemically to unreasonable conditions. A more elegant adaptation may be more dangerous because it makes the mismatch invisible.

The Bottom Line

NEUROPUMP is not ready to build. Its proposed transdermal dosing is unproven, its reward effect is uncertain, its safety case rests heavily on animal data, and its VLM cannot reliably verify an internal cognitive state. The September NIH trial will answer only the first human questions about single-dose oral mitragynine. Anyone presenting the concept as a safer Adderall replacement is skipping years of evidence.

But as a thought experiment, it exposes something real. The campus stimulant market is not merely a drug problem. It is a market signal that sustained cognitive labor often loses the competition for human attention. We have spent two decades making distraction more rewarding. NEUROPUMP asks what happens when the same closed-loop optimization is pointed at learning instead.

The answer will not be settled by whether mitragynine has a respiratory ceiling or whether a camera can detect fake studying. It will be settled by a more difficult boundary: when technology can choose which behavior feels good, does that expand human freedom—or transfer control of motivation to whoever defines “productive”?