Jacques Carolan | The future of brain health

Beatrice:

I’m very happy to be joined today by Jacques Carolan. I think we’ll just dive straight in. You are a Program Director at ARIA, which is a very interesting organization. Could you start by telling us who you are and what you do as a Program Director at ARIA?

Jacques:
Absolutely—thank you so much. I’m excited to be here. ARIA is a UK-based R&D funding agency with the express mission of pursuing research in areas that seem speculative but, if true, could be transformational on a global scale. We were established by an Act of Parliament, which gives us broad scope to pursue bold, long-term visions. Some listeners might be familiar with DARPA—that’s a reasonable first-order approximation of what we do.

Our work is organized around programs: mission-driven R&D efforts where, if we can demonstrate certain things, we can really move the dial in areas of massive impact. A unique part of the model is the Program Director role. I’m a Program Director—my job is to set the research vision, direct funds, and then, once the program is running and projects are selected, to manage them. I speak with teams regularly, make sure things are going well, double down when something works, and pivot or shut down projects when needed.

We’re working across a number of areas: next-gen computing hardware, AI safety, robotics—and I joined ARIA to stand up the neurotechnology program, which is very exciting.

Beatrice:
Neurotech has been booming—at Foresight, it’s our fastest-growing program with the most interest. What do you work on in that program? What opportunities are you looking at?

Jacques:
For listeners not familiar with neurotechnology, a quick primer: it’s any way to interface with the brain—either to read out signals or to modulate brain activity in a targeted way.

Why do this? Take individuals with severe motor conditions who can’t communicate or move, perhaps due to spinal cord injury. You can implant electrodes into the brain’s movement regions and translate intentions into motor outputs—say, moving a cursor. That lets people access computers or control objects. Many listeners will know companies like Neuralink; there’s huge excitement in the space.

When I think about brain disorders, I think broadly—beyond severe movement disorders—to neurological and neuropsychiatric conditions: anxiety and mood disorders, addiction, psychosis. The spectrum is much broader. We’re seeing early signals that neurotechnologies can address a wider range of conditions than movement alone.

So at ARIA, the opportunity is: if we can build next-generation precision neurotechnologies, we can potentially address a much broader patient population than before.

Beatrice:
It’s probably more advanced than many people think. We’ll talk much more about neurotech today, but I’d also love to hear about you. You have an unusual background—you were in physics before this. How did you make that jump?

Jacques:
It’s been a journey. I spent the first part of my career in applied physics, exploring how to use light for computation—ranging from quantum computers to optical computing for AI. I enjoyed the problems and the people, but two things struck me. First, the likelihood that the widget I was building—an artificial atom, a single-photon source—would be useful was small. Second, even if it were useful, the time horizons and capital needed were massive.

I wanted to build technologies that could help people in the near term. Around the pandemic, I wondered if I could become a neuroscientist. I’ve always been fascinated by the brain and how physical substrates compute. I spent six months reading papers, textbooks, and talking to people. Physicists said, “This is a dumb idea—you’ve got a great career.” Neuroscientists said, “This is awesome—you should do it.” Turns out many physicists move into neuroscience.

I found a job in a systems neuroscience lab. I went from microchip fabrication to mouse brain surgery, optogenetics, and imaging. For context, I don’t even have high-school biology; I learned it all for the first time. I was the only physicist in the room, but I learned a lot. Coming from hardcore engineering, I felt the tools we used to study the brain were relatively primitive. I saw an opportunity to close the gap—bring cutting-edge tech to the brain. That’s when ARIA was hiring its first cohort of Program Directors, I pitched this vision, and ended up here.

Beatrice:
Such an inspiring story of shifting careers without neuro credentials on paper.

Jacques:
I had no neuroscience publications, but the lab’s lasers were similar to those I used in my PhD; I could build optical imaging systems. My advice: take something you’re good at and apply it to a new area. Also, talk to people who’ve switched careers—we’ll all likely have many careers.

Beatrice:
Let’s talk about what you focus on now. One of your main areas is precision neurotechnologies. What do you mean by that?

Jacques:
Neurotechnologies, broadly, aim to address a wide set of neurological and neuropsychiatric conditions—anxiety and mood disorders, epilepsy, addiction, pain—conditions many of us know or have experienced. The societal burden is huge.

From basic neuroscience (e.g., large-scale transcriptomics in mouse and other brains) and human neurophysiology, we’re discovering many of these conditions are disorders of circuits—distributed networks around the brain. For a motor BCI, you target primary motor cortex; for many other conditions, it’s not one region but a distributed network.

So gaining circuit-level access is key. We launched the Precision Neurotechnologies program to increase the precision and circuit-level access of current tools. Compared to drugs—which are imprecise and flood the whole body—precision promises fewer side effects and more personalized treatment. That’s our bet.

Beatrice:
Makes sense—precision is especially important in the brain. What new tools are you developing? Any concrete examples?

Jacques:
Helpful context: as a Program Director, I set the north star—I don’t know exactly how to get there. If I did, I’d start a company. We set direction, then run open calls. Seeing the applications, I realized we had something special. A few examples:

1. Ultrasound to read and modulate mood
   A collaboration among Forrest Neurotech, the NHS, and the University of Plymouth: can we read out brain state (via blood-flow changes akin to Doppler) and target specific circuit elements to perturb mood—all non-invasively? Neurons have mechanosensitive ion channels—ultrasound can “squidge” them and induce firing reversibly, without damage. This is in a clinical trial.
2. Closed-loop gene therapies
   A team at UCL is developing closed-loop gene therapies initially for epilepsy. The therapy senses increased activity (e.g., a seizure) and upregulates potassium channels—putting the brakes on neurons—to stop the seizure. We’re funding them to expand access in different brain regions, explore ultrasound activation, and apply to other conditions.
3. Biohybrid interfaces
   Implanted electrodes are mechanically mismatched to soft brain tissue, leading to scarring and signal issues, and they don’t provide cell-type access. If you could grow neurons to serve as an interface, you might achieve better circuit specificity. We’re funding teams to explore something akin to a living DBS electrode.

The idea is that different conditions require different levels of precision. ARIA can spread bets across approaches and support multiple paths to success.

Beatrice:
DBS stands for deep brain stimulation, right?

Jacques:
Yes—Deep Brain Stimulation.

Beatrice:
What are the bottlenecks in current neurotech?

Jacques:
There are many—that’s what makes it exciting. One I think about a lot: the trade-off between precision and invasiveness. The closer you are to neurons, the better you can record from or stimulate them. Precision is a proxy for functionality. How do we break the trade-off—achieve less-invasive tech with the functionality of high-performing implantables? That might require new physics, new biophysics, or new biologics. It’s an open question.

Beatrice:
You’ve compared today’s neurotech challenges to the evolution of cardiology. Why is that useful?

Jacques:
If neurotech can be massively impactful, we should take an honest look at scale—how many people will actually access it, and what are the barriers?

A good case study is DBS for Parkinson’s. It’s FDA-approved for ~20–25 years, reimbursed in the US, and we understand responses reasonably well. But if you plot procedures over time versus Parkinson’s prevalence, we’re massively underpenetrated, well below 0.5% in the US—our best-case context. We’re not increasing at the right rate.

Two buckets of blockers:

• Efficacy: Parkinson’s is neurodegenerative; stimulation mainly treats motor symptoms, not the disease. If it were a slam dunk, penetration would be higher. This motivates Precision Neurotechnologies—better circuit tools to address disease.
• Procedural burden: I recently observed a DBS procedure. It’s a serious surgery with infection and bleeding risks—driving patient and clinician hesitancy. There are also workforce constraints; in the UK, perhaps two dozen functional neurosurgeons can perform it.

So, where have we massively scaled an intervention before? In the 1950s, treating heart block required open-heart surgery (thoracotomy) to suture electrodes—hugely invasive, last resort. The invention of the transvenous lead transformed it into a low-risk outpatient procedure, now performed millions of times. That solved invasiveness, drove device innovation (early pacemakers were terrible—nuclear batteries, broken leads—but improved rapidly), and scaled from a handful to hundreds of thousands of procedures in ~20 years.

What’s the analog for the brain? It’s not perfect—we better know heart targets—but it’s a useful lens: solving invasiveness can unlock scale and innovation.

Beatrice:
Will reducing invasiveness shift interventions from last resort to proactive brain health?

Jacques:
There will be buckets. For mild to moderate cases, people won’t undergo serious brain surgery. But for treatment-resistant patients, implants may be necessary. In parallel, we can pursue proactive brain health. Over time, there may be overlap, and I do think how we think about brain health will change.

Beatrice:
Could brain health become like physical fitness—proactive and guided?

Jacques:
I think so. We’re seeing uptake in wearables. The key is the science underneath. We’re getting better evidence that different circuits underlie different conditions. For example, work from Stanford (e.g., scanning people diagnosed with depression and analyzing functional connectivity) shows subtypes of depression that better predict treatment response. That’s an existence proof that understanding brain state improves outcomes.

In the future, lower-resolution at-home devices might guide behavior (“go for a walk,” “do this therapy”), though that’s an open question. The science suggests it’s plausible.

Beatrice:
When you talk about your work, which conditions come up most?

Jacques:
Mood and anxiety disorders, epilepsy, addiction—and a big one: chronic pain. Many people live with chronic pain; there are early signals that very targeted brain modulation can alleviate it. As we do more, we’ll likely add more conditions. We’re at the beginning of understanding how targeted interaction improves a range of disorders.

Beatrice:
How about the overlap with AI? Can AI accelerate discovery in neurotech and neuroscience?

Jacques:
Absolutely. The Stanford example—deriving biotypes from big connectivity data—uses AI. But they start with biological priors about implicated circuits. That’s important. A pure black-box approach can fail in a system as complex as the brain. Start with priors.

We also see AI in DBS work. For example, in treatment-resistant depression, groups implant electrodes in mood-related regions and record electrophysiology. Some studies show electrophysiological signatures preceding depressive relapse by weeks. If that’s true, you could warn patients and adjust care. AI underpins much of this, but with biological priors and clinician involvement.

Beatrice:
Looking ahead 20 years, what would make you proudest?

Jacques:
By design, I don’t know which projects will succeed. I’d love to see our technologies in the clinic helping people, but there’s uncertainty. Another thing ARIA can do is build community—bring together ambitious technologists and scientists to dream big and tackle hard problems. Even early in the program, there’s so much happening: activation partners, a neurotechnology accelerator, exploring Focused Research Organizations in the UK. There’s a sense of a phase change. If that proves true in 20 years, I’ll be delighted.

Beatrice:
Are there similar initiatives to ARIA elsewhere?

Jacques:
Yes—SPRIND in Germany; related efforts in Japan; something in NATO’s orbit; and in the US, multiple ARPA agencies (ARPA-H for health, ARPA-E for energy, DARPA, etc.). These models require a strong talent base. The UK has an ambitious, entrepreneurial community—early-career people who want to build. ARIA gives permission to dream bigger.

Also, you don’t have to be from the UK to work with ARIA. We’re very international in hiring Program Directors. For funding, our responsibility is to UK taxpayers, so most work is in the UK, but we can fund international partnerships where the majority is UK-based and a top partner (e.g., US) enables otherwise impossible work.

Beatrice:
Back to your first program: on your site you talk about Measure–Model–Perturb. I had to look up “perturb.” Can you walk us through that framework?

Jacques:
One awesome thing at ARIA is we publish documents and get feedback—we build in public. Measure–Model–Perturb asks: with better circuit-level tools, can we move the brain from a pathological state to a more physiological one? That’s a control problem.

You have a dynamic system (the brain). Define a brain state (a low-dimensional representation of many neurons’ firing). You want to nudge from, say, a seizure state to a non-seizure state.

• Measure: read out the brain.
• Model: build a computational model that tells you how to perturb to reach the desired state.
• Perturb: apply the intervention.
• Then measure again, update the model, and iterate.

It’s about predictably driving the brain from unhealthy to healthy states, leveraging advances in computational methods and AI.

Beatrice:
You also set precision metrics. Why is that kind of systematic framework important, and what do you measure?

Jacques:
ARIA is metrics-driven. We ask teams to do hard things on short timelines, so we de-risk with concrete metrics—so we know if it’s working. Applicants might say it was painful to write, but it matters.

I identified four key aspects of circuit interaction:

1. Cell-type specificity: The brain has thousands of cell types—excitatory, inhibitory—doing radically different downstream things. Specificity matters.
2. Field of view: Access as much of the brain as possible.
3. Voxel size: How precisely can you target specific circuit elements?
4. Number of voxels / channels: How many independent “handholds” do you have on the brain?

Why? Consider Hebbian plasticity (“neurons that fire together wire together”). If a circuit is under-regulated, turning on two parts might strengthen connections; if over-regulated, timing-based interventions might down-regulate it. Circuit-level access is essential—hence the (annoying) equation in the document.

Beatrice:
Precision Neurotechnologies is your first program in Scalable Neural Interfaces. You’ve hinted at a second program?

Jacques:
You nailed the ARIA terminology. We have the broad area Scalable Neural Interfaces, with the first program (Precision Neurotechnologies). Outside that, we have seed projects—smaller, more speculative bets (to be announced soon). I’ve also been thinking about a second program; I floated ideas in a Substack post on massively scalable neurotechnologies to gather feedback.

It’s become clear we need radically different ways to access the brain. In the 1950s, the transistor inspired breakthroughs in cardiology (transistorized metronome → pacemaker). Today’s frontier technologies are engineered biology and frontier hardware.

Questions we’re asking: Can we build very small devices that traverse blood vessels to reach target brain regions? Can we use cerebrospinal fluid routes? Can we leverage cellular systems—delivered peripherally—that autonomously home to brain targets? There’s a lot of white space here. We haven’t launched this program yet; we’re gathering feedback.

Beatrice:
These technologies touch something intimate—our brains and identities. What ethical and societal challenges do you see, and how do we make sure tech is distributed and used safely?

Jacques:
Initially, these technologies will be used in severe, treatment-refractory patients. We’re a long way from healthy people electively undergoing serious brain surgery. But ethical issues remain—identity, cognition, equity. We’ve funded a whole stream at ARIA to explore this: building more equitable technologies, engaging people with lived experience, and integrating those insights into our program.

Beatrice:
You’ve done patient engagement work. What have you learned?

Jacques:
It’s complicated and individual. What I might consider non-invasive could feel invasive to someone else. Wearing a headset might be intrusive; in some cultures, mental illness has serious stigma, and a visible device can be a blocker. Shaving one’s head can be a blocker. The key is engaging people with lived experience to understand these factors.

We help our teams work with affected individuals to identify key needs. That leads to better technologies and products, and it motivates teams—meeting the people who need these tools puts a fire under you.

Beatrice:
At the ARIA Summit, you had patients on stage. I recall someone with neural stimulation for depression—it was powerful.

Jacques:
That’s John Nelson—now a good friend. I urge listeners to watch his talk (search “John Nelson ARIA Summit”). Hearing him describe wanting to die every day, nothing helping, and then a treatment that saved his life—it still gives me chills. It reminds me why we need to build these technologies faster; delays cost lives. It sounds extreme, but that’s the mission.

Beatrice:
Have you had “wow” moments in research?

Jacques:
A few. One is seeing calcium indicators expressed in a live mouse. You image with a two-photon laser; neurons light up when they fire. There’s a mouse on the table, laser on its head, you record neurons, and when you clap, neurons light up. Seeing the brain as an input-output machine was incredible.

Another was at a functional neurosurgery clinic at Queen Square in the UK. An older woman with essential tremor underwent a thalamotomy (ablating the thalamus, a relay station). This procedure is done while awake. Before, she couldn’t draw a spiral. They targeted the thalamus and applied a small electrical current for about 60 seconds, and the tremor was almost completely suppressed. She could bring her hand to her face—she could drive again, enjoy a drink—things we take for granted. It was as close to magic as I’ve seen.

Beatrice:
Do you have favorite YouTube clips for people to see?

Jacques:
Work from Sergey … (on patients with motor neuron disease) decoding speech is incredible—implanting electrodes in speech-related motor areas (lips, etc.) and translating brain signals into speech in real time. Watching families speak with loved ones again is emotional. Also, many videos on tremor suppression are striking.

Beatrice:
If everything goes right, what’s your best existential-hope future for neurotech?

Jacques:
A world where neurological and neuropsychiatric diseases are no longer stigmatized; we view them as physical brain disorders and have treatments. We can identify subtypes and predict therapy response—it won’t always be an implant; it might be CBT. The science is nearly there, and now it’s about access. In the next 10 years, I believe our approach to brain health and treatment will radically change.

Beatrice:
That’s an important point: if we can show where behaviors come from, people don’t have to feel “crazy”—they can see the physical basis.

Jacques:
Exactly—100%.

Beatrice:
Where should listeners go to follow or get involved?

Jacques:
Visit aria.org.uk—you’ll find upcoming program and seed announcements. Click Follow for updates. Also check out the ARIA Substack—we publish monthly. We’d love your listeners to get involved.

Beatrice:
Perfect. Thank you so much, Jacques. It was great to hear about the exciting and important work you’re doing. I’m very curious to follow along.

Jacques:
Thank you—it’s been a real pleasure.