012 - Neurotechnology
Brainwaves, behavior & blood
Editors Note: This blog article is a summary of conversations at an in-person event held in San Francisco on 2024-03-31 facilitated by
The future of mental health lies not in categories but in biology. The promise of neurotechnology is oft-fetishized, but the path forward requires a sober disentangling of social context from physiology, and an embrace of new approaches to treatment.
As French philosopher Michel Foucault argued in his seminal work "Madness and Civilization," our understanding of mental illness has long been shaped more by societal norms than by science. The Diagnostic and Statistical Manual of Mental Disorders (DSM), the bible of psychiatry, defines disorders based on clusters of symptoms - observable behaviors that deviate from the norm. But this symptomatic approach fails to capture the complex biological underpinnings of mental health.
Many have sought to move beyond this limited paradigm. The “Research Domain Criteria” (RDoC) was created in the early 2010s to conceptualize mental illness as brain disorders and change how they are diagnosed. It was a large part of former NIMH director Thomas Insel’s strategy for changing how we understand and treat mental illness, leading to large investments in brain imaging technologies like fMRI. More than a decade later, it hasn’t impacted clinical diagnoses much (DSM is still overwhelmingly used) and fMRI itself hasn’t transformed our understanding of mental health.
All is not lost however! RDoC is still a worthwhile perspective and fMRI is not the end of biologically-inspired explorations of our mental life. Researchers are turning to new tools and technologies. Brain organoids, lab-grown tissues that mimic the structure and function of the brain, offer a promising platform for probing the molecular basis of neurological diseases. Advances in computational modeling and artificial intelligence may help unravel the intricate interplay of genes, proteins, and neural circuits that give rise to both healthy cognition and psychopathology.
The search for biological underpinnings also extends beyond the brain. The molecular landscape of mental health is vast and largely uncharted, encompassing not just neurotransmitters and receptors, but the myriad biochemical pathways that regulate mood, cognition, and behavior. Researchers are increasingly looking to peripheral tissues like blood and cerebrospinal fluid for biomarkers of psychiatric disease. Genetic studies are uncovering the complex interplay of risk alleles and environmental factors that shape individual susceptibility to disorders like schizophrenia and autism.
All this said, there is a growing recognition that mental health cannot, or should not, be reduced to molecules alone. Behavior, the final output of the mind, is a rich source of data that can inform diagnosis and treatment. This level of analysis and intervention may be cheaper, more efficient, and more effective, depending on the context. In addition, while behavioral analyses has traditionally been underinvested in compared to biological treatments, there are reasons to believe behavioral data will be transformed dramatically by the current technological moment, both due to the ubiquity of high resolution behavioral data sources (smartphones, online data) and our ability to process the data.
For instance, digital phenotyping (the use of smartphones and wearables to passively collect behavioral data) offers a window into the daily rhythms of activity, sleep, and social interaction that are often disrupted in mental illness. This offers a far richer “behaviorome” than we’ve had before. Advances in machine learning also expands the kind of behavioral data that is relevant and useful. Voice recordings, journals, online posting, and even our presentation in a natural conversation are all kinds of behavior that likely hold signal relevant for diagnoses and intervention, soon available due to AI. A massive change may come with the advent of LLMs. Rather than needing to compress human behavior into lossy forms like coded surveys, we can process more complex behavioral data. Perhaps with machine learning we can take the raw audio signal of a person’s interactions (or self-report) and have a rich view of their emotional and cognitive state, and how this state has changed over time.
Of course, these advancements and hopes are embedded in an unequal world which favors those who are best represented in medicine, science, and the datasets themselves. Ensuring equitable access to diagnosis and treatment will require grappling with the thorny issues of diversity and inclusion in medical research. Historical biases in datasets and algorithms risk perpetuating disparities in care. Passive data collection through smartphones and wearables holds immense potential for personalizing interventions, but raises urgent questions about privacy and consent.
Even more fundamentally, disentangling the threads of nature and nurture in mental health remains a daunting challenge. The experiences of childhood, the learned behaviors and coping mechanisms of adulthood, the stigmas and stressors of social context - all these shape the contours of the mind as surely as any chemical imbalance. Diagnostic categories based solely on biology will always be incomplete.
Nonetheless, a new era of precision psychiatry is on the horizon. Blood tests for depression, brain implants for addiction, psychedelics for PTSD - the future of mental health care will be multimodal and personalized, integrating molecular data with brain imaging, genetic screening with cognitive behavioral therapy, passive sensing with active intervention. The road ahead is long and winding, but the destination is a world in which the mind is no longer a black box, but a knowable, treatable entity - a world in which mental healthcare is not a guessing game, but a science.
As the pioneering psychologist William James wrote in his 1890 treatise "The Principles of Psychology": "The greatest revolution of our generation is the discovery that human beings by changing the inner attitudes of their minds, can change the outer aspects of their lives." More than a century later, we stand poised to fulfill that promise - not through the power of positive thinking, but through the marriage of mind and molecule, of silicon and synapse, of blood and behavior. The neuro-fetishists may yet be proven right, but only if we have the wisdom to see the brain in context, and neurotechnology as one kind of advancement supporting the human mind’s improvement.
Notes from the conversation
Applied neuroscience can sometimes devolve into "neuro-fetishism" without providing real value.
Moving away from symptomatic behavioral categories (like the DSM) towards a more biological understanding of mental health issues could improve diagnosis and treatment.
Mind reading technology is a potential future application of neurotechnology, distinct from healthcare applications.
Computational modeling and AI could help understand the complex biological systems involved in mental health.
Organoids are a promising tool for drug discovery, but are not yet good enough for understanding disease mechanisms.
Ketamine is an example of a drug with potential for treating depression, but predicting individual response is challenging.
Molecular data may be more useful than brain imaging for understanding and treating neurodegenerative diseases.
Diversity in datasets is important for ensuring equitable access to treatments.
Single-patient trials and continuous monitoring could enable more personalized approaches to mental healthcare.
Passive data collection through devices like smartphones could provide valuable behavioral insights relevant to mental health.
Form factor is a major challenge for consumer adoption of neurotechnology devices.
Open-source and decentralized approaches to neurotechnology development could accelerate progress.
Childhood experiences and learned behaviors play a significant role in adult mental health, beyond just biology.
fMRI mind reading has progressed to the point of being able to reconstruct images from brain activity.
Blood biomarkers could provide valuable insights into mental health.
There is a gap between the tech industry and the biopharma industry that needs to be bridged for successful commercialization of neurotechnology.
The brain's growth trajectory in autism may differ from neurotypical development.
Directly sampling diseased brain tissue is challenging, so alternative approaches like molecular analysis of proxy tissues are being explored.
Neurotechnology could enable faster drug response detection, accelerating research.
Social context and stigma need to be disentangled from underlying physiology in mental health research.
Questions
How can we ensure that neurotechnology provides real value beyond "neuro-fetishism"?
What biological markers could replace symptomatic behavioral categories for mental health diagnosis?
What level of signal fidelity would be needed for effective mind reading technology?
How can computational models incorporate the complex social and environmental factors that influence mental health?
What advances are needed to make organoids more useful for understanding disease mechanisms?
How can we better predict individual responses to drugs like ketamine?
What molecular data sources are most promising for understanding neurodegenerative diseases?
How can we ensure diversity in mental health datasets to enable equitable access to treatments?
What are the ethical considerations around single-patient trials and continuous monitoring for mental health?
How can passive behavioral data be used effectively without compromising privacy?
What form factors for neurotechnology devices would enable widespread consumer adoption?
How can open-source and decentralized approaches be incentivized and supported?
What role should childhood experiences and learned behaviors play in mental health diagnosis and treatment?
What are the limits of fMRI mind reading, and what other technologies could enable similar insights?
Which blood biomarkers are most promising for mental health applications?
How can the tech industry and biopharma industry collaborate more effectively?
What can we learn from the differences in brain development between autism and neurotypical trajectories?
What alternative approaches to direct brain tissue sampling are most promising?
How can faster drug response detection be integrated into mental health treatment protocols?
How can social context and stigma be effectively accounted for in mental health research?