Scientific path
My scientific path has moved across neurophysiology, complex systems physics, computational neuroscience, biological computation, and NeuroAI. The institutional path matters, because each place sharpened a different part of the same question:
How do biological systems organize physical dynamics into computation?
That question did not begin as an abstract philosophical position. It grew out of concrete problems in neural data: sleep rhythms, electromagnetic fields, field potentials, cortical excitability, excitation/inhibition balance, seizures, thalamocortical dynamics, and the difficulty of interpreting collective neural activity across scales.
Over time, these problems pushed me toward a broader view: the brain is not merely an information-processing device implemented in tissue. It is a physical system whose computation is realized through dynamics, constraints, geometry, fields, cell types, recurrent interactions, and multiscale organization.
Early work: fields, sleep rhythms, and thalamocortical dynamics
My early research developed around multimodal neurophysiology, electromagnetic source localization, sleep rhythms, and thalamocortical oscillations.
This formative period included work at the HMS/MGH/MIT Martinos Center for Biomedical Imaging, the UCSD Multimodal Imaging Laboratory, and the MGH Cortical Physiology Laboratory.
The scientific problem was already a physical one: what do macroscopic measurements of brain activity actually measure? EEG, MEG, iEEG, and field potentials do not simply report “neural activity” in the abstract. They are shaped by source geometry, tissue properties, extracellular media, synchrony, spatial scale, and the organization of cortical and thalamocortical circuits.
This period shaped my long-standing interest in the relation between neural events and the physical measurements through which we observe them. Sleep rhythms — spindles, K-complexes, slow waves, and state transitions — became an especially important setting for this question because they expose the brain as a dynamical system moving between regimes.
Doctoral work: complex systems physics and cortical dynamics
I pursued my Ph.D. in the physics of complex systems and computational neuroscience at the Laboratory of Computational Neuroscience in UNIC-CNRS and Sorbonne Université.
My doctoral work focused on the electromagnetic signatures of human cortical dynamics during wakefulness and sleep. The central themes were spectral dynamics, frequency scaling, self-organized criticality, invasive ensemble recordings, and the network organization of excitation and inhibition.
This was the stage at which the physics of neural activity became central to my thinking. I became interested not only in what neural signals correlate with, but in what kind of physical processes generate them. Why do neural recordings show scaling? What does it mean for cortical dynamics to approach critical-like regimes? How should one interpret field potentials when the extracellular medium itself is not an idealized passive resistor? How do excitation and inhibition organize the state-space of cortical activity?
These questions pushed me away from treating neural data as merely a statistical object and toward treating it as the observable trace of a physical dynamical system.
E/I balance, seizures, and state-dependent cortical computation
After the Ph.D., I continued this line of work as an independent institute technology fellow at Harvard’s Wyss Institute for Biologically Inspired Engineering and the New England Complex Systems Institute.
This period extended my work on neural ensembles toward state-dependent rhythmic activity, multiscale excitation/inhibition balance, and pathological dynamics such as seizures.
The central lesson was that neural computation cannot be separated from neural state. The same circuit can operate differently depending on the balance of excitation and inhibition, the sleep-wake regime, the history of recent activity, the degree of synchrony, and the proximity to pathological transitions.
This work strengthened a theme that still runs through my current research: computation in the brain is not only a matter of connectivity or representation. It is also a matter of dynamical regime.
MIT Physics and CBMM: information, causality, and excitable media
I later joined MIT Physics and the Center for Brains, Minds and Machines, where my work increasingly connected neuroscience, computation, and physics.
At this stage, I became more interested in information-theoretic and causal descriptions of biological dynamics. I studied emergent phenomena and causally linked dynamics across macroscopic and microscopic scales in excitable media, and continued work on pairwise and higher-order interactions of inhibitory and excitatory neurons during wakefulness and sleep.
This period broadened the question from “what are the dynamics of cortical activity?” to “how do multiscale biological dynamics compute, coordinate, and control?”
That shift is important. It made computation less of a metaphor and more of a physical problem: how does a system with local interactions, distributed constraints, recurrent dynamics, and state-dependent transitions produce organized function?
Biological computation and compositionality
At the Allen Discovery Center at Tufts, my work expanded from neural computation toward biological computation more generally.
There, I developed mathematical frameworks for thinking about physical and biological computation, including category-theoretic approaches to compositionality, regulatory mechanisms, and adaptive biological organization. I also worked on models of cellular homeostasis and reconfiguration in changing environments.
This phase helped crystallize a broader view: biological systems compute not only because they process signals, but because they compose physical processes into reusable, adaptive, multiscale organizations. Cells, tissues, neural circuits, and organisms all pose versions of the same problem: how can physical systems preserve identity, adapt to perturbation, and generate functional transformations?
This is where my interest in physical computation, biological computation, and compositionality became explicitly connected.
Current work: MIT, McGovern, compneuro@MIT, and N³HUB
I am currently at MIT, focusing on complex systems approaches to collective neural computation, foundations of physical and biological computation, and bio-inspired intelligence.
My current work is connected to compneuro@MIT and N³HUB, which is organized around three core axes:
- NeuroPhysics
- NeuroComputation
- NeuroDynamics
These three axes form the core of what I call the N³ program. On my own site, I extend this to N⁴ by adding NeuroAI: the attempt to translate principles from neural and biological computation into artificial systems.
The current research program brings together several threads:
- statistical physics of cortical dynamics;
- excitation/inhibition balance across states and scales;
- field potentials and the biophysics of neural measurements;
- sleep rhythms, spindles, K-complexes, waves, and state transitions;
- higher-order motifs and correlations in neural recordings;
- generative models of neural activity;
- data-driven dynamical modeling;
- physical computation and compositionality;
- biologically inspired AI and neural inductive biases.
The unifying question is still the same:
How do matter, dynamics, and organization give rise to computation?
N⁴ is my current name for that program. It is not meant to be a slogan. It is a way of holding together the different faces of the same object: the brain as a physical, dynamical, computational system, and biological computation as a guide for future artificial intelligence.
The arc
Looking back, the path has not been a shift from one field to another. It has been a progressive enlargement of the same problem.
Field potentials and sleep rhythms led to the physics of neural measurement.
The physics of neural measurement led to cortical dynamics and criticality.
Cortical dynamics led to excitation/inhibition, seizures, and state-dependent computation.
State-dependent computation led to information, causality, and excitable media.
Excitable media and biological regulation led to physical computation and compositionality.
Physical computation now leads back to neuroscience and forward to NeuroAI.
That is the arc of the work: from neural signals to physical computation; from cortical dynamics to biological organization; from the brain as an object of measurement to the brain as a model of computation.