Among the deepest motivations in science are the beauty and logic of mathematics and physics, as well as questions that seem to reach beyond them. For example, are events ultimately predetermined or random, and where, in the physical world, including AI machinery, is there room for consciousness or free will? Such thought experiments can inspire one to study neuroscience, where mathematics, electromagnetism, electronics, and medicine converge in the effort to characterize, understand, and influence the human brain.
Particularly striking examples of this convergence are electroencephalography (EEG), magnetoencephalography (MEG) and transcranial magnetic stimulation (TMS). With EEG and MEG, one can observe brain activity by measuring electromagnetic fields outside the head. In contrast, TMS allows modulating brain activity with brief, strong magnetic-field pulses that induce electric currents in the brain, depolarizing neuronal membranes and triggering action potentials. This method is used in basic research, diagnostics, and therapy. However, conventional TMS therapy is still largely nonindividualized: pulses are typically delivered repeatedly to a single target, with limited adaptation to the patient’s brain state or response.
We have developed the first multi-locus TMS (mTMS) devices, collaborating closely also with the University of São Paulo. These instruments can change the location and orientation of the stimulating electric field electronically, in less than 1 ms, without moving the coils. This has required overcoming considerable engineering challenges: TMS systems operate with currents of thousands of amperes and instantaneous power in the megawatt range. Our current five-coil system enables electronic targeting and electric-field orientation control within a cortical region approximately 30 mm in diameter.The combination of simultaneous mTMS and EEG creates a bidirectional, non-invasive link to the brain. EEG can reveal the brain’s state and its immediate response to stimulation, while computer algorithms can use this information in real time to select the next stimulus. Our long-term goal is a closed-loop system that interacts with the brain somewhat like a therapist: listening, responding, and continuously adapting the approach. By bringing together physics, mathematics, neuroscience, and artificial intelligence, this approach may substantially improve the precision, individualization, and therapeutic efficacy of brain stimulation.