Because existing legal and technical frameworks regulate identifiable data but not the inferences multimodal systems generate, a governance gap arises where opacity, personalization, and clinical consequence meet. This session aims to clarify what multimodal systems can infer, how those inferences shape neuromodulation and computational psychiatry, and the ethical, clinical, and regulatory challenges they introduce. It will assess implications for consent, explanation, transparency, and legitimacy, and outline criteria for patient-aligned, trustworthy use.
- Kristin Kostick-Quenet (organiser, in person)
- Marcello Ienca (in-person)
- Georg Stark (in-person)
- Laura Cabrera (in-person)
- Phillipp Kellmeyer (in-person)
Agenda
- Introduction / 5 mins
- Dr. Kostick-Quenet on novel ethical concerns and regulatory considerations / 8 mins
- Dr. Ienca on how multimodal inference challenges existing norms / 8 mins
- Dr. Starke on the challenge of providing user-centric explanations / 8 mins
- Dr. Cabrera on how inferred states pose novel challenges to traditional informed consent processes / 8 mins
- Dr. Kellmeyer will focus on the role of foundation models in multimodal inference in neuroscience / 8 mins
- Moderated discussion / 10 mins
- Reflections / 5 mins
Attend
IMT School for Advanced Studies
Sagrestia Room
Thursday, April 16, 2026
9:00-10:00 CEST
Time check »
Neuroethics 2026
Building Bridges and Cultivating Community
This session is part of the Neuroethics 2026 conference organized by INS and SINe, and hosted by the IMT School for Advanced Studies. Register for the conference to attend in-person at IMT School for Advanced Studies in Lucca, Tuscany, Italy, or to participate virtually.
Description
Efforts to integrate multimodal data into neuromodulation, especially adaptive deep brain stimulation (DBS), are rapidly expanding. Neural, physiological, behavioral, linguistic, and ambient indicators can be fused to generate inferences about an individual’s brain and behavior, enabling more precise identification of pathological states and responsive neuromodulation strategies. Such multimodal inferences introduce new forms of knowledge about individuals that are neither raw data nor traditional biomarkers, but probabilistic constructs with ethical significance. While these systems offer clinical promise, they raise concerns that extend beyond familiar privacy debates: the resulting inferences are model-generated representations of states or traits that did not previously exist, were likely never explicitly consented to be created or acted upon, and can function as durable, sensitive labels that follow individuals across clinical settings, shaping stimulation parameters, therapeutic decisions, and downstream understandings of prognosis, responsibility, or identity.
Because existing legal and technical frameworks regulate identifiable data but not the inferences multimodal systems generate, a governance gap arises where opacity, personalization, and clinical consequence meet. This session aims to clarify what multimodal systems can infer, how those inferences shape neuromodulation and computational psychiatry, and the ethical, clinical, and regulatory challenges they introduce. It will assess implications for consent, explanation, transparency, and legitimacy, and outline criteria for patient-aligned, trustworthy use.
Together, these perspectives underscore the need to treat multimodal inference itself as an object of both scientific and ethical inquiry, examining which inputs yield which emergent representations, distinguishing mechanistically interpretable signals from opaque inferences, and documenting how inferences and their ethical impacts evolve as systems adapt. Transparent reporting, participatory design, and careful communication can help ensure that multimodal experimentation accelerates patient-aligned discovery and innovation without compromising trust. This panel integrates empirical, participatory, and conceptual perspectives to advance INS’s commitment to interdisciplinary neuroethics that bridge technical innovation with lived experience and clinical realities.
Recording
A recording of this session will be made available to registered attendees.