Research

Neural Basis of Confidence in Decisions

How does the brain make decisions and assess confidence?

We pioneered a computational behavioral approach to isolate distinct cognitive processes and translate them across species. By establishing a rigorous framework linking human confidence reports to statistical decision confidence, we identified the first neural correlates of confidence in animals (Kepecs et al., 2008, Nature). Our studies revealed that orbitofrontal cortex neurons encode a modality-general confidence signal that predicts confidence-driven behaviors (Masset et al., 2020, Cell). We also discovered a novel confidence-guided learning mechanism conserved across species, underscoring its broad behavioral relevance (Lak et al., 2020, eLife).

In ongoing we are decoding cortical algorithms underlying confidence computation directly from spikes, recorded using high density electrophysiology, using advanced machine learning techniques. We are imaging orbitofrontal populations to identify fundamental cortical algorithms implemented by specific cell types to gain insights into how these cognitive operations are implemented.

Cortical Circuits and Interneurons

How do specific cell types contribute to cortical processing and behavior?

We identified cell-type-specific cortical principles that advance our understanding of cortical information processing. We found that VIP-expressing interneurons mediate disinhibitory control through a conserved circuit across cortical regions (Pi et al., 2014, Nature). Additionally, we discovered that VIP interneurons are activated by both reward and punishment, regulating cortical processing and plasticity (Szadai et al., 2020, eLife). These findings link the activity of specific neural types to behavioral decisions.

Currently we are probing the relationship between cortical interneurons and neuromodulators using miniscope imaging to observe large populations of neurons during behavior and to map to transcriptomics. We are exploring how specific cell types implement fundamental cortical algorithms.

Neuromodulatory Systems

How do neuromodulators like dopamine and acetylcholine influence behavior and mental health?

Neuromodulators can reconfigure circuit operations; we are focusing on the cholinergic basal forebrain and midbrain dopamine systems. Our research explores the computations of neuromodulatory systems, focusing on the cholinergic basal forebrain and midbrain dopamine systems. Recent studies have expanded into the modularity of the midbrain dopamine system and its coordination with the cholinergic system.

We investigated how elevated dopamine levels contribute to high-confidence false perceptions, offering insights into the neural mechanisms underlying psychotic disorders. This work demonstrates dopamine’s causal role in hallucination-like perception (Schmack et al., 2021, Science).

Currently we are dissecting the functions of striatal region-specific roles of midbrain dopamine and probing the dopamine in the context of psychosis and Parkinson’s disease.

Computational Psychiatry

Can we objectively diagnose psychiatric conditions and map them across species to improve treatments?

We developed objective measures for subjective behaviors like confidence and apathy, bridging human experiences with quantitative metrics. By creating computational-behavioral frameworks, we have modeled psychiatric phenomena across species, facilitating translational research. For instance, we operationalized apathy—a dimension of depression—as effort sensitivity, a measurable parameter across species. We are expanding our quantitative behavioral approach to develop objective diagnostics and animal models, addressing critical challenges in psychiatry.

Neuro-Immune Interactions

What role does the brain play in immunology and diseases like cancer and autoimmune disorders?

We initiated research exploring brain-body interactions, focusing on how the brain communicates with the immune system. We identified a circuit by which cancer-induced inflammation leads to fatigue and depressive symptoms, providing insights into chronic inflammation’s impact on behavior (Starosta et al., under review). Our studies mapped connections from sympathetic nerves to breast tumors, demonstrating how sympathetic activation increases tumor growth (Monje et al., 2020, Cell). We are also investigating rheumatoid arthritis and endometriosis to understand neuro-immune mechanisms affecting behavior and disease.

Neuro-Inspired Artificial Intelligence

Can insights from neural circuits bridge critical gaps in artificial intelligence?

We are developing neuro-AI approaches to integrate biologically-inspired confidence assessments into AI architectures, addressing limitations in current systems. By applying computational models to decode complex neural data and understand behavior, we aim to enhance AI systems with capabilities reflecting cognitive processes found in the brain.

Advancing Neuroscience with Technology

What technological innovations can expand our ability to probe neural circuits and brain-body physiology?

Technology is driving neurobiology into uncharted territories, revealing and altering brain circuits like never before. In collaboration with others and through our own efforts, we’re constantly exploring what new techniques can advance brain research.

Working with Michal Lipson’s group, we developed a nanophotonics probe, a reconfigurable optical interface that enables high-density neural stimulation in freely moving rodents (Mohanty et al., 2020). With Song Hu’s team, we are creating an optical fiber technique capable of stimulating over 1,000 sites with a single fiber. A persistent challenge in unbiased neural targeting is the tropism of viruses—how specific viruses preferentially infect certain cell types. To address this, we improved viral tracing methods for the CAV2 retrograde virus (Li et al., 2018), and we continue to explore new approaches to expand the frontiers of neurotechnology.