Research
We record from many neurons across the depth of the cortex while primates perform demanding visual and social tasks, and use those data to find general principles of how information flows through cortical circuits.
Neural mechanisms of active visual perception
Vision is an active process. We select behaviorally relevant stimuli for detailed inspection while ignoring distractors. Two mechanisms drive this selection: spatial attention, which concentrates limited perceptual resources on regions of interest, and saccades, the rapid eye movements we make two to three times every second to bring objects onto high-acuity central vision.
Visual information is processed by a hierarchy of brain areas that together make up about 30% of the neocortex. Neurons in these areas encode attributes such as shape, color and motion, and they sit within a layered (laminar) architecture. The patterns of connectivity between layers are stereotyped and shared across senses, which suggests a canonical cortical circuit. Attention and eye movements continuously modulate information processing in these circuits.
By studying the dynamics of laminar circuits in animals engaged in challenging visual tasks, we aim to uncover fundamental principles of information flow through the cortex. This work can help explain conditions in which that flow is disrupted, such as ADHD and schizophrenia. We use high-density laminar electrophysiology, advanced data analysis and computational modeling.
Questions we are asking
- What are the principles of hierarchical processing in the sensory neocortex?
- How does spatial attention modulate information flow in the laminar cortical circuit?
- How do eye movements modulate laminar cortical dynamics?
- How does the presence of contextual stimuli change visual information processing?
Neural mechanisms of social cognition
Primates, including humans, live in complex social groups and interact richly with one another. Existing approaches to studying social cognition are often limited in ecological validity or fragmented in scope.
To address this, we have developed methods for accurate 3D tracking of freely moving animals as they interact in controlled social environments. Combined with modern computational and wireless neural recording techniques, these methods let us quantify the behavioral dynamics of social networks and the neural dynamics that support social behavior.
This is a collaboration with the Chang and Jadi labs at Yale.