RESEARCH
Our vision
Animals display a remarkable array of complex adaptive behaviors. They are able to learn from experience and, based on it, make new inferences and flexibly guide decisions in response to changing environmental demands. These behaviors are supported by the finely tuned dynamics of neuronal ensembles distributed across brain circuits. We seek to understand the algorithmic and mechanistic underpinnings of these complex behaviors at the computational, circuit, and cellular levels.
Because tackling these questions requires multiple complementary approaches and novel methods, we develop and implement sophisticated electrophysiological, optogenetic, imaging and computational techniques. Drawing from both computational theories of learning and ethology, we implement novel spatial and social naturalistic behavioral paradigms with a variety of rodent species. We use large-scale electrophysiology and imaging to record hundreds to thousands of neurons across multiple brain areas during behavior. To test the role of specific cells types and brain patterns we perform closed-loop optogenetic manipulations of neural activity, allowing refinement and updating of the theories guiding experiment. In doing so, our experiments inform new theories and models of neural mechanisms of behavior, allowing a continual dialogue between experiment and theory.
Research topics
1- Learning & memory guided decision making
One of the main goals of our lab is to better understand how previously acquired memories guide future decisions. To learn from individual experiences, animals must be able to extract common principles and apply them to new instances, that is, to generalize. This generalization process allows animals to perform inferences and make predictions about possible outcomes of their actions. We hypothesize that such predictive inferences rely on internal models of the world (or ‘cognitive maps’), which are supported by the same neural circuits and mechanisms as spatial navigation.
A brain region necessary for memory and spatial navigation is the hippocampus. During learning, memory traces are initially encoded in the hippocampus and subsequently consolidated during sleep. A key physiological mechanism underlying this process is the synchronous reactivation of hippocampal cells that participated in a recent experience. Hippocampal reactivations are coordinated by network events known as sharp-wave ripples (SWR). These reactivations occur in the form of sequences that recapitulate experience in a temporally compressed manner (“replay”). We (publications) and others have shown that SWRs also occur during behavior and contribute to the planning of future actions. SWRs broadcast hippocampal memory representations across the brain. Coordination between the hippocampus and associated cortical areas is necessary for memory-guided behavior, and its impairment in neural disorders leads to cognitive deficits. We want to understand how the rest of the brain reads-out the hippocampal memory code and uses it to guide behavior
2- Flexible social behaviors
Social behaviors are a key aspect of many animal species, including rodents and humans. In species that live in groups, the ability to recognize a conspecific and remember previous interactions is an essential part of adaptive behavior. Animals navigate their “social space” by learning hierarchical structures, keeping track of interactions with different emotional valence, establishing and defending territories for survival, and more. To understand the cellular mechanisms of social behaviors it is crucial to perform well-controlled physiological studies in naturalistic conditions. A good animal model for tackling these questions are rodents, since they are social animals and amenable to behavioral assays and neural circuit interrogations.
We focus on the hippocampus, where we found cells that encode conspecific identity and form social memory traces that reactivate during SWRs. This area receives strong neuromodulatory inputs from subcortical structures. Neuromodulators, such as serotonin and acetylcholine, regulate both social behaviors and memory processes. Their fluctuations correlate with brain state transitions and enhance the emotional relevance of salient information, potentially through the induction of synaptic plasticity. We investigate how hippocampal-subcortical interactions support social cognition in ecologically relevant conditions.
3- Neural dynamics & brain computation
We seek to identify fundamental principles of brain computation that support complex behaviors. The long-term aspiration is to understand the neuronal mechanisms of intelligent behavior. We believe that one of the keys to understanding these phenomena relies on the dynamic coordination of distributed cell ensembles that can flexibly support changing behavioral demands.
We can now record hundreds of individual neurons simultaneously across brain areas in behaving animals. However, making sense of this wealth of data is one of the biggest bottlenecks in modern neuroscience. We work on two fronts. We develop and apply new analysis methods to large-scale neural datasets borrowing tools from machine-learning, information theory and dynamical systems. We also build computational models to understand how brain circuits process information and develop mathematical frameworks to explain how brain computations support flexible behaviors.
We believe that the study of the biological basis of intelligence can strongly benefit from recent advances in artificial intelligence, and also the other way around. Discoveries of neural algorithms can inform the development of AI systems. We thus keep a keen interest in both fields.
4- Closed-loop manipulations of neural dynamics in brain disease
In most neuropsychiatric disorders, traditional therapies involve irreversible surgical or pharmacological interventions that cannot be rapidly adjusted to changes in symptoms. Alternatively, closed-loop methods synchronized to neural activity provide intervention only when needed, by detecting and correcting abnormal neurological patterns. We are developing applications of closed-loop manipulations of specific neural patterns as a method to ameliorate cognitive deficits in brain disorders.
Several neuropsychiatric disorders characterized by memory deficits, such as Alzheimer’s disease, schizophrenia or epilepsy, have been related to abnormal sharp-wave ripples and disrupted hippocampo–cortical coordination in humans and mouse disease models. We showed that closed-loop optogenetic ripple prolongation is effective in improving memory in wild-type rats and mice. Now we are applying similar types of manipulations to different genetic models of neural disease. We also aim to understand the circuit and brain dynamics alterations that lead to the cognitive and behavioral symptoms in these diseases.
Techniques
We employ a multi-disciplinary approach, including the development and application of cutting-edge experimental and computational techniques, to understand the algorithms and mechanisms that support flexible behavior.
Large-scale electrophysiology
We employ large-scale silicon probes and Neuropixels to record hundreds of single neurons and brain oscillations across multiple brain areas in behaving rodents.
Optogenetics
We use several optogenetic tools to manipulate the activity of genetically defined cell types in order to dissect their contribution to specific behaviors or brain dynamics.
Imaging
Using both multi-photon microscopes or single-photon mini endoscopes, we image the activity of cell ensembles and axons. We also employ fiber-photometry to measure the fluctuations of neuromodulators during behaviors and sleep.
Behavior
Traditional behavioral tasks have been restricted to a few highly simplified paradigms. We develop novel behavioral tasks for complex spatial learning, inference and naturalistic social behaviors.
Data analysis and theory
Our experiments are guided by theory and computational models. We are also constantly developing and implementing novel analysis methods for large-scale neural datasets and the measure and control of behaviors.