The ultimate goal of my lab is to understand the relationship between neuronal activity dynamics and the states of cognitive experience, like memories and thoughts. We focus on the hippocampal/entorhinal system, which is critically involved in the formation and retrieval of memories. Studies of this circuit have traditionally focused on navigation, uncovering a rich representation of spatial variables in the firing of its neurons. Yet, spatial location is only one component of memory. Beyond location, memories contain a mix of sensory, temporal, behavioral and emotional information. We are trying to understand how all these diverse types of information are encoded in the hippocampal/entorhinal system and bound together into a coherent representation.
Rat virtual reality. In one line of research, we are developing and using rat virtual reality (VR) systems in order to obtain unprecedented experimental control of the animal’s sensory environment. VR allows rapid, closed-loop manipulations of the animal’s sensory and behavioral context, coupled with uninterrupted neural recordings and manipulations. We are beginning to understand how behaviorally relevant sensory information is organized in the activity of the hippocampal/entorhinal circuit. By training rats to perform sophisticated memory-guided tasks in VR, we also aim to understand how the states of activity in this circuit contribute to behaviors that require memory retrieval.
Food caching by birds. In a parallel line of research, we will study how memories are processed during a fascinating and highly specialized behavior – the caching and subsequent retrieval of hidden food items by black-capped chickadees. Chickadees use their hippocampal formation to perform this memory-guided task. However, compared to its mammalian counterpart, the avian hippocampal system is anatomically simpler and more tractable on the circuit level. Using automated laboratory caching arenas inspired by VR, combined with miniaturized technologies for small birds, we plan to investigate neural representations of cache memories and the contributions of these representations to behavior.
There are profound differences in behavior between the sexes, among individuals, and across species. Much effort has been devoted to the observation and theory of behavioral variation, but the molecular, genetic, and neurobiological mechanisms that generate and maintain such diversity are largely unknown. My lab studies the mechanisms of behavioral variation from genetic and neurobiological angles – by identifying specific genes involved and how they impact the brain, and by characterizing functional variation in neuronal circuits. We then characterize the common themes that emerge to describe the evolution of behavior.
Work in my lab focuses primarily on deer mice (genus Peromyscus), an excellent system to study natural variation in behavior because (1) there is a large diversity of behavior within and among species and many of these species are interfertile, permitting forward genetic analyses; (2) the genome of multiple Peromyscus species has been sequenced; (3) Peromyscus is diverged from laboratory mice and rats, providing an opportunity to discover biological features that differ from traditional model species; nonetheless, many tools developed for laboratory mice and rats also work in Peromyscus; and (4) Peromyscus mice can breed in the laboratory in the same conditions as laboratory mice, allowing us to perform controlled experiments.
In addition to our primary work in Peromyscus mice, we harness behavioral variation in other vertebrates to uncover the molecular, genetic and neurobiological mechanisms that drive behavioral evolution.
A challenge in understanding the brain lies in going beyond descriptions of single brain areas to studying how multiple areas interact within a network to produce emergent behaviors. In the cerebral cortex, these network interactions are thought to be mediated by bidirectional connections between cortical areas. However, despite anatomical evidence of strongly recurrent network connectivity in cortex, the functional role of recurrent cortical processing is not well understood. The goal of the lab is to elucidate the computational purpose of the recurrent network architecture of cortex.
Using vision as a model system, the lab studies how recurrent processing supports our ability to visually recognize objects, a computationally challenging task at which humans excel. Bidirectional connections between the hierarchically arranged stages of visual cortex could serve at least two fundamental computational purposes: they may dynamically modify neural activity in a process of online visual inference (Bayesian inference) between cortical stages, and they may provide top-down signals for driving learning (error backpropagation) across the cortical hierarchy. To explore these hypotheses, experiments measure the content and downstream impact of messages passed across the network during object recognition and learning.
Our experimental platform, centered around the common marmoset, will use advanced tools such as cellular imaging and targeted optogenetics to examine the neural subpopulations involved in transmission of information between high-level visual cortical areas. Positioned at the intersection of biological vision, neuroengineering, and machine learning, the lab’s environment fosters an interplay between experiments, novel techniques, and neural network models in an effort to reveal the computations implemented in cortical networks.
Understanding brain-computational mechanisms by testing deep neural network models with massively multivariate brain-activity data
Recent advances in neural network modelling have enabled major strides in computer vision and other artificial intelligence applications. Artificial neural networks are inspired by the brain and their computations could be implemented in biological neurons. Although designed with engineering goals, this technology provides the basis for tomorrow’s computational neuroscience, engaging complex cognitive tasks and high-level cortical representations. We are entering an exciting new era, in which we will be able to build neurobiologically faithful feedforward and recurrent computational models of how biological brains perform high-level feats of intelligence.
The objective of the lab is to understand the brain information processing that enables visual perception, object recognition, and scene understanding. Vision is of interest in its own right, but also provides a model for understanding, more generally, how the brain computes and how it might perform probabilistic inference through parallel and recurrent computations.
The lab uses massively multivariate measurements of brain activity along with behavioural data to test models of brain information processing that perform visual tasks. To explain visual processing, the models must meet computational challenges comparable to those biological visual systems face in the real world. The models therefore need to contain rich visual knowledge about the world and have substantial computational power. Building such models requires the methods of machine learning and artificial intelligence. We take a top-down approach to modelling, starting with models that perform the task, but abstract from much of the biological detail. We then attempt to reveal the aspects of human task performance and brain activity that these models fail to explain. This motivates adjustments to the architecture and the design of the units. Architectures and units must be plausibly implementable with biological neurons. Their design is chosen as required by function and inspired by biology, so as to better explain brain and behavioural data. The lab develops neural net models, statistical inference and visualisation techniques, and visual stimuli and tasks, and measures brain activity with fMRI and MEG in humans and with array recordings in nonhuman primates.