Abstract: Since the discovery of rapid eye movement (REM) sleep, the nature of the eye movements that characterize this sleep phase has remained elusive. Do they reveal gaze shifts in the virtual environment of dreams or simply reflect random brainstem activity? We harnessed the head direction (HD) system of the mouse thalamus, a neuronal population whose activity reports, in awake mice, their actual HD as they explore their environment and, in sleeping mice, their virtual HD. We discovered that the direction and amplitude of rapid eye movements during REM sleep reveal the direction and amplitude of the ongoing changes in virtual HD. Thus, rapid eye movements disclose gaze shifts in the virtual world of REM sleep, thereby providing a window into the cognitive processes of the sleeping brain.Such coordination between eye movements and virtual HD suggests the involvement of the deep layers of the superior colliculus (dSC), a midbrain motor command center that drives eye and head movement to shift gazes during awake navigation. Here, we show that the dSC issues motor commands during REM sleep, e.g., turn left, that are similar to those issued in the awake behaving animal. Strikingly, these motor commands, despite not being executed, shift the internal representation of HD as if the animal had turned.  Thus, during REM sleep, the brain simulates actions by issuing motor commands that, while not executed, have consequences as if they had been. These studies suggest that the sleeping brain, while disengaged from the external world, uses its internal model of the world to simulate interactions with it.

Senzai Lab

Abstract: Computational processes in neural systems emerge through learning across multiple timescales; from evolution and development to immediate, in-context adaptation. Yet fundamental questions remain: Which neural architectures confer evolutionary advantages? How do experiences shape circuit dynamics? What principles govern how specific computations arise during training? My group addresses these questions using simulated recurrent neural networks. Building on a decade of research across multiple labs, we focus on fixed point structures, termed "dynamical motifs”, that serve as computational primitives. We've discovered that these motifs can be flexibly composed to solve diverse tasks, with rapid learning often involving novel recombination of existing motifs rather than construction of entirely new dynamics. However, the principles governing motif composition remain poorly understood, motivating our simulation-based approach.I will present two ongoing projects that illustrate this framework:

1. Dynamical motifs underlying foraging behavior: How fundamental dynamical motifs support naturalistic decision-making and navigation.
2. How task structure shapes computational dynamics: The relationship between problem structure and the organization of dynamical systems that solve it.

Laura Driscoll profile

This visit and seminar will be scheduled for a later date to be announced.

Fairhall lab website

Abstract:  From serving a volleyball to playing the piano - one of our brain’s most remarkable feats is the ability to learn a sheer endless number of motor skills. Despite their importance, how our brain learns and generates such skills is poorly understood. While many nodes of the brain’s distributed motor network have been identified, their functions and interactions remain often unclear. We probe this network through the lens of complex, highly stereotyped and spatiotemporally precise movement patterns trained in rats. We have found that the basal ganglia play critical, unexpected roles in both skill learning and execution, and regulate the transition from variable to stereotyped movement patterns throughout learning. Furthermore, we are exploring how the brain solves the challenge to form, store and recall the memories for our countless skills, using the same neural substrates. Together, our results shine new light on the mechanisms and circuits underlying our motor skills.

Wolff Lab

Abstract

Around 380 million years ago, as vertebrates ventured onto land, vision changed dramatically. Air is far more transparent than water, and at the water-to-land transition we also see a marked increase in eye size. Together these factors yielded an enormous extension of visual range, resulting in a million-fold expansion in the volume of space that could be visually monitored. With objects detectable much farther away, animals suddenly had more time to act.

I argue that this elongated sensory horizon shifted the advantage from fast, reflexive responses---effective underwater when threats emerge at about a body length---to multi-step action sequences that are planned ahead. In particular, partially cluttered terrestrial settings (e.g., savanna-like mixes of open zones and cover) create many viable future paths, some of which will avoid mortal threat while others will lead to death. In such environments, selecting among imagined futures---planning---should pay off.

I will first present motivating simulation results showing that planning yields large benefits specifically in mid-clutter terrestrial regimes, whereas in very simple or very cluttered spaces habit-based action control is equally or more effective. To test these predictions behaviorally, we've built a robot–rodent interaction arena with reconfigurable obstacles that let us dial spatial complexity up or down. An autonomous robot acts as a mobile threat while mice navigate toward safety. I will share initial behavioral findings from this paradigm, including path diversification and pauses that appear to support look-ahead, as well as preliminary hippocampal recordings acquired during behavior. We also show some initial work comparing state-of-the-art reinforcement learning algorithms to animal behavior with interesting implications for improving AI. Together, these results outline a tractable experimental program for linking expanded terrestrial sensory horizons to the emergence of planning---and, potentially, to key components of mind.

Malcolm A. MacIver profile

Abstract: “A central goal shared by neuroscience and robotics is to understand how systems can navigate and act autonomously in complex environments. Although extensive research has revealed how the visual system segments natural scenes into distinct components—insights that have inspired advances in computer vision and robotics—the next crucial challenge remains: learning the properties of these objects and responding appropriately. In this talk, I will present our work using the fruit fly Drosophila melanogaster to investigate how the brain learns about objects and other animals in its environment, and how it uses that information to guide behavior. By integrating quantitative behavioral analysis, genetic manipulation, connectomics, and neural recordings, we aim to uncover the neural mechanisms that enable flexible, adaptive interactions with the world.”

EPFL Ramdya Lab