Check out our new paper, A large majority of awake hippocampal sharp-wave ripples feature spatial trajectories with momentum by Emma Krause, a previous graduate student in the lab, that just appeared online at Neuron.
Hippocampus contains cells known as place cells that are active in different parts of the environment, and thus are thought to collectively encode one’s location in space. During rest, these cells feature bursts of activity, known as sharp wave ripples (more precisely, that’s when the local electric field shows a particular activity pattern in a specific frequency band), during which these place cells are re-activated in a particular sequence. Previous work has shown that, for a fraction of sharp wave ripples, the place cell activity sequence simulates trajectories through the environment - events known as replay events. These events are thought to support planning, memory recall, and storage.
Particularly in open fields (rather than mazes with narrow corridors), only a small fraction of sharp wave ripples (~1/4 - 1/3) have been previously classified as replay events. It was unclear what the remaining sharp wave ripples were doing. In this work, we developed a novel approach based on Bayesian statistics to analyze the neural data underlying associated with sharp wave ripples (using the rodent dataset from Pfeiffer & Foster (2013, 2015)). First, this allowed us to show that not just a small fraction, but almost all sharp wave ripples constitute replay events. Second, we showed that the dynamics of the trajectories encoded in replay events featured momentum - that is, ballistic motion, just as the motion of an animal actually moving through the environment. The latter finding has consequences for the potential neural mechanisms underlying these events. Furthermore, it stands in contrast to a previous study that characterized the trajectory dynamics of replay events while animals were sleeping, and found that they lacked momentum, but instead followed Brownian motion (Stella et al., 2019).
Our study suggests that almost all, if not all, sharp wave ripples encode trajectories through space, such that they appear to have one single role rather than multiple roles. Hippocampal place cells have been found to not only encode spatial locations but also sequences of other abstract variables as long as they were relevant for the task. For example, in a task in which an animal heard a tone of slowly increasing frequency and needed to perform an action once the tone reached a particular, learned frequency, place cell activity appears to tile the space of task-relevant tone frequencies (Aronov, Nevers & Tank, 2017). In such instances, would replay events simulate these frequency sweeps always in a consistent order (akin to trajectories through abstract task space)? Furthermore, mimicking momentum of spatial trajectories, would the dynamics of these simulated sweeps mirror that of experienced sweeps? More generally, do these replays thus constitute a general simulation mechanism for variables that are relevant for the current task, including its dynamics? Knowing answers to these questions would help us narrow down the neural mechanisms underlying these replays, and understand their role for goal-directed behavior, memory, and planning.