How the Hippocampus Builds Predictive Spatial Maps
Post by Giulia Baracchini
What’s the science?
Place cells, a type of neuron in the brain’s hippocampus, are involved in recreating spatial maps of the external world and thus play a key role in spatial learning and memory. Spatial learning requires the reactivation of place cells after spatial encoding has occurred, a phenomenon called place cell replay. However, place cells’ spatial representations are not static. Rodent studies have shown how place cells’ spatial representations are formed as animals learn and adapt to their changing environments. How place cell replay processes dynamically evolve during spatial learning remains unexplored. This week, in PNAS, Igata and colleagues tested how place cell replays change as rats learn a spatial task.
How did they do it?
Rats were trained to run from a starting area to a checkpoint where they were given an intermediate reward, and then to a goal area where they received a final reward (pre-learning phase). After a few trials, the authors changed the location of the intermediate checkpoint and reward, requiring the rats to update their navigation strategies in order to obtain the reward (replacement phase). The animals eventually learned to run along the new path where they would receive a reward (post-learning phase).
While the rats were performing the task, the authors recorded the spiking activity and theta-sequences (a measure of the spatial organization of the place cells) of multiple place cells located in the dorsal hippocampal CA1 region. They quantified the presence, frequency, directionality and sequence strength of sharp-wave ripple (SWR)-associated synchronous spikes, which are bursts of activity during which hippocampal place cells are activated. These reactivations of hippocampal place cells are referred to as place cell replays. They then used a statistical model (Bayesian decoding) to estimate how the rats’ spatial behaviour was represented by place cell replays. Lastly, to provide evidence for a causal link between learning-related replays, and animal behaviour during spatial learning, the authors transiently suppressed SWR-associated synchronous spikes after the replacement phase. They did so by selectively stimulating the ventral hippocampal commissure and delivering closed-loop feedback electrical stimulation.
What did they find?
The authors found that throughout the different learning stages, rats built a spatial representation of the environment by primarily recruiting (i) stable (i.e., similar across stages) sets of hippocampal place cells along with (ii) sets of place cells showing context-dependent properties (i.e., encoding specific locations). By feeding information about these cells’ preferred locations into their Bayesian model, the authors could successfully reconstruct an animal’s position in space. These findings demonstrate the role of the hippocampus in creating abstract, generalized memory maps.
As the rats were updating their navigational strategies, the authors found a significant increase in hippocampal theta-sequences and sequential SWR-associated synchronous spikes, compared to other learning phases. Interestingly, sequential place cell replays occurred for prioritized experiences only, in other words only salient and reward-related locations were replayed by separate synchronous events. Such preferential replay events were found to be greater for newly rewarded locations. While the rats learned about the new intermediate checkpoint area, most of these replay events represented the new path in the later phase of learning, even before the animals started taking the new path. The authors also found that the content and the directionality of the place cell replays changed as a function of learning over time. Together, these findings highlight the role of the hippocampus in building predictive maps of the environment that dynamically evolve as learning takes place. Finally, the authors reported that suppressing SWR-associated synchronized events impaired learning, suggesting that place cell replays are causally involved in the stabilization of newly learned behaviours.
What’s the impact?
This study highlights the key role of hippocampal place cell replays in building predictive, dynamic maps of the external environment. Importantly, the hippocampus replays salient or prioritized experiences to effectively encode them into memory. Further, such maps change as a function of learning and predict future behaviour.
Igata et al. Prioritized experience replays on a hippocampal predictive map for learning. Proceedings of the National Academy of Sciences (2020). Access the original scientific publication here.