Silencing hippocampal CA2 reduces behavioral flexibility in spatial learning.

The hippocampus is a key structure involved in learning and remembering spatial information. However, the extent to which hippocampal region CA2 is involved in these processes remains unclear. Here, we show that chronically silencing dorsal CA2 impairs reversal learning in the Morris water maze. After platform relocation, CA2-silenced mice spent more time in the vicinity of the old platform location and less time in the new target quadrant. Accordingly, behavioral strategy analysis revealed increased perseverance in navigating to the old location during the first day and an increased use of non-spatial strategies during the second day of reversal learning. Confirming previous indirect indications, these results demonstrate that CA2 is recruited when mice must flexibly adapt their behavior as task contingencies change. We discuss how these findings can be explained by recent theories of CA2 function and outline testable predictions to understand the underlying neural mechanisms. Demonstrating a direct involvement of CA2 in spatial learning, this work lends further support to the notion that CA2 plays a fundamental role in hippocampal information processing.

Brian2Loihi: An emulator for the neuromorphic chip Loihi using the spiking neural network simulator Brian.

Developing intelligent neuromorphic solutions remains a challenging endeavor. It requires a solid conceptual understanding of the hardware's fundamental building blocks. Beyond this, accessible and user-friendly prototyping is crucial to speed up the design pipeline. We developed an open source Loihi emulator based on the neural network simulator Brian that can easily be incorporated into existing simulation workflows. We demonstrate errorless Loihi emulation in software for a single neuron and for a recurrently connected spiking neural network. On-chip learning is also reviewed and implemented, with reasonable discrepancy due to stochastic rounding. This work provides a coherent presentation of Loihi's computational unit and introduces a new, easy-to-use Loihi prototyping package with the aim to help streamline conceptualization and deployment of new algorithms.

Neuromodulator-dependent synaptic tagging and capture retroactively controls neural coding in spiking neural networks.

Events that are important to an individual's life trigger neuromodulator release in brain areas responsible for cognitive and behavioral function. While it is well known that the presence of neuromodulators such as dopamine and norepinephrine is required for memory consolidation, the impact of neuromodulator concentration is, however, less understood. In a recurrent spiking neural network model featuring neuromodulator-dependent synaptic tagging and capture, we study how synaptic memory consolidation depends on the amount of neuromodulator present in the minutes to hours after learning. We find that the storage of rate-based and spike timing-based information is controlled by the level of neuromodulation. Specifically, we find better recall of temporal information for high levels of neuromodulation, while we find better recall of rate-coded spatial patterns for lower neuromodulation, mediated by the selection of different groups of synapses for consolidation. Hence, our results indicate that in minutes to hours after learning, the level of neuromodulation may alter the process of synaptic consolidation to ultimately control which type of information becomes consolidated in the recurrent neural network.

CA2 beyond social memory: Evidence for a fundamental role in hippocampal information processing.

Hippocampal region CA2 has received increased attention due to its importance in social recognition memory. While its specific function remains to be identified, there are indications that CA2 plays a major role in a variety of situations, widely extending beyond social memory. In this targeted review, we highlight lines of research which have begun to converge on a more fundamental role for CA2 in hippocampus-dependent memory processing. We discuss recent proposals that speak to the computations CA2 may perform within the hippocampal circuit.

A local circadian clock for memory?

The master clock, suprachiasmatic nucleus, is believed to control peripheral circadian oscillators throughout the brain and body. However, recent data suggest there is a circadian clock involved in learning and memory, potentially housed in the hippocampus, which is capable of acting independently of the master clock. Curiously, the hippocampal clock appears to be influenced by the master clock and by hippocampal dependent learning, while under certain conditions it may also revert to its endogenous circadian rhythm. Here we propose a mechanism by which the hippocampal clock could locally determine the nature of its entrainment. We introduce a novel theoretical framework, inspired by but extending beyond the hippocampal memory clock, which provides a new perspective on how circadian clocks throughout the brain coordinate their rhythms. Importantly, a local clock for memory would suggest that hippocampal-dependent learning at the same time every day should improve memory, opening up a range of possibilities for non-invasive therapies to alleviate the detrimental effects of circadian rhythm disruption on human health.

Robust Trajectory Generation for Robotic Control on the Neuromorphic Research Chip Loihi.

Neuromorphic hardware has several promising advantages compared to von Neumann architectures and is highly interesting for robot control. However, despite the high speed and energy efficiency of neuromorphic computing, algorithms utilizing this hardware in control scenarios are still rare. One problem is the transition from fast spiking activity on the hardware, which acts on a timescale of a few milliseconds, to a control-relevant timescale on the order of hundreds of milliseconds. Another problem is the execution of complex trajectories, which requires spiking activity to contain sufficient variability, while at the same time, for reliable performance, network dynamics must be adequately robust against noise. In this study we exploit a recently developed biologically-inspired spiking neural network model, the so-called anisotropic network. We identified and transferred the core principles of the anisotropic network to neuromorphic hardware using Intel's neuromorphic research chip Loihi and validated the system on trajectories from a motor-control task performed by a robot arm. We developed a network architecture including the anisotropic network and a pooling layer which allows fast spike read-out from the chip and performs an inherent regularization. With this, we show that the anisotropic network on Loihi reliably encodes sequential patterns of neural activity, each representing a robotic action, and that the patterns allow the generation of multidimensional trajectories on control-relevant timescales. Taken together, our study presents a new algorithm that allows the generation of complex robotic movements as a building block for robotic control using state of the art neuromorphic hardware.

Selective neuromodulation and mutual inhibition within the CA3-CA2 system can prioritize sequences for replay.

To make optimal use of previous experiences, important neural activity sequences must be prioritized during hippocampal replay. Integrating insights about the interplay between CA3 and CA2, we propose a conceptual framework that allows the two regions to control which sequences are reactivated. We suggest that neuromodulatory-gated plasticity and mutual inhibition enable discrete assembly sequences in both regions to support each other while suppressing competing sequences. This perspective provides a coherent interpretation for a variety of seemingly disconnected functional properties of CA2 and paves the way for a more general understanding of CA2.

Rats in a levered T-maze task show evidence of time-place discriminations in two different measures.

It is difficult for rats to learn to go to an arm of a T-maze to receive food that is dependent on the time of day, unless the amount of food in each daily session is different. In the same task, rats show evidence of time-place discriminations if they are required to press levers in the arms of the T-maze, but learning is only evident when the first lever press is considered, and not the first arm visited. These data suggest that rats struggle to use time as a discriminative stimulus unless the rewards/events differ in some dimension, or unless the goal locations can be visited prior to making a response. If both of these conditions are met in the same task, it might be possible to compare time-place learning in two different measures that essentially indicate performance before and after entering the arms of the T-maze. In the present study, we investigated time-place learning in rats with a levered T-maze task in which the amounts of food varied depending on the time of day. The first arm choices and first lever presses both indicated that the rats had acquired time-place discriminations, and both of these measures became significantly different from chance during the same block. However, there were subtle differences between the two measures, which suggest that time-place discrimination is aided by visiting the goal locations.

In a daily time-place learning task, time is only used as a discriminative stimulus if each daily session is associated with a distinct spatial location.

It is difficult for rats to acquire daily time-place (TP) learning tasks. One theory suggests that rats do not use time of day as a stimulus signaling a specific response. In the present study, we tested rats' ability to use time of day as a discriminative stimulus. A fixed-interval procedure was used in which one lever provided reinforcement on a FI-5-s schedule in morning sessions, and the same lever provided reinforcement on a FI-30-s schedule in afternoon sessions. Because only one place was used in this paradigm, the rats could only use time of day to acquire the task. Mean responses during the first 5 s of the first trial in each session indicated that the rats did not discriminate between the two sessions. In Phase II, a different lever location was used for each of the two daily sessions, which meant that both spatial and temporal information could be used to acquire the task. The rats readily acquired the task in this phase, and probe trials indicated that the rats were using a combination of spatial and temporal information to discriminate between the two different trial types. When the spatial cue was removed in Phase III, rats no longer discriminated the two sessions, suggesting that time can only be used as a discriminative stimulus when each daily session is associated with a distinct spatial location.