Correction: Data-driven analyses of motor impairments in animal models of neurological disorders.

[This corrects the article DOI: 10.1371/journal.pbio.3000516.].

Regional variation in cholinergic terminal activity determines the non-uniform occurrence of cortical slow waves during REM sleep in mice.

Sleep consists of two basic stages: non-rapid eye movement (NREM) and rapid eye movement (REM) sleep. NREM sleep is characterized by slow high-amplitude cortical electroencephalogram (EEG) signals, while REM sleep is characterized by desynchronized cortical rhythms. Despite this, recent electrophysiological studies have suggested the presence of slow waves (SWs) in local cortical areas during REM sleep. Electrophysiological techniques, however, have been unable to resolve the regional structure of these activities because of relatively sparse sampling. Here, we map functional gradients in cortical activity during REM sleep using mesoscale imaging in mice and show local SW patterns occurring mainly in somatomotor and auditory cortical regions with minimum presence within the default mode network. The role of the cholinergic system in local desynchronization during REM sleep is also explored by calcium imaging of cholinergic activity within the cortex and analyzing structural data. We demonstrate weaker cholinergic projections and terminal activity in regions exhibiting frequent SWs during REM sleep.

Sensory experience selectively reorganizes the late component of evoked responses.

In response to sensory stimulation, the cortex exhibits an early transient response followed by late and slower activation. Recent studies suggest that the early component represents features of the stimulus while the late component is associated with stimulus perception. Although very informative, these studies only focus on the amplitude of the evoked responses to study its relationship with sensory perception. In this work, we expand upon the study of how patterns of evoked and spontaneous activity are modified by experience at the mesoscale level using voltage and extracellular glutamate transient recordings over widespread regions of mouse dorsal neocortex. We find that repeated tactile or auditory stimulation selectively modifies the spatiotemporal patterns of cortical activity, mainly of the late evoked response in anesthetized mice injected with amphetamine and also in awake mice. This modification lasted up to 60 min and results in an increase in the amplitude of the late response after repeated stimulation and in an increase in the similarity between the spatiotemporal patterns of the late early evoked response. This similarity increase occurs only for the evoked responses of the sensory modality that received the repeated stimulation. Thus, this selective long-lasting spatiotemporal modification of the cortical activity patterns might provide evidence that evoked responses are a cortex-wide phenomenon. This work opens new questions about how perception-related cortical activity changes with sensory experience across the cortex.

Challenges of a small world analysis for the continuous monitoring of behavior in mice.

Documenting a mouse's "real world" behavior in the "small world" of a laboratory cage with continuous video recordings offers insights into phenotypical expression of mouse genotypes, development and aging, and neurological disease. Nevertheless, there are challenges in the design of a small world, the behavior selected for analysis, and the form of the analysis used. Here we offer insights into small world analyses by describing how acute behavioral procedures can guide continuous behavioral methodology. We show how algorithms can identify behavioral acts including walking and rearing, circadian patterns of action including sleep duration and waking activity, and the organization of patterns of movement into home base activity and excursions, and how they are altered with aging. We additionally describe how specific tests can be incorporated within a mouse's living arrangement. We emphasize how machine learning can condense and organize continuous activity that extends over extended periods of time.

Spatiotemporal structure of sensory-evoked and spontaneous activity revealed by mesoscale imaging in anesthetized and awake mice.

Stimuli-evoked and spontaneous brain activity propagates across the cortex in diverse spatiotemporal patterns. Despite extensive studies, the relationship between spontaneous and evoked activity is poorly understood. We investigate this relationship by comparing the amplitude, speed, direction, and complexity of propagation trajectories of spontaneous and evoked activity elicited with visual, auditory, and tactile stimuli using mesoscale wide-field imaging in mice. For both spontaneous and evoked activity, the speed and direction of propagation is modulated by the amplitude. However, spontaneous activity has a higher complexity of the propagation trajectories. For low stimulus strengths, evoked activity amplitude and speed is similar to that of spontaneous activity but becomes dissimilar at higher stimulus strengths. These findings are consistent with observations that primary sensory areas receive widespread inputs from other cortical regions, and during rest, the cortex tends to reactivate traces of complex multisensory experiences that might have occurred in exhibition of different behaviors.

Deep reinforcement learning to study spatial navigation, learning and memory in artificial and biological agents.

Despite the recent advancements and popularity of deep learning that has resulted from the advent of numerous industrial applications, artificial neural networks (ANNs) still lack crucial features from their biological counterparts that could improve their performance and their potential to advance our understanding of how the brain works. One avenue that has been proposed to change this is to strengthen the interaction between artificial intelligence (AI) research and neuroscience. Since their historical beginnings, ANNs and AI, in general, have developed in close alignment with both neuroscience and psychology. In addition to deep learning, reinforcement learning (RL) is another approach that is strongly linked to AI and neuroscience to understand how learning is implemented in the brain. In a recently published article, Botvinick et al. (Neuron, 107:603-616, 2020) explain why deep reinforcement learning (DRL) is important for neuroscience as a framework to study learning, representations and decision making. Here, I summarise Botvinick et al.'s main arguments and frame them in the context of the study of learning, memory and spatial navigation. I believe that applying this approach to study spatial navigation can provide useful insights for the understanding of how the brain builds, processes and stores representations of the outside world to extract knowledge.

The Neuroscience of Spatial Navigation and the Relationship to Artificial Intelligence.

Recent advances in artificial intelligence (AI) and neuroscience are impressive. In AI, this includes the development of computer programs that can beat a grandmaster at GO or outperform human radiologists at cancer detection. A great deal of these technological developments are directly related to progress in artificial neural networks-initially inspired by our knowledge about how the brain carries out computation. In parallel, neuroscience has also experienced significant advances in understanding the brain. For example, in the field of spatial navigation, knowledge about the mechanisms and brain regions involved in neural computations of cognitive maps-an internal representation of space-recently received the Nobel Prize in medicine. Much of the recent progress in neuroscience has partly been due to the development of technology used to record from very large populations of neurons in multiple regions of the brain with exquisite temporal and spatial resolution in behaving animals. With the advent of the vast quantities of data that these techniques allow us to collect there has been an increased interest in the intersection between AI and neuroscience, many of these intersections involve using AI as a novel tool to explore and analyze these large data sets. However, given the common initial motivation point-to understand the brain-these disciplines could be more strongly linked. Currently much of this potential synergy is not being realized. We propose that spatial navigation is an excellent area in which these two disciplines can converge to help advance what we know about the brain. In this review, we first summarize progress in the neuroscience of spatial navigation and reinforcement learning. We then turn our attention to discuss how spatial navigation has been modeled using descriptive, mechanistic, and normative approaches and the use of AI in such models. Next, we discuss how AI can advance neuroscience, how neuroscience can advance AI, and the limitations of these approaches. We finally conclude by highlighting promising lines of research in which spatial navigation can be the point of intersection between neuroscience and AI and how this can contribute to the advancement of the understanding of intelligent behavior.

Spatiotemporal patterns of neocortical activity around hippocampal sharp-wave ripples.

A prevalent model is that sharp-wave ripples (SWR) arise 'spontaneously' in CA3 and propagate recent memory traces outward to the neocortex to facilitate memory consolidation there. Using voltage and extracellular glutamate transient recording over widespread regions of mice dorsal neocortex in relation to CA1 multiunit activity (MUA) and SWR, we find that the largest SWR-related modulation occurs in retrosplenial cortex; however, contrary to the unidirectional hypothesis, neocortical activation exhibited a continuum of activation timings relative to SWRs, varying from leading to lagging. Thus, contrary to the model in which SWRs arise 'spontaneously' in the hippocampus, neocortical activation often precedes SWRs and may thus constitute a trigger event in which neocortical information seeds associative reactivation of hippocampal 'indices'. This timing continuum is consistent with a dynamics in which older, more consolidated memories may in fact initiate the hippocampal-neocortical dialog, whereas reactivation of newer memories may be initiated predominantly in the hippocampus.

Data-driven analyses of motor impairments in animal models of neurological disorders.

Behavior provides important insights into neuronal processes. For example, analysis of reaching movements can give a reliable indication of the degree of impairment in neurological disorders such as stroke, Parkinson disease, or Huntington disease. The analysis of such movement abnormalities is notoriously difficult and requires a trained evaluator. Here, we show that a deep neural network is able to score behavioral impairments with expert accuracy in rodent models of stroke. The same network was also trained to successfully score movements in a variety of other behavioral tasks. The neural network also uncovered novel movement alterations related to stroke, which had higher predictive power of stroke volume than the movement components defined by human experts. Moreover, when the regression network was trained only on categorical information (control = 0; stroke = 1), it generated predictions with intermediate values between 0 and 1 that matched the human expert scores of stroke severity. The network thus offers a new data-driven approach to automatically derive ratings of motor impairments. Altogether, this network can provide a reliable neurological assessment and can assist the design of behavioral indices to diagnose and monitor neurological disorders.

Low-cost solution for rodent home-cage behaviour monitoring.

In the current research on measuring complex behaviours/phenotyping in rodents, most of the experimental design requires the experimenter to remove the animal from its home-cage environment and place it in an unfamiliar apparatus (novel environment). This interaction may influence behaviour, general well-being, and the metabolism of the animal, affecting the phenotypic outcome even if the data collection method is automated. Most of the commercially available solutions for home-cage monitoring are expensive and usually lack the flexibility to be incorporated with existing home-cages. Here we present a low-cost solution for monitoring home-cage behaviour of rodents that can be easily incorporated to practically any available rodent home-cage. To demonstrate the use of our system, we reliably predict the sleep/wake state of mice in their home-cage using only video. We validate these results using hippocampal local field potential (LFP) and electromyography (EMG) data. Our approach provides a low-cost flexible methodology for high-throughput studies of sleep, circadian rhythm and rodent behaviour with minimal experimenter interference.

High-performance, inexpensive setup for simultaneous multisite recording of electrophysiological signals and mesoscale voltage imaging in the mouse cortex.

Simultaneous recording of optical and electrophysiological signals from multiple cortical areas may provide crucial information to expand our understanding of cortical function. However, the insertion of multiple electrodes into the brain may compromise optical imaging by both restricting the field of view and interfering with the approaches used to stabilize the specimen. Existing methods that combine electrophysiological recording and optical imaging in vivo implement either multiple surface electrodes, silicon probes, or a single electrode for deeper recordings. To address such limitation, we built a microelectrode array (hyperdrive, patent US5928143 A) compatible with wide-field imaging that allows insertion of up to 12 probes into a large brain area (8 mm diameter). The hyperdrive is comprised of a circle of individual microdrives where probes are positioned at an angle leaving a large brain area unobstructed for wide-field imaging. Multiple tetrodes and voltage-sensitive dye imaging were used for acute simultaneous registration of spontaneous and evoked cortical activity in anesthetized mice. The electrophysiological signals were used to extract local field potential (LFP) traces, multiunit, and single-unit spiking activity. To demonstrate our approach, we compared LFP and VSD signals over multiple regions of the cortex and analyzed the relationship between single-unit and global cortical population activities. The study of the interactions between cortical activity at local and global scales, such as the one presented in this work, can help to expand our knowledge of brain function.

Formation and reverberation of sequential neural activity patterns evoked by sensory stimulation are enhanced during cortical desynchronization.

Memory formation is hypothesized to involve the generation of event-specific neural activity patterns during learning and the subsequent spontaneous reactivation of these patterns. Here, we present evidence that these processes can also be observed in urethane-anesthetized rats and are enhanced by desynchronized brain state evoked by tail pinch, subcortical carbachol infusion, or systemic amphetamine administration. During desynchronization, we found that repeated tactile or auditory stimulation evoked unique sequential patterns of neural firing in somatosensory and auditory cortex and that these patterns then reoccurred during subsequent spontaneous activity, similar to what we have observed in awake animals. Furthermore, the formation of these patterns was blocked by an NMDA receptor antagonist, suggesting that the phenomenon depends on synaptic plasticity. These results suggest that anesthetized animals with a desynchronized brain state could serve as a convenient model for studying stimulus-induced plasticity to improve our understanding of memory formation and replay in the brain.