Rapid, systematic updating of movement by accumulated decision evidence.

Acting in the natural world requires not only deciding among multiple options but also converting decisions into motor commands. How the dynamics of decision formation influence the fine kinematics of response movement remains, however, poorly understood. Here we investigate how the accumulation of decision evidence shapes the response orienting trajectories in a task where freely-moving rats combine prior expectations and auditory information to select between two possible options. Response trajectories and their motor vigor are initially determined by the prior. Rats movements then incorporate sensory information as early as 60 ms after stimulus onset by accelerating or slowing depending on how much the stimulus supports their initial choice. When the stimulus evidence is in strong contradiction, rats change their mind and reverse their initial trajectory. Human subjects performing an equivalent task display a remarkably similar behavior. We encapsulate these results in a computational model that, by mapping the decision variable onto the movement kinematics at discrete time points, captures subjects' choices, trajectories and changes of mind. Our results show that motor responses are not ballistic. Instead, they are systematically and rapidly updated, as they smoothly unfold over time, by the parallel dynamics of the underlying decision process.

Recurrent networks endowed with structural priors explain suboptimal animal behavior.

The strategies found by animals facing a new task are determined both by individual experience and by structural priors evolved to leverage the statistics of natural environments. Rats quickly learn to capitalize on the trial sequence correlations of two-alternative forced choice (2AFC) tasks after correct trials but consistently deviate from optimal behavior after error trials. To understand this outcome-dependent gating, we first show that recurrent neural networks (RNNs) trained in the same 2AFC task outperform rats as they can readily learn to use across-trial information both after correct and error trials. We hypothesize that, although RNNs can optimize their behavior in the 2AFC task without any a priori restrictions, rats' strategy is constrained by a structural prior adapted to a natural environment in which rewarded and non-rewarded actions provide largely asymmetric information. When pre-training RNNs in a more ecological task with more than two possible choices, networks develop a strategy by which they gate off the across-trial evidence after errors, mimicking rats' behavior. Population analyses show that the pre-trained networks form an accurate representation of the sequence statistics independently of the outcome in the previous trial. After error trials, gating is implemented by a change in the network dynamics that temporarily decouple the categorization of the stimulus from the across-trial accumulated evidence. Our results suggest that the rats' suboptimal behavior reflects the influence of a structural prior that reacts to errors by isolating the network decision dynamics from the context, ultimately constraining the performance in a 2AFC laboratory task.

Towards the next generation of recurrent network models for cognitive neuroscience.

Recurrent neural networks (RNNs) trained with machine learning techniques on cognitive tasks have become a widely accepted tool for neuroscientists. In this short opinion piece, we discuss fundamental challenges faced by the early work of this approach and recent steps to overcome such challenges and build next-generation RNN models for cognition. We propose several essential questions that practitioners of this approach should address to continue to build future generations of RNN models.

Temporal Sharpening of Sensory Responses by Layer V in the Mouse Primary Somatosensory Cortex.

The timing of stimulus-evoked spikes encodes information about sensory stimuli. Here we studied the neural circuits controlling this process in the mouse primary somatosensory cortex. We found that brief optogenetic activation of layer V pyramidal cells just after whisker deflection modulated the membrane potential of neurons and interrupted their long-latency whisker responses, increasing their accuracy in encoding whisker deflection time. In contrast, optogenetic inhibition of layer V during either passive whisker deflection or active whisking decreased accuracy in encoding stimulus or touch time, respectively. Suppression of layer V pyramidal cells increased reaction times in a texture discrimination task. Moreover, two-color optogenetic experiments revealed that cortical inhibition was efficiently recruited by layer V stimulation and that it mainly involved activation of parvalbumin-positive rather than somatostatin-positive interneurons. Layer V thus performs behaviorally relevant temporal sharpening of sensory responses through circuit-specific recruitment of cortical inhibition.

High-Accuracy Detection of Neuronal Ensemble Activity in Two-Photon Functional Microscopy Using Smart Line Scanning.

Two-photon functional imaging using genetically encoded calcium indicators (GECIs) is one prominent tool to map neural activity. Under optimized experimental conditions, GECIs detect single action potentials in individual cells with high accuracy. However, using current approaches, these optimized conditions are never met when imaging large ensembles of neurons. Here, we developed a method that substantially increases the signal-to-noise ratio (SNR) of population imaging of GECIs by using galvanometric mirrors and fast smart line scan (SLS) trajectories. We validated our approach in anesthetized and awake mice on deep and dense GCaMP6 staining in the mouse barrel cortex during spontaneous and sensory-evoked activity. Compared to raster population imaging, SLS led to increased SNR, higher probability of detecting calcium events, and more precise identification of functional neuronal ensembles. SLS provides a cheap and easily implementable tool for high-accuracy population imaging of neural GCaMP6 signals by using galvanometric-based two-photon microscopes.

Synchronization in the prefrontal-striatal circuit tracks behavioural choice in a go-no-go task in rats.

Rodent striatum is involved in sensory-motor transformations and reward-related learning. Lesion studies suggest dorsolateral striatum, dorsomedial striatum and nucleus accumbens underlie stimulus-response transformations, goal-directed behaviour and reward expectation, respectively. In addition, prefrontal inputs likely control these functions. Here, we set out to study how reward-driven behaviour is mediated by the coordinated activity of these structures in the intact brain. We implemented a discrimination task requiring rats to either respond or suppress responding on a lever after the presentation of auditory cues in order to obtain rewards. Single unit activity in the striatal subregions and pre-limbic cortex was recorded using tetrode arrays. Striatal units showed strong onset responses to auditory cues paired with an opportunity to obtain reward. Cue-onset responses in both striatum and cortex were significantly modulated by previous errors suggesting a role of these structures in maintaining appropriate motivation or action selection during ongoing behaviour. Furthermore, failure to respond to the reward-paired tones was associated with higher pre-trial coherence among striatal subregions and between cortex and striatum suggesting a task-negative corticostriatal network whose activity may be suppressed to enable processing of reward-predictive cues. Our findings highlight that coordinated activity in a distributed network including both pre-limbic cortex and multiple striatal regions underlies reward-related decisions.

An inhibitory gate for state transition in cortex.

Large scale transitions between active (up) and silent (down) states during quiet wakefulness or NREM sleep regulate fundamental cortical functions and are known to involve both excitatory and inhibitory cells. However, if and how inhibition regulates these activity transitions is unclear. Using fluorescence-targeted electrophysiological recording and cell-specific optogenetic manipulation in both anesthetized and non-anesthetized mice, we found that two major classes of interneurons, the parvalbumin and the somatostatin positive cells, tightly control both up-to-down and down-to-up state transitions. Inhibitory regulation of state transition was observed under both natural and optogenetically-evoked conditions. Moreover, perturbative optogenetic experiments revealed that the inhibitory control of state transition was interneuron-type specific. Finally, local manipulation of small ensembles of interneurons affected cortical populations millimetres away from the modulated region. Together, these results demonstrate that inhibition potently gates transitions between cortical activity states, and reveal the cellular mechanisms by which local inhibitory microcircuits regulate state transitions at the mesoscale.

Interspersed Distribution of Selectivity to Kinematic Stimulus Features in Supragranular Layers of Mouse Barrel Cortex.

Neurons in the primary sensory regions of neocortex have heterogeneous response properties. The spatial arrangement of neurons with particular response properties is a key aspect of population representations and can shed light on how local circuits are wired. Here, we investigated how neurons with sensitivity to different kinematic features of whisker stimuli are distributed across local circuits in supragranular layers of the barrel cortex. Using 2-photon calcium population imaging in anesthetized mice, we found that nearby neurons represent diverse kinematic features, providing a rich population representation at the local scale. Neurons interspersed in space therefore responded differently to a common stimulus kinematic feature. Conversely, neurons with similar feature selectivity were located no closer to each other than predicted by a random distribution null hypothesis. This finding relied on defining a null hypothesis that was specific for testing the spatial distribution of tuning across neurons. We also measured how neurons sensitive to specific features were distributed relative to barrel boundaries, and found no systematic organization. Our results are compatible with randomly distributed selectivity to kinematic features, with no systematic ordering superimposed upon the whisker map.

Variable Temporal Integration of Stimulus Patterns in the Mouse Barrel Cortex.

Making sense of the world requires distinguishing temporal patterns and sequences lasting hundreds of milliseconds or more. How cortical circuits integrate over time to represent specific sensory sequences remains elusive. Here we assessed whether neurons in the barrel cortex (BC) integrate information about temporal patterns of whisker movements. We performed cell-attached recordings in anesthetized mice while delivering whisker deflections at variable intervals and compared the information carried by neurons about the latest interstimulus interval (reflecting sensitivity to instantaneous frequency) and earlier intervals (reflecting integration over timescales up to several hundred milliseconds). Neurons carried more information about the latest interval than earlier ones. The amount of temporal integration varied with neuronal responsiveness and with the cortical depth of the recording site, that is, with laminar location. A subset of neurons in the upper layers displayed the strongest integration. Highly responsive neurons in the deeper layers encoded the latest interval but integrated particularly weakly. Under these conditions, BC neurons act primarily as encoders of current stimulation parameters; however, our results suggest that temporal integration over hundreds of milliseconds can emerge in some neurons within BC.

Statistical wiring of thalamic receptive fields optimizes spatial sampling of the retinal image.

It is widely assumed that mosaics of retinal ganglion cells establish the optimal representation of visual space. However, relay cells in the visual thalamus often receive convergent input from several retinal afferents and, in cat, outnumber ganglion cells. To explore how the thalamus transforms the retinal image, we built a model of the retinothalamic circuit using experimental data and simple wiring rules. The model shows how the thalamus might form a resampled map of visual space with the potential to facilitate detection of stimulus position in the presence of sensor noise. Bayesian decoding conducted with the model provides support for this scenario. Despite its benefits, however, resampling introduces image blur, thus impairing edge perception. Whole-cell recordings obtained in vivo suggest that this problem is mitigated by arrangements of excitation and inhibition within the receptive field that effectively boost contrast borders, much like strategies used in digital image processing.

Flowers and weeds: cell-type specific pruning in the developing visual thalamus.

In the first weeks of vertebrate postnatal life, neural networks in the visual thalamus undergo activity-dependent refinement thought to be important for the development of functional vision. This process involves pruning of synaptic connections between retinal ganglion cells and excitatory thalamic neurons that relay signals on to visual areas of the cortex. A recent report in Neural Development shows that this does not occur in inhibitory neurons, questioning our current understanding of the development of mature neural circuits.

Contact repulsion controls the dispersion and final distribution of Cajal-Retzius cells.

Cajal-Retzius (CR) cells play a fundamental role in the development of the mammalian cerebral cortex. They control the formation of cortical layers by regulating the migration of pyramidal cells through the release of Reelin. The function of CR cells critically depends on their regular distribution throughout the surface of the cortex, but little is known about the events controlling this phenomenon. Using time-lapse video microscopy in vivo and in vitro, we found that movement of CR cells is regulated by repulsive interactions, which leads to their random dispersion throughout the cortical surface. Mathematical modeling reveals that contact repulsion is both necessary and sufficient for this process, which demonstrates that complex neuronal assemblies may emerge during development through stochastic events. At the molecular level, we found that contact repulsion is mediated by Eph/ephrin interactions. Our observations reveal a mechanism that controls the even distribution of neurons in the developing brain.

Sensory input drives multiple intracellular information streams in somatosensory cortex.

Stable perception arises from the interaction between sensory inputs and internal activity fluctuations in cortex. Here we analyzed how different types of activity contribute to cortical sensory processing at the cellular scale. We performed whole-cell recordings in the barrel cortex of anesthetized rats while applying ongoing whisker stimulation and measured the information conveyed about the time-varying stimulus by different types of input (membrane potential) and output (spiking) signals. We found that substantial, comparable amounts of incoming information are carried by two types of membrane potential signal: slow, large (up-down state) fluctuations, and faster (>20 Hz), smaller-amplitude synaptic activity. Both types of activity fluctuation are therefore significantly driven by the stimulus on an ongoing basis. Each stream conveys essentially independent information. Output (spiking) information is contained in spike timing not just relative to the stimulus but also relative to membrane potential fluctuations. Information transfer is favored in up states relative to down states. Thus, slow, ongoing activity fluctuations and finer-scale synaptic activity generate multiple channels for incoming and outgoing information within barrel cortex neurons during ongoing stimulation.