Visual pursuit behavior in mice maintains the pursued prey on the retinal region with least optic flow.

Mice have a large visual field that is constantly stabilized by vestibular ocular reflex (VOR) driven eye rotations that counter head-rotations. While maintaining their extensive visual coverage is advantageous for predator detection, mice also track and capture prey using vision. However, in the freely moving animal quantifying object location in the field of view is challenging. Here, we developed a method to digitally reconstruct and quantify the visual scene of freely moving mice performing a visually based prey capture task. By isolating the visual sense and combining a mouse eye optic model with the head and eye rotations, the detailed reconstruction of the digital environment and retinal features were projected onto the corneal surface for comparison, and updated throughout the behavior. By quantifying the spatial location of objects in the visual scene and their motion throughout the behavior, we show that the prey image consistently falls within a small area of the VOR-stabilized visual field. This functional focus coincides with the region of minimal optic flow within the visual field and consequently area of minimal motion-induced image-blur, as during pursuit mice ran directly toward the prey. The functional focus lies in the upper-temporal part of the retina and coincides with the reported high density-region of Alpha-ON sustained retinal ganglion cells.

Mice have a lot to keep an eye on. To survive, they need to dodge predators looming on land and from the skies, while also hunting down the small insects that are part of their diet. To do this, they are helped by their large panoramic field of vision, which stretches from behind and over their heads to below their snouts. To stabilize their gaze when they are on the prowl, mice reflexively move their eyes to counter the movement of their head: in fact, they are unable to move their eyes independently. This raises the question: what part of their large visual field of view do these rodents use when tracking a prey, and to what advantage? This is difficult to investigate, since it requires simultaneously measuring the eye and head movements of mice as they chase and capture insects. In response, Holmgren, Stahr et al. developed a new technique to record the precise eye positions, head rotations and prey location of mice hunting crickets in surroundings that were fully digitized at high resolution. Combining this information allowed the team to mathematically recreate what mice would see as they chased the insects, and to assess what part of their large visual field they were using. This revealed that, once a cricket had entered any part of the mice's large field of view, the rodents shifted their head ­ but not their eyes ­ to bring the prey into both eye views, and then ran directly at it. If the insect escaped, the mice repeated that behavior. During the pursuit, the cricket's position was mainly held in a small area of the mouse's view that corresponds to a specialized region in the eye which is thought to help track objects. This region also allowed the least motion-induced image blur when the animals were running forward. The approach developed by Holmgren, Stahr et al. gives a direct insight into what animals see when they hunt, and how this constantly changing view ties to what happens in the eyes. This method could be applied to other species, ushering in a new wave of tools to explore what freely moving animals see, and the relationship between behaviour and neural circuitry.

Training deep neural density estimators to identify mechanistic models of neural dynamics.

Mechanistic modeling in neuroscience aims to explain observed phenomena in terms of underlying causes. However, determining which model parameters agree with complex and stochastic neural data presents a significant challenge. We address this challenge with a machine learning tool which uses deep neural density estimators-trained using model simulations-to carry out Bayesian inference and retrieve the full space of parameters compatible with raw data or selected data features. Our method is scalable in parameters and data features and can rapidly analyze new data after initial training. We demonstrate the power and flexibility of our approach on receptive fields, ion channels, and Hodgkin-Huxley models. We also characterize the space of circuit configurations giving rise to rhythmic activity in the crustacean stomatogastric ganglion, and use these results to derive hypotheses for underlying compensation mechanisms. Our approach will help close the gap between data-driven and theory-driven models of neural dynamics.

Computational neuroscientists use mathematical models built on observational data to investigate what's happening in the brain. Models can simulate brain activity from the behavior of a single neuron right through to the patterns of collective activity in whole neural networks. Collecting the experimental data is the first step, then the challenge becomes deciding which computer models best represent the data and can explain the underlying causes of how the brain behaves. Researchers usually find the right model for their data through trial and error. This involves tweaking a model's parameters until the model can reproduce the data of interest. But this process is laborious and not systematic. Moreover, with the ever-increasing complexity of both data and computer models in neuroscience, the old-school approach of building models is starting to show its limitations. Now, Gonçalves, Lueckmann, Deistler et al. have designed an algorithm that makes it easier for researchers to fit mathematical models to experimental data. First, the algorithm trains an artificial neural network to predict which models are compatible with simulated data. After initial training, the method can rapidly be applied to either raw experimental data or selected data features. The algorithm then returns the models that generate the best match. This newly developed machine learning tool was able to automatically identify models which can replicate the observed data from a diverse set of neuroscience problems. Importantly, further experiments showed that this new approach can be scaled up to complex mechanisms, such as how a neural network in crabs maintains its rhythm of activity. This tool could be applied to a wide range of computational investigations in neuroscience and other fields of biology, which may help bridge the gap between 'data-driven' and 'theory-driven' approaches.

Characterizing Retinal Ganglion Cell Responses to Electrical Stimulation Using Generalized Linear Models.

The ability to preferentially stimulate different retinal pathways is an important area of research for improving visual prosthetics. Recent work has shown that different classes of retinal ganglion cells (RGCs) have distinct linear electrical input filters for low-amplitude white noise stimulation. The aim of this study is to provide a statistical framework for characterizing how RGCs respond to white-noise electrical stimulation. We used a nested family of Generalized Linear Models (GLMs) to partition neural responses into different components-progressively adding covariates to the GLM which captured non-stationarity in neural activity, a linear dependence on the stimulus, and any remaining non-linear interactions. We found that each of these components resulted in increased model performance, but that even the non-linear model left a substantial fraction of neural variability unexplained. The broad goal of this paper is to provide a much-needed theoretical framework to objectively quantify stimulus paradigms in terms of the types of neural responses that they elicit (linear vs. non-linear vs. stimulus-independent variability). In turn, this aids the prosthetic community in the search for optimal stimulus parameters that avoid indiscriminate retinal activation and adaptation caused by excessively large stimulus pulses, and avoid low fidelity responses (low signal-to-noise ratio) caused by excessively weak stimulus pulses.

Intranasal Oxytocin and Vasopressin Modulate Divergent Brainwide Functional Substrates.

The neuropeptides oxytocin (OXT) and vasopressin (AVP) have been identified as modulators of emotional social behaviors and associated with neuropsychiatric disorders characterized by social dysfunction. Experimental and therapeutic use of OXT and AVP via the intranasal route is the subject of extensive clinical research. However, the large-scale functional substrates directly engaged by these peptides and their functional dynamics remain elusive. By using cerebral blood volume (CBV) weighted fMRI in the mouse, we show that intranasal administration of OXT rapidly elicits the transient activation of cortical regions and a sustained activation of hippocampal and forebrain areas characterized by high oxytocin receptor density. By contrast, intranasal administration of AVP produced a robust and sustained deactivation in cortico-parietal, thalamic and mesolimbic regions. Importantly, intravenous administration of OXT and AVP did not recapitulate the patterns of modulation produced by intranasal dosing, supporting a central origin of the observed functional changes. In keeping with this notion, hippocampal local field potential recordings revealed multi-band power increases upon intranasal OXT administration. We also show that the selective OXT-derivative TGOT reproduced the pattern of activation elicited by OXT and that the deletion of OXT receptors does not affect AVP-mediated deactivation. Collectively, our data document divergent modulation of brainwide neural systems by intranasal administration of OXT and AVP, an effect that involves key substrates of social and emotional behavior. The observed divergence calls for a deeper investigation of the systems-level mechanisms by which exogenous OXT and AVP modulate brain function and exert their putative therapeutic effects.