Recent success in training artificial agents and robots derives from a combination of direct learning of behavioral policies and indirect learning via value functions. Policy learning and value learning employ distinct algorithms that optimize behavioral performance and reward prediction, respectively. In animals, behavioral learning and the role of mesolimbic dopamine signaling have been extensively evaluated with respect to reward prediction; however, to date there has been little consideration of how direct policy learning might inform our understanding. Here we used a comprehensive dataset of orofacial and body movements to understand how behavioral policies evolve as naive, head-restrained mice learned a trace conditioning paradigm. Individual differences in initial dopaminergic reward responses correlated with the emergence of learned behavioral policy, but not the emergence of putative value encoding for a predictive cue. Likewise, physiologically-calibrated manipulations of mesolimbic dopamine produced multiple effects inconsistent with value learning but predicted by a neural network-based model that used dopamine signals to set an adaptive rate, not an error signal, for behavioral policy learning. This work provides strong evidence that phasic dopamine activity can regulate direct learning of behavioral policies, expanding the explanatory power of reinforcement learning models for animal learning.

AMPA-type receptors (AMPARs) are rapidly inserted into synapses undergoing long-term potentiation (LTP) to increase synaptic transmission, but how AMPAR-containing vesicles are selectively trafficked to these synapses during LTP is not known. Here we developed a strategy to label AMPAR GluA1 subunits expressed from the endogenous loci of rat hippocampal neurons such that the motion of GluA1-containing vesicles in time-lapse sequences can be characterized using single-particle tracking and mathematical modeling. We find that GluA1-containing vesicles are confined and concentrated near sites of stimulation-induced plasticity. We show that confinement is mediated by actin polymerization, which hinders the active transport of GluA1-containing vesicles along the length of the dendritic shaft by modulating the rheological properties of the cytoplasm. Actin polymerization also facilitates myosin-mediated transport of GluA1-containing vesicles to exocytic sites. We conclude that neurons utilize F-actin to increase vesicular GluA1 reservoirs and promote exocytosis proximal to the sites of neuronal activity.

To establish functional connectivity between two candidate neurons that might form a circuit element, a common approach is to activate an optogenetic tool such as Chrimson in the candidate pre-synaptic neuron and monitor fluorescence of the calcium-sensitive indicator GCaMP in a candidate post-synaptic neuron. While performing such experiments, we found that low levels of leaky Chrimson expression can lead to strong artifactual GCaMP signals in presumptive postsynaptic neurons even when Chrimson is not intentionally expressed in any particular neurons. Withholding all-trans retinal, the chromophore required as a co-factor for Chrimson response to light, eliminates GCaMP signal but does not provide an experimental control for leaky Chrimson expression. Leaky Chrimson expression appears to be an inherent feature of current Chrimson transgenes, since artifactual connectivity was detected with Chrimson transgenes integrated into three different genomic locations (two insertions tested in larvae; a third insertion tested in the adult fly). These false-positive signals may complicate the interpretation of functional connectivity experiments. We illustrate how a no-Gal4 negative control improves interpretability of functional connectivity assays. We also propose a simple but effective procedure to identify experimental conditions that minimize potentially incorrect interpretations caused by leaky Chrimson expression.

Memory guides the choices an animal makes across widely varying conditions in dynamic environments. Consequently, the most adaptive choice depends on the options available. How can a single memory support optimal behavior across different sets of choice options? We address this using olfactory learning in Drosophila. Even when we restrict an odor-punishment association to a single set of synapses using optogenetics, we find that flies still show choice behavior that depends on the options it encounters. Here we show that how the odor choices are presented to the animal influences memory recall itself. Presenting two similar odors in sequence enabled flies to not only discriminate them behaviorally but also at the level of neural activity. However, when the same odors were encountered as solitary stimuli, no such differences were detectable. These results show that memory recall is not simply a comparison to a static learned template, but can be adaptively modulated by stimulus dynamics.

To flexibly navigate, many animals rely on internal spatial representations that persist when the animal is standing still in darkness, and update accurately by integrating the animal's movements in the absence of localizing sensory cues. Theories of mammalian head direction cells have proposed that these dynamics can be realized in a special class of networks that maintain a localized bump of activity via structured recurrent connectivity, and that shift this bump of activity via angular velocity input. Although there are many different variants of these so-called ring attractor networks, they all rely on large numbers of neurons to generate representations that persist in the absence of input and accurately integrate angular velocity input. Surprisingly, in the fly, Drosophila melanogaster, a head direction representation is maintained by a much smaller number of neurons whose dynamics and connectivity resemble those of a ring attractor network. These findings challenge our understanding of ring attractors and their putative implementation in neural circuits. Here, we analyzed failures of angular velocity integration that emerge in small attractor networks with only a few computational units. Motivated by the peak performance of the fly head direction system in darkness, we mathematically derived conditions under which small networks, even with as few as 4 neurons, achieve the performance of much larger networks. The resulting description reveals that by appropriately tuning the network connectivity, the network can maintain persistent representations over the continuum of head directions, and it can accurately integrate angular velocity inputs. We then analytically determined how performance degrades as the connectivity deviates from this optimally-tuned setting, and we find a trade-off between network size and the tuning precision needed to achieve persistence and accurate integration. This work shows how even small networks can accurately track an animal's movements to guide navigation, and it informs our understanding of the functional capabilities of discrete systems more broadly.

Gustatory sensory neurons detect caloric and harmful compounds in potential food and convey this information to the brain to inform feeding decisions. To examine the signals that gustatory neurons transmit and receive, we reconstructed gustatory axons and their synaptic sites in the adult brain, utilizing a whole-brain electron microscopy volume. We reconstructed 87 gustatory projections from the proboscis labellum in the right hemisphere and 57 from the left, representing the majority of labellar gustatory axons. Gustatory neurons contain a nearly equal number of interspersed pre- and postsynaptic sites, with extensive synaptic connectivity among gustatory axons. Morphology- and connectivity-based clustering revealed six distinct groups, likely representing neurons recognizing different taste modalities. The vast majority of synaptic connections are between neurons of the same group. This study resolves the anatomy of labellar gustatory projections, reveals that gustatory projections are segregated based on taste modality, and uncovers synaptic connections that may alter the transmission of gustatory signals.

To interpret the sensory environment, the brain combines ambiguous sensory measurements with context-specific prior experience. But environmental contexts can change abruptly and unpredictably, resulting in uncertainty about the current context. Here we address two questions: how should context-specific prior knowledge optimally guide the interpretation of sensory stimuli in changing environments, and do human decision-making strategies resemble this optimum? We probe these questions with a task in which subjects report the orientation of ambiguous visual stimuli that were drawn from three dynamically switching distributions, representing different environmental contexts. We derive predictions for an ideal Bayesian observer that leverages the statistical structure of the task to maximize decision accuracy and show that its decisions are biased by task context. The magnitude of this decision bias is not a fixed property of the sensory measurement but depends on the observer's belief about the current context. The model therefore predicts that decision bias will grow with the reliability of the context cue, the stability of the environment, and with the number of trials since the last context switch. Analysis of human choice data validates all three predictions, providing evidence that the brain continuously updates probabilistic representations of the environment to best interpret an uncertain, ever-changing world.

While we think of neurons as having a fixed identity, many show spectacular plasticity. Metamorphosis drives massive changes in the fly brain; neurons that persist into adulthood often change in response to the steroid hormone ecdysone. Besides driving remodeling, ecdysone signaling can also alter the differentiation status of neurons. The three sequentially born subtypes of mushroom body (MB) Kenyon cells (γ, followed by α'/β', and finally α/β) serve as a model of temporal fating. γ neurons are also used as a model of remodeling during metamorphosis. As γ neurons are the only functional Kenyon cells in the larval brain, they serve the function of all three adult subtypes. Correspondingly, larval γ neurons have a similar morphology to α'/β' and α/β neurons-their axons project dorsally and medially. During metamorphosis, γ neurons remodel to form a single medial projection. Both temporal fate changes and defects in remodeling therefore alter γ-neuron morphology in similar ways. Mamo, a broad-complex, tramtrack, and bric-à-brac/poxvirus and zinc finger (BTB/POZ) transcription factor critical for temporal specification of α'/β' neurons, was recently described as essential for γ remodeling. In a previous study, we noticed a change in the number of adult Kenyon cells expressing γ-specific markers when mamo was manipulated. These data implied a role for Mamo in γ-neuron fate specification, yet mamo is not expressed in γ neurons until pupariation, well past γ specification. This indicates that mamo has a later role in ensuring that γ neurons express the correct Kenyon cell subtype-specific genes in the adult brain.

The three-dimensional (3D) genome structure plays a fundamental role in gene regulation and cellular functions. Recent studies in 3D genomics inferred the very basic functional chromatin folding structures known as chromatin loops, the long-range chromatin interactions that are mediated by protein factors and dynamically extruded by cohesin. We combined the use of FISH staining of a very short (33 kb) chromatin fragment, interferometric photoactivated localization microscopy (iPALM), and traveling salesman problem-based heuristic loop reconstruction algorithm from an image of the one of the strongest CTCF-mediated chromatin loops in human lymphoblastoid cells. In total, we have generated thirteen good quality images of the target chromatin region with 2-22 nm oligo probe localization precision. We visualized the shape of the single chromatin loops with unprecedented genomic resolution which allowed us to study the structural heterogeneity of chromatin looping. We were able to compare the physical distance maps from all reconstructed image-driven computational models with contact frequencies observed by ChIA-PET and Hi-C genomic-driven methods to examine the concordance between single cell imaging and population based genomic data.

Topographic maps, the systematic spatial ordering of neurons by response tuning, are common across species. In Drosophila, the lobula columnar (LC) neuron types project from the optic lobe to the central brain, where each forms a glomerulus in a distinct position. However, the advantages of this glomerular arrangement are unclear. Here, we examine the functional and spatial relationships of 10 glomeruli using single-neuron calcium imaging. We discover novel detectors for objects smaller than the lens resolution (LC18) and for complex line motion (LC25). We find that glomeruli are spatially clustered by selectivity for looming versus drifting object motion and ordered by size tuning to form a topographic visual feature map. Furthermore, connectome analysis shows that downstream neurons integrate from sparse subsets of possible glomeruli combinations, which are biased for glomeruli encoding similar features. LC neurons are thus an explicit example of distinct feature detectors topographically organized to facilitate downstream circuit integration.