Oscillations and sparsening of odor representations in the mushroom body

An insect-based bioelectronic sensing system combining flexible dual-sided microelectrode array and insect olfactory circuitry for human lung cancer detection

Early detection of lung cancer significantly enhances treatment outcomes, yet current screening methods are limited by accessibility, sensitivity, and cost. This study introduces a bioelectronic sensing platform that integrates the highly sensitive locust olfactory system with a flexible dual-sided microelectrode array (MEA), for robust, noninvasive, and label-free detection of volatile lung cancer biomarkers. Using an innovative folding-annealing fabrication technique and PEDOT:PSS surface functionalization, we developed flexible, dual-sided MEAs with high electrode densities of 463, 687, and 766 channels/mm2 across prototypes, maintaining low impedance (within 4 × 104 Ω). These MEAs demonstrated mechanical flexibility and stability, enabling direct insertion into locust brain tissue without mechanical reinforcement and facilitating precise recording of neural activity in the antennal lobe triggered by lung cancer-related volatile organic compounds (VOCs) from low concentration (1 ppm). Advanced dimensionality reduction techniques applied to the electrophysiological recordings identified distinct neural response patterns to each VOC biomarker and the complex "scent" emitted from various cell lines. Using high-dimensional population neuronal response analysis with a leave-one-trial-out approach, the platform achieved a 100 % classification success rate for unknown VOCs. Additionally, varying concentrations (ppm-ppb) of individual VOC biomarkers were detected and classified with an accuracy of 86 %. The system was further tested for its ability to detect and classify human lung cancer cell lines based on the unique "scent" of cultured cells, including two non-small cell lung cancer (NSCLC) and two small cell lung cancer (SCLC) types. Quantitative assessments demonstrated that the platform achieved a classification accuracy of 85 % across these cell lines. These results substantiate the platform's potential for enhancing clinical diagnostics through the accurate identification of lung cancer stages and cell types. By integrating biological sensory systems with advanced bioelectronics, this study introduces a novel and efficient approach to lung cancer biomarker detection. It provides a non-invasive, brain-based cancer screening method, offering an accessible and innovative solution for early lung cancer diagnosis.

Plasticity in inhibitory networks improves pattern separation in early olfactory processing

Distinguishing between nectar and non-nectar odors is challenging for animals due to shared compounds and varying ratios in complex mixtures. Changes in nectar production throughout the day and over the animal's lifetime add to the complexity. The honeybee olfactory system, containing fewer than 1000 principal neurons in the early olfactory relay, the antennal lobe (AL), must learn to associate diverse volatile blends with rewards. Previous studies identified plasticity in the AL circuits, but its role in odor learning remains poorly understood. Using a biophysical computational model, tuned by in vivo electrophysiological data, and live imaging of the honeybee's AL, we explored the neural mechanisms of plasticity in the AL. Our findings revealed that when trained with a set of rewarded and unrewarded odors, the AL inhibitory network suppresses responses to shared chemical compounds while enhancing responses to distinct compounds. This results in improved pattern separation and a more concise neural code. Our calcium imaging data support these predictions. Analysis of a graph convolutional neural network performing an odor categorization task revealed a similar mechanism for contrast enhancement. Our study provides insights into how inhibitory plasticity in the early olfactory network reshapes the coding for efficient learning of complex odors.

Plasticity in inhibitory networks improves pattern separation in early olfactory processing

Distinguishing between nectar and non-nectar odors is challenging for animals due to shared compounds and varying ratios in complex mixtures. Changes in nectar production throughout the day - and potentially many times within a forager's lifetime - add to the complexity. The honeybee olfactory system, containing fewer than 1,000 principal neurons in the early olfactory relay, the antennal lobe (AL), must learn to associate diverse volatile blends with rewards. Previous studies identified plasticity in the AL circuits, but its role in odor learning remains poorly understood. Using a biophysical computational network model, tuned by in vivo electrophysiological data, and live imaging of the honeybee's AL, we explored the neural mechanisms and functions of plasticity in the early olfactory system. Our findings revealed that when trained with a set of rewarded and unrewarded odors, the AL inhibitory network suppresses shared chemical compounds while enhancing responses to distinct compounds. This results in improved pattern separation and a more concise neural code. Our calcium imaging data support these predictions. Analysis of a graph convolutional neural network performing an odor categorization task revealed a similar mechanism for contrast enhancement. Our study provides insights into how inhibitory plasticity in the early olfactory network reshapes the coding for efficient learning of complex odors.

Optogenetic interrogation of the lateral-line sensory system reveals mechanisms of pattern separation in the zebrafish brain

The ability of animals to interact with their environment hinges on the brain's capacity to distinguish between patterns of sensory information and accurately attribute them to specific sensory organs. The mechanisms by which neuronal circuits discriminate and encode the source of sensory signals remain elusive. To address this, we utilized as a model the posterior lateral line system of larval zebrafish, which is used to detect water currents. This system comprises a series of mechanosensory organs called neuromasts, which are innervated by neurons from the posterior lateral line ganglion. By combining single-neuromast optogenetic stimulation with whole-brain calcium imaging, we developed a novel approach to investigate how inputs from neuromasts are processed. Upon stimulating individual neuromasts, we observed that neurons in the brain of the zebrafish show diverse selectivity properties despite a lack of topographic organization in second-order circuits. We further demonstrated that complex combinations of neuromast stimulation are represented by sparse ensembles of neurons within the medial octavolateralis nucleus (MON) and found that neuromast input can be integrated nonlinearly. Our approach offers an innovative method for spatiotemporally interrogating the zebrafish lateral line system and presents a valuable model for studying whole-brain sensory encoding.

Driver lines for studying associative learning in Drosophila

The mushroom body (MB) is the center for associative learning in insects. In Drosophila, intersectional split-GAL4 drivers and electron microscopy (EM) connectomes have laid the foundation for precise interrogation of the MB neural circuits. However, investigation of many cell types upstream and downstream of the MB has been hindered due to lack of specific driver lines. Here we describe a new collection of over 800 split-GAL4 and split-LexA drivers that cover approximately 300 cell types, including sugar sensory neurons, putative nociceptive ascending neurons, olfactory and thermo-/hygro-sensory projection neurons, interneurons connected with the MB-extrinsic neurons, and various other cell types. We characterized activation phenotypes for a subset of these lines and identified a sugar sensory neuron line most suitable for reward substitution. Leveraging the thousands of confocal microscopy images associated with the collection, we analyzed neuronal morphological stereotypy and discovered that one set of mushroom body output neurons, MBON08/MBON09, exhibits striking individuality and asymmetry across animals. In conjunction with the EM connectome maps, the driver lines reported here offer a powerful resource for functional dissection of neural circuits for associative learning in adult Drosophila.

Mosaic evolution of a learning and memory circuit in Heliconiini butterflies

How do neural circuits accommodate changes that produce cognitive variation? We explore this question by analyzing the evolutionary dynamics of an insect learning and memory circuit centered within the mushroom body. Mushroom bodies are composed of a conserved wiring logic, mainly consisting of Kenyon cells, dopaminergic neurons, and mushroom body output neurons. Despite this conserved makeup, there is huge diversity in mushroom body size and shape across insects. However, empirical data on how evolution modifies the function and architecture of this circuit are largely lacking. To address this, we leverage the recent radiation of a Neotropical tribe of butterflies, the Heliconiini (Nymphalidae), which show extensive variation in mushroom body size over comparatively short phylogenetic timescales, linked to specific changes in foraging ecology, life history, and cognition. To understand how such an extensive increase in size is accommodated through changes in lobe circuit architecture, we combined immunostainings of structural markers, neurotransmitters, and neural injections to generate new, quantitative anatomies of the Nymphalid mushroom body lobe. Our comparative analyses across Heliconiini demonstrate that some Kenyon cell sub-populations expanded at higher rates than others in Heliconius and identify an additional increase in GABA-ergic feedback neurons, which are essential for non-elemental learning and sparse coding. Taken together, our results demonstrate mosaic evolution of functionally related neural systems and cell types and identify that evolutionary malleability in an architecturally conserved parallel circuit guides adaptation in cognitive ability.

Human olfactory perception embeds fine temporal resolution within a single sniff

A sniff in humans typically lasts one to three seconds and is commonly considered to produce a long-exposure shot of the chemical environment that sets the temporal limit of olfactory perception. To break this limit, we devised a sniff-triggered apparatus that controls odorant deliveries within a sniff with a precision of 18 milliseconds. Using this apparatus, we show through rigorous psychophysical testing of 229 participants (649 sessions) that two odorants presented in one order and its reverse become perceptually discriminable when the stimulus onset asynchrony is merely 60 milliseconds (Cohen's d = 0.48; 95% confidence interval, (55, 59); 120-millisecond difference). Discrimination performance improves with the length of stimulus onset asynchrony and is independent of explicit knowledge of the temporal order of odorants or the relative amount of odorant molecules accumulated in a sniff. Our findings demonstrate that human olfactory perception is sensitive to chemical dynamics within a single sniff and provide behavioural evidence for a temporal code of odour identity.

Precision detection of select human lung cancer biomarkers and cell lines using honeybee olfactory neural circuitry as a novel gas sensor

Human breath contains biomarkers (odorants) that can be targeted for early disease detection. It is well known that honeybees have a keen sense of smell and can detect a wide variety of odors at low concentrations. Here, we employ honeybee olfactory neuronal circuitry to classify human lung cancer volatile biomarkers at different concentrations and their mixtures at concentration ranges relevant to biomarkers in human breath from parts-per-billion to parts-per-trillion. We also validated this brain-based sensing technology by detecting human non-small cell lung cancer (NSCLC) and small cell lung cancer (SCLC) cell lines using the 'smell' of the cell cultures. Different lung cancer biomarkers evoked distinct spiking response dynamics in the honeybee antennal lobe neurons indicating that those neurons encoded biomarker-specific information. By investigating lung cancer biomarker-evoked population neuronal responses from the honeybee antennal lobe, we classified individual human lung cancer biomarkers successfully (88% success rate). When we mixed six lung cancer biomarkers at different concentrations to create 'synthetic lung cancer' vs. 'synthetic healthy' human breath, honeybee population neuronal responses were able to classify those complex breath mixtures reliably with exceedingly high accuracy (93-100% success rate with a leave-one-trial-out classification method). Finally, we employed this sensor to detect human NSCLC and SCLC cell lines and we demonstrated that honeybee brain olfactory neurons could distinguish between lung cancer vs. healthy cell lines and could differentiate between different NSCLC and SCLC cell lines successfully (82% classification success rate). These results indicate that the honeybee olfactory system can be used as a sensitive biological gas sensor to detect human lung cancer.

Analysis of fast calcium dynamics of honey bee olfactory coding

Odour processing exhibits multiple parallels between vertebrate and invertebrate olfactory systems. Insects, in particular, have emerged as relevant models for olfactory studies because of the tractability of their olfactory circuits. Here, we used fast calcium imaging to track the activity of projection neurons in the honey bee antennal lobe (AL) during olfactory stimulation at high temporal resolution. We observed a heterogeneity of response profiles and an abundance of inhibitory activities, resulting in various response latencies and stimulus-specific post-odour neural signatures. Recorded calcium signals were fed to a mushroom body (MB) model constructed implementing the fundamental features of connectivity between olfactory projection neurons, Kenyon cells (KC), and MB output neurons (MBON). The model accounts for the increase of odorant discrimination in the MB compared to the AL and reveals the recruitment of two distinct KC populations that represent odorants and their aftersmell as two separate but temporally coherent neural objects. Finally, we showed that the learning-induced modulation of KC-to-MBON synapses can explain both the variations in associative learning scores across different conditioning protocols used in bees and the bees' response latency. Thus, it provides a simple explanation of how the time contingency between the stimulus and the reward can be encoded without the need for time tracking. This study broadens our understanding of olfactory coding and learning in honey bees. It demonstrates that a model based on simple MB connectivity rules and fed with real physiological data can explain fundamental aspects of odour processing and associative learning.

Ring-shaped odor coding in the antennal lobe of migratory locusts

The representation of odors in the locust antennal lobe with its >2,000 glomeruli has long remained a perplexing puzzle. We employed the CRISPR-Cas9 system to generate transgenic locusts expressing the genetically encoded calcium indicator GCaMP in olfactory sensory neurons. Using two-photon functional imaging, we mapped the spatial activation patterns representing a wide range of ecologically relevant odors across all six developmental stages. Our findings reveal a functionally ring-shaped organization of the antennal lobe composed of specific glomerular clusters. This configuration establishes an odor-specific chemotopic representation by encoding different chemical classes and ecologically distinct odors in the form of glomerular rings. The ring-shaped glomerular arrangement, which we confirm by selective targeting of OR70a-expressing sensory neurons, occurs throughout development, and the odor-coding pattern within the glomerular population is consistent across developmental stages. Mechanistically, this unconventional spatial olfactory code reflects the locust-specific and multiplexed glomerular innervation pattern of the antennal lobe.

A lightweight data-driven spiking neuronal network model of Drosophila olfactory nervous system with dedicated hardware support

Data-driven spiking neuronal network (SNN) models enable in-silico analysis of the nervous system at the cellular and synaptic level. Therefore, they are a key tool for elucidating the information processing principles of the brain. While extensive research has focused on developing data-driven SNN models for mammalian brains, their complexity poses challenges in achieving precision. Network topology often relies on statistical inference, and the functions of specific brain regions and supporting neuronal activities remain unclear. Additionally, these models demand huge computing facilities and their simulation speed is considerably slower than real-time. Here, we propose a lightweight data-driven SNN model that strikes a balance between simplicity and reproducibility. The model is built using a qualitative modeling approach that can reproduce key dynamics of neuronal activity. We target the Drosophila olfactory nervous system, extracting its network topology from connectome data. The model was successfully implemented on a small entry-level field-programmable gate array and simulated the activity of a network in real-time. In addition, the model reproduced olfactory associative learning, the primary function of the olfactory system, and characteristic spiking activities of different neuron types. In sum, this paper propose a method for building data-driven SNN models from biological data. Our approach reproduces the function and neuronal activities of the nervous system and is lightweight, acceleratable with dedicated hardware, making it scalable to large-scale networks. Therefore, our approach is expected to play an important role in elucidating the brain's information processing at the cellular and synaptic level through an analysis-by-construction approach. In addition, it may be applicable to edge artificial intelligence systems in the future.

Functional Imaging of Learning-Induced Plasticity in the Central Nervous System with Genetically Encoded Reporters in Drosophila

Learning and memory allow animals to adjust their behavior based on the predictive value of their past experiences. Memories often exist in complex representations, spread across numerous cells and synapses in the brain. Studying relatively simple forms of memory provides insights into the fundamental processes that underlie multiple forms of memory. Associative learning occurs when an animal learns the relationship between two previously unrelated sensory stimuli, such as when a hungry animal learns that a particular odor is followed by a tasty reward. Drosophila is a particularly powerful model to study how this type of memory works. The fundamental principles are widely shared among animals, and there is a wide range of genetic tools available to study circuit function in flies. In addition, the olfactory structures that mediate associative learning in flies, such as the mushroom body and its associated neurons, are anatomically organized, relatively well-characterized, and readily accessible to imaging. Here, we review the olfactory anatomy and physiology of the olfactory system, describe how plasticity in the olfactory pathway mediates learning and memory, and explain the general principles underlying calcium imaging approaches.

Sensory encoding and memory in the mushroom body: signals, noise, and variability

To survive in changing environments, animals need to learn to associate specific sensory stimuli with positive or negative valence. How do they form stimulus-specific memories to distinguish between positively/negatively associated stimuli and other irrelevant stimuli? Solving this task is one of the functions of the mushroom body, the associative memory center in insect brains. Here we summarize recent work on sensory encoding and memory in the Drosophila mushroom body, highlighting general principles such as pattern separation, sparse coding, noise and variability, coincidence detection, and spatially localized neuromodulation, and placing the mushroom body in comparative perspective with mammalian memory systems.

What do the mushroom bodies do for the insect brain? Twenty-five years of progress

In 1998, a special edition of Learning & Memory was published with a discrete focus of synthesizing the state of the field to provide an overview of the function of the insect mushroom body. While molecular neuroscience and optical imaging of larger brain areas were advancing, understanding the basic functioning of neuronal circuits, particularly in the context of the mushroom body, was rudimentary. In the past 25 years, technological innovations have allowed researchers to map and understand the in vivo function of the neuronal circuits of the mushroom body system, making it an ideal model for investigating the circuit basis of sensory encoding, memory formation, and behavioral decisions. Collaborative efforts within the community have played a crucial role, leading to an interactive connectome of the mushroom body and accessible genetic tools for studying mushroom body circuit function. Looking ahead, continued technological innovation and collaborative efforts are likely to further advance our understanding of the mushroom body and its role in behavior and cognition, providing insights that generalize to other brain structures and species.

Evolution of neural circuitry and cognition

Neural circuits govern the interface between the external environment, internal cues and outwardly directed behaviours. To process multiple environmental stimuli and integrate these with internal state requires considerable neural computation. Expansion in neural network size, most readily represented by whole brain size, has historically been linked to behavioural complexity, or the predominance of cognitive behaviours. Yet, it is largely unclear which aspects of circuit variation impact variation in performance. A key question in the field of evolutionary neurobiology is therefore how neural circuits evolve to allow improved behavioural performance or innovation. We discuss this question by first exploring how volumetric changes in brain areas reflect actual neural circuit change. We explore three major axes of neural circuit evolution-replication, restructuring and reconditioning of cells and circuits-and discuss how these could relate to broader phenotypes and behavioural variation. This discussion touches on the relevant uses and limitations of volumetrics, while advocating a more circuit-based view of cognition. We then use this framework to showcase an example from the insect brain, the multi-sensory integration and internal processing that is shared between the mushroom bodies and central complex. We end by identifying future trends in this research area, which promise to advance the field of evolutionary neurobiology.

Olfactory system structure and function in newly hatched and adult locusts

An important question in neuroscience is how sensory systems change as animals grow and interact with the environment. Exploring sensory systems in animals as they develop can reveal how networks of neurons process information as the neurons themselves grow and the needs of the animal change. Here we compared the structure and function of peripheral parts of the olfactory pathway in newly hatched and adult locusts. We found that populations of olfactory sensory neurons (OSNs) in hatchlings and adults responded with similar tunings to a panel of odors. The morphologies of local neurons (LNs) and projection neurons (PNs) in the antennal lobes (ALs) were very similar in both age groups, though they were smaller in hatchlings, they were proportional to overall brain size. The odor evoked responses of LNs and PNs were also very similar in both age groups, characterized by complex patterns of activity including oscillatory synchronization. Notably, in hatchlings, spontaneous and odor-evoked firing rates of PNs were lower, and LFP oscillations were lower in frequency, than in the adult. Hatchlings have smaller antennae with fewer OSNs; removing antennal segments from adults also reduced LFP oscillation frequency. Thus, consistent with earlier computational models, the developmental increase in frequency is due to increasing intensity of input to the oscillation circuitry. Overall, our results show that locusts hatch with a fully formed olfactory system that structurally and functionally matches that of the adult, despite its small size and lack of prior experience with olfactory stimuli.

Minimal circuit motifs for second-order conditioning in the insect mushroom body

In well-established first-order conditioning experiments, the concurrence of a sensory cue with reinforcement forms an association, allowing the cue to predict future reinforcement. In the insect mushroom body, a brain region central to learning and memory, such associations are encoded in the synapses between its intrinsic and output neurons. This process is mediated by the activity of dopaminergic neurons that encode reinforcement signals. In second-order conditioning, a new sensory cue is paired with an already established one that presumably activates dopaminergic neurons due to its predictive power of the reinforcement. We explored minimal circuit motifs in the mushroom body for their ability to support second-order conditioning using mechanistic models. We found that dopaminergic neurons can either be activated directly by the mushroom body's intrinsic neurons or via feedback from the output neurons via several pathways. We demonstrated that the circuit motifs differ in their computational efficiency and robustness. Beyond previous research, we suggest an additional motif that relies on feedforward input of the mushroom body intrinsic neurons to dopaminergic neurons as a promising candidate for experimental evaluation. It differentiates well between trained and novel stimuli, demonstrating robust performance across a range of model parameters.

Odorant detection in a locust exhibits unusually low redundancy

Olfactory coding, from insects to humans, is canonically considered to involve considerable across-fiber coding already at the peripheral level, thereby allowing recognition of vast numbers of odor compounds. We show that the migratory locust has evolved an alternative strategy built on highly specific odorant receptors feeding into a complex primary processing center in the brain. By collecting odors from food and different life stages of the locust, we identified 205 ecologically relevant odorants, which we used to deorphanize 48 locust olfactory receptors via ectopic expression in Drosophila. Contrary to the often broadly tuned olfactory receptors of other insects, almost all locust receptors were found to be narrowly tuned to one or very few ligands. Knocking out a single receptor using CRISPR abolished physiological and behavioral responses to the corresponding ligand. We conclude that the locust olfactory system, with most olfactory receptors being narrowly tuned, differs from the so-far described olfactory systems.

Scalable Electrophysiology of Millimeter-Scale Animals with Electrode Devices

Millimeter-scale animals such as Caenorhabditis elegans, Drosophila larvae, zebrafish, and bees serve as powerful model organisms in the fields of neurobiology and neuroethology. Various methods exist for recording large-scale electrophysiological signals from these animals. Existing approaches often lack, however, real-time, uninterrupted investigations due to their rigid constructs, geometric constraints, and mechanical mismatch in integration with soft organisms. The recent research establishes the foundations for 3-dimensional flexible bioelectronic interfaces that incorporate microfabricated components and nanoelectronic function with adjustable mechanical properties and multidimensional variability, offering unique capabilities for chronic, stable interrogation and stimulation of millimeter-scale animals and miniature tissue constructs. This review summarizes the most advanced technologies for electrophysiological studies, based on methods of 3-dimensional flexible bioelectronics. A concluding section addresses the challenges of these devices in achieving freestanding, robust, and multifunctional biointerfaces.

Interactions between specialized gain control mechanisms in olfactory processing

Gain control is a process that adjusts a system's sensitivity when input levels change. Neural systems contain multiple mechanisms of gain control, but we do not understand why so many mechanisms are needed or how they interact. Here, we investigate these questions in the Drosophila antennal lobe, where we identify several types of inhibitory interneurons with specialized gain control functions. We find that some interneurons are nonspiking, with compartmentalized calcium signals, and they specialize in intra-glomerular gain control. Conversely, we find that other interneurons are recruited by strong and widespread network input; they specialize in global presynaptic gain control. Using computational modeling and optogenetic perturbations, we show how these mechanisms can work together to improve stimulus discrimination while also minimizing temporal distortions in network activity. Our results demonstrate how the robustness of neural network function can be increased by interactions among diverse and specialized mechanisms of gain control.

Evolution of the neuronal substrate for kin recognition in social Hymenoptera

In evolutionary terms, life is about reproduction. Yet, in some species, individuals forgo their own reproduction to support the reproductive efforts of others. Social insect colonies for example, can contain up to a million workers that actively cooperate in tasks such as foraging, brood care and nest defence, but do not produce offspring. In such societies the division of labour is pronounced, and reproduction is restricted to just one or a few individuals, most notably the queen(s). This extreme eusocial organisation exists in only a few mammals, crustaceans and insects, but strikingly, it evolved independently up to nine times in the order Hymenoptera (including ants, bees and wasps). Transitions from a solitary lifestyle to an organised society can occur through natural selection when helpers obtain a fitness benefit from cooperating with kin, owing to the indirect transmission of genes through siblings. However, this process, called kin selection, is vulnerable to parasitism and opportunistic behaviours from unrelated individuals. An ability to distinguish kin from non-kin, and to respond accordingly, could therefore critically facilitate the evolution of eusociality and the maintenance of non-reproductive workers. The question of how the hymenopteran brain has adapted to support this function is therefore a fundamental issue in evolutionary neuroethology. Early neuroanatomical investigations proposed that social Hymenoptera have expanded integrative brain areas due to selection for increased cognitive capabilities in the context of processing social information. Later studies challenged this assumption and instead pointed to an intimate link between higher social organisation and the existence of developed sensory structures involved in recognition and communication. In particular, chemical signalling of social identity, known to be mediated through cuticular hydrocarbons (CHCs), may have evolved hand in hand with a specialised chemosensory system in Hymenoptera. Here, we compile the current knowledge on this recognition system, from emitted identity signals, to the molecular and neuronal basis of chemical detection, with particular emphasis on its evolutionary history. Finally, we ask whether the evolution of social behaviour in Hymenoptera could have driven the expansion of their complex olfactory system, or whether the early origin and conservation of an olfactory subsystem dedicated to social recognition could explain the abundance of eusocial species in this insect order. Answering this question will require further comparative studies to provide a comprehensive view on lineage-specific adaptations in the olfactory pathway of Hymenoptera.

The mushroom body output encodes behavioral decision during sensory-motor transformation

Animals form a behavioral decision by evaluating sensory evidence on the background of past experiences and the momentary motivational state. In insects, we still lack understanding of how and at which stage of the recurrent sensory-motor pathway behavioral decisions are formed. The mushroom body (MB), a central brain structure in insects1 and crustaceans,2,3 integrates sensory input of different modalities4,5,6 with the internal state, the behavioral state, and external sensory context7,8,9,10 through a large number of recurrent, mostly neuromodulatory inputs,11,12 implicating a functional role for MBs in state-dependent sensory-motor transformation.13,14 A number of classical conditioning studies in honeybees15,16 and fruit flies17,18,19 have provided accumulated evidence that at its output, the MB encodes the valence of a sensory stimulus with respect to its behavioral relevance. Recent work has extended this notion of valence encoding to the context of innate behaviors.8,20,21,22 Here, we co-analyzed a defined feeding behavior and simultaneous extracellular single-unit recordings from MB output neurons (MBONs) in the cockroach in response to timed sensory stimulation with odors. We show that clear neuronal responses occurred almost exclusively during behaviorally responded trials. Early MBON responses to the sensory stimulus preceded the feeding behavior and predicted its occurrence or non-occurrence from the single-trial population activity. Our results therefore suggest that at its output, the MB does not merely encode sensory stimulus valence. We hypothesize instead that the MB output represents an integrated signal of internal state, momentary environmental conditions, and experience-dependent memory to encode a behavioral decision.

Effects of stochastic coding on olfactory discrimination in flies and mice

Sparse coding can improve discrimination of sensory stimuli by reducing overlap between their representations. Two factors, however, can offset sparse coding's benefits: similar sensory stimuli have significant overlap and responses vary across trials. To elucidate the effects of these 2 factors, we analyzed odor responses in the fly and mouse olfactory regions implicated in learning and discrimination-the mushroom body (MB) and the piriform cortex (PCx). We found that neuronal responses fall along a continuum from extremely reliable across trials to extremely variable or stochastic. Computationally, we show that the observed variability arises from noise within central circuits rather than sensory noise. We propose this coding scheme to be advantageous for coarse- and fine-odor discrimination. More reliable cells enable quick discrimination between dissimilar odors. For similar odors, however, these cells overlap and do not provide distinguishing information. By contrast, more unreliable cells are decorrelated for similar odors, providing distinguishing information, though these benefits only accrue with extended training with more trials. Overall, we have uncovered a conserved, stochastic coding scheme in vertebrates and invertebrates, and we identify a candidate mechanism, based on variability in a winner-take-all (WTA) inhibitory circuit, that improves discrimination with training.

Neuromorphic sequence learning with an event camera on routes through vegetation

For many robotics applications, it is desirable to have relatively low-power and efficient onboard solutions. We took inspiration from insects, such as ants, that are capable of learning and following routes in complex natural environments using relatively constrained sensory and neural systems. Such capabilities are particularly relevant to applications such as agricultural robotics, where visual navigation through dense vegetation remains a challenging task. In this scenario, a route is likely to have high self-similarity and be subject to changing lighting conditions and motion over uneven terrain, and the effects of wind on leaves increase the variability of the input. We used a bioinspired event camera on a terrestrial robot to collect visual sequences along routes in natural outdoor environments and applied a neural algorithm for spatiotemporal memory that is closely based on a known neural circuit in the insect brain. We show that this method is plausible to support route recognition for visual navigation and more robust than SeqSLAM when evaluated on repeated runs on the same route or routes with small lateral offsets. By encoding memory in a spiking neural network running on a neuromorphic computer, our model can evaluate visual familiarity in real time from event camera footage.

A biologically inspired auto-associative network with sparse temporal population coding

Associative system has attracted increasing attention for it can store basic information and then infer details to match perception with an efficient self-organization algorithm. However, the implementation of the associative system with the application of real-world data is relatively difficult. To address this issue, we propose a novel biologically inspired auto-associative (BIAA) network to explore the structure, encoding and formation of associative memory as well as to extend the ability to real-world application. Our network is constructed by imitating the organization of the cortical minicolumns where each minicolumn contains plenty of parallel biological spiking neurons. To allow the network to learn and predict one symbol per theta cycle, we incorporate synaptic delay and theta oscillation into the neuron dynamic process. Subsequently, we design a sparse temporal population (STP) coding scheme that allows each input symbol to be represented as stable, unique, and easily recallable sparsely distributed representations. By combining associative learning dynamics with the STP coding, our network realizes efficient storage and inference in an ordered manner. Experimental results indicate that the proposed network successfully performs sequence retrieval from partial text and sequence recovery from distorted information. BIAA network provides new insight into introducing biologically inspired mechanisms into associative system and has enormous potential for hardware and software applications.

Beyond ℓ1 sparse coding in V1

Growing evidence indicates that only a sparse subset from a pool of sensory neurons is active for the encoding of visual stimuli at any instant in time. Traditionally, to replicate such biological sparsity, generative models have been using the ℓ1 norm as a penalty due to its convexity, which makes it amenable to fast and simple algorithmic solvers. In this work, we use biological vision as a test-bed and show that the soft thresholding operation associated to the use of the ℓ1 norm is highly suboptimal compared to other functions suited to approximating ℓp with 0 ≤ p < 1 (including recently proposed continuous exact relaxations), in terms of performance. We show that ℓ1 sparsity employs a pool with more neurons, i.e. has a higher degree of overcompleteness, in order to maintain the same reconstruction error as the other methods considered. More specifically, at the same sparsity level, the thresholding algorithm using the ℓ1 norm as a penalty requires a dictionary of ten times more units compared to the proposed approach, where a non-convex continuous relaxation of the ℓ0 pseudo-norm is used, to reconstruct the external stimulus equally well. At a fixed sparsity level, both ℓ0- and ℓ1-based regularization develop units with receptive field (RF) shapes similar to biological neurons in V1 (and a subset of neurons in V2), but ℓ0-based regularization shows approximately five times better reconstruction of the stimulus. Our results in conjunction with recent metabolic findings indicate that for V1 to operate efficiently it should follow a coding regime which uses a regularization that is closer to the ℓ0 pseudo-norm rather than the ℓ1 one, and suggests a similar mode of operation for the sensory cortex in general.

Redundant representations are required to disambiguate simultaneously presented complex stimuli

A pedestrian crossing a street during rush hour often looks and listens for potential danger. When they hear several different horns, they localize the cars that are honking and decide whether or not they need to modify their motor plan. How does the pedestrian use this auditory information to pick out the corresponding cars in visual space? The integration of distributed representations like these is called the assignment problem, and it must be solved to integrate distinct representations across but also within sensory modalities. Here, we identify and analyze a solution to the assignment problem: the representation of one or more common stimulus features in pairs of relevant brain regions-for example, estimates of the spatial position of cars are represented in both the visual and auditory systems. We characterize how the reliability of this solution depends on different features of the stimulus set (e.g., the size of the set and the complexity of the stimuli) and the details of the split representations (e.g., the precision of each stimulus representation and the amount of overlapping information). Next, we implement this solution in a biologically plausible receptive field code and show how constraints on the number of neurons and spikes used by the code force the brain to navigate a tradeoff between local and catastrophic errors. We show that, when many spikes and neurons are available, representing stimuli from a single sensory modality can be done more reliably across multiple brain regions, despite the risk of assignment errors. Finally, we show that a feedforward neural network can learn the optimal solution to the assignment problem, even when it receives inputs in two distinct representational formats. We also discuss relevant results on assignment errors from the human working memory literature and show that several key predictions of our theory already have support.

Mushroom body output neurons MBON-a1/a2 define an odor intensity channel that regulates behavioral odor discrimination learning in larval Drosophila

The sensitivity of animals to sensory input must be regulated to ensure that signals are detected and also discriminable. However, how circuits regulate the dynamic range of sensitivity to sensory stimuli is not well understood. A given odor is represented in the insect mushroom bodies (MBs) by sparse combinatorial coding by Kenyon cells (KCs), forming an odor quality representation. To address how intensity of sensory stimuli is processed at the level of the MB input region, the calyx, we characterized a set of novel mushroom body output neurons that respond preferentially to high odor concentrations. We show that a pair of MB calyx output neurons, MBON-a1/2, are postsynaptic in the MB calyx, where they receive extensive synaptic inputs from KC dendrites, the inhibitory feedback neuron APL, and octopaminergic sVUM1 neurons, but relatively few inputs from projection neurons. This pattern is broadly consistent in the third-instar larva as well as in the first instar connectome. MBON-a1/a2 presynaptic terminals innervate a region immediately surrounding the MB medial lobe output region in the ipsilateral and contralateral brain hemispheres. By monitoring calcium activity using jRCamP1b, we find that MBON-a1/a2 responses are odor-concentration dependent, responding only to ethyl acetate (EA) concentrations higher than a 200-fold dilution, in contrast to MB neurons which are more concentration-invariant and respond to EA dilutions as low as 10^-4. Optogenetic activation of the calyx-innervating sVUM1 modulatory neurons originating in the SEZ (Subesophageal zone), did not show a detectable effect on MBON-a1/a2 odor responses. Optogenetic activation of MBON-a1/a2 using CsChrimson impaired odor discrimination learning compared to controls. We propose that MBON-a1/a2 form an output channel of the calyx, summing convergent sensory and modulatory input, firing preferentially to high odor concentration, and might affect the activity of downstream MB targets.

Abstract representations emerge naturally in neural networks trained to perform multiple tasks

Humans and other animals demonstrate a remarkable ability to generalize knowledge across distinct contexts and objects during natural behavior. We posit that this ability to generalize arises from a specific representational geometry, that we call abstract and that is referred to as disentangled in machine learning. These abstract representations have been observed in recent neurophysiological studies. However, it is unknown how they emerge. Here, using feedforward neural networks, we demonstrate that the learning of multiple tasks causes abstract representations to emerge, using both supervised and reinforcement learning. We show that these abstract representations enable few-sample learning and reliable generalization on novel tasks. We conclude that abstract representations of sensory and cognitive variables may emerge from the multiple behaviors that animals exhibit in the natural world, and, as a consequence, could be pervasive in high-level brain regions. We also make several specific predictions about which variables will be represented abstractly.

What neural oscillations can and cannot do for syntactic structure building

Understanding what someone says requires relating words in a sentence to one another as instructed by the grammatical rules of a language. In recent years, the neurophysiological basis for this process has become a prominent topic of discussion in cognitive neuroscience. Current proposals about the neural mechanisms of syntactic structure building converge on a key role for neural oscillations in this process, but they differ in terms of the exact function that is assigned to them. In this Perspective, we discuss two proposed functions for neural oscillations - chunking and multiscale information integration - and evaluate their merits and limitations taking into account a fundamentally hierarchical nature of syntactic representations in natural languages. We highlight insights that provide a tangible starting point for a neurocognitive model of syntactic structure building.

SpaRCe: Improved Learning of Reservoir Computing Systems Through Sparse Representations

"Sparse" neural networks, in which relatively few neurons or connections are active, are common in both machine learning and neuroscience. While, in machine learning, "sparsity" is related to a penalty term that leads to some connecting weights becoming small or zero, in biological brains, sparsity is often created when high spiking thresholds prevent neuronal activity. Here, we introduce sparsity into a reservoir computing network via neuron-specific learnable thresholds of activity, allowing neurons with low thresholds to contribute to decision-making but suppressing information from neurons with high thresholds. This approach, which we term "SpaRCe," optimizes the sparsity level of the reservoir without affecting the reservoir dynamics. The read-out weights and the thresholds are learned by an online gradient rule that minimizes an error function on the outputs of the network. Threshold learning occurs by the balance of two opposing forces: reducing interneuronal correlations in the reservoir by deactivating redundant neurons, while increasing the activity of neurons participating in correct decisions. We test SpaRCe on classification problems and find that threshold learning improves performance compared to standard reservoir computing. SpaRCe alleviates the problem of catastrophic forgetting, a problem most evident in standard echo state networks (ESNs) and recurrent neural networks in general, due to increasing the number of task-specialized neurons that are included in the network decisions.

Harnessing insect olfactory neural circuits for detecting and discriminating human cancers

There is overwhelming evidence that presence of cancer alters cellular metabolic processes, and these changes are manifested in emitted volatile organic compound (VOC) compositions of cancer cells. Here, we take a novel forward engineering approach by developing an insect olfactory neural circuit-based VOC sensor for cancer detection. We obtained oral cancer cell culture VOC-evoked extracellular neural responses from in vivo insect (locust) antennal lobe neurons. We employed biological neural computations of the antennal lobe circuitry for generating spatiotemporal neuronal response templates corresponding to each cell culture VOC mixture, and employed these neuronal templates to distinguish oral cancer cell lines (SAS, Ca9-22, and HSC-3) vs. a non-cancer cell line (HaCaT). Our results demonstrate that three different human oral cancers can be robustly distinguished from each other and from a non-cancer oral cell line. By using high-dimensional population neuronal response analysis and leave-one-trial-out methodology, our approach yielded high classification success for each cell line tested. Our analyses achieved 76-100% success in identifying cell lines by using the population neural response (n = 194) collected for the entire duration of the cell culture study. We also demonstrate this cancer detection technique can distinguish between different types of oral cancers and non-cancer at different time-matched points of growth. This brain-based cancer detection approach is fast as it can differentiate between VOC mixtures within 250 ms of stimulus onset. Our brain-based cancer detection system comprises a novel VOC sensing methodology that incorporates entire biological chemosensory arrays, biological signal transduction, and neuronal computations in a form of a forward-engineered technology for cancer VOC detection.

Spike frequency adaptation facilitates the encoding of input gradient in insect olfactory projection neurons

The olfactory system in insects has evolved to process the dynamic changes in the concentration of food odors or sex pheromones to localize the nutrients or conspecific mating partners. Experimental studies have suggested that projection neurons (PNs) in insects encode not only the stimulus intensity but also its rate-of-change (input gradient). In this study, we aim to develop a simple computational model for a PN to understand the mechanism underlying the coding of the rate-of-change information. We show that the spike frequency adaptation is a potential key mechanism for reproducing the phasic response pattern of the PN in Drosophila. We also demonstrate that this adaptation mechanism enables the PN to encode the rate-of-change of the input firing rate. Finally, our model predicts that the PN exhibits the intensity-invariant response for the pulse and ramp odor stimulus. These results suggest that the developed model is useful for investigating the coding principle underlying olfactory information processing in insects.

Drosophila olfaction: past, present and future

Among the many wonders of nature, the sense of smell of the fly Drosophila melanogaster might seem, at first glance, of esoteric interest. Nevertheless, for over a century, the 'nose' of this insect has been an extraordinary system to explore questions in animal behaviour, ecology and evolution, neuroscience, physiology and molecular genetics. The insights gained are relevant for our understanding of the sensory biology of vertebrates, including humans, and other insect species, encompassing those detrimental to human health. Here, I present an overview of our current knowledge of D. melanogaster olfaction, from molecules to behaviours, with an emphasis on the historical motivations of studies and illustration of how technical innovations have enabled advances. I also highlight some of the pressing and long-term questions.

In Search for the Retrievable Memory Trace in an Insect Brain

The search strategy for the memory trace and its semantics is exemplified for the case of olfactory learning in the honeybee brain. The logic of associative learning is used to guide the experimental approach into the brain by identifying the anatomical and functional convergence sites of the conditioned stimulus and unconditioned stimulus pathways. Two of the several convergence sites are examined in detail, the antennal lobe as the first-order sensory coding area, and the input region of the mushroom body as a higher order integration center. The memory trace is identified as the pattern of associative changes on the level of synapses. The synapses are recruited, drop out, and change the transmission properties for both specifically associated stimulus and the non-associated stimulus. Several rules extracted from behavioral studies are found to be mirrored in the patterns of synaptic change. The strengths and the weaknesses of the honeybee as a model for the search for the memory trace are addressed in a comparison with Drosophila. The question is discussed whether the memory trace exists as a hidden pattern of change if it is not retrieved and whether an external reading of the content of the memory trace may ever be possible. Doubts are raised on the basis that the retrieval circuits are part of the memory trace. The concept of a memory trace existing beyond retrieval is defended by referring to two well-documented processes also in the honeybee, memory consolidation during sleep, and transfer of memory across brain areas.

Visualization of learning-induced synaptic plasticity in output neurons of the Drosophila mushroom body γ-lobe

By learning, through experience, which stimuli coincide with dangers, it is possible to predict outcomes and act pre-emptively to ensure survival. In insects, this process is localized to the mushroom body (MB), the circuitry of which facilitates the coincident detection of sensory stimuli and punishing or rewarding cues and, downstream, the execution of appropriate learned behaviors. Here, we focused our attention on the mushroom body output neurons (MBONs) of the γ-lobes that act as downstream synaptic partners of the MB γ-Kenyon cells (KCs) to ask how the output of the MB γ-lobe is shaped by olfactory associative conditioning, distinguishing this from non-associative stimulus exposure effects, and without the influence of downstream modulation. This was achieved by employing a subcellularly localized calcium sensor to specifically monitor activity at MBON postsynaptic sites. Therein, we identified a robust associative modulation within only one MBON postsynaptic compartment (MBON-γ1pedc > α/β), which displayed a suppressed postsynaptic response to an aversively paired odor. While this MBON did not undergo non-associative modulation, the reverse was true across the remainder of the γ-lobe, where general odor-evoked adaptation was observed, but no conditioned odor-specific modulation. In conclusion, associative synaptic plasticity underlying aversive olfactory learning is localized to one distinct synaptic γKC-to-γMBON connection.

Pairing-Dependent Plasticity in a Dissected Fly Brain Is Input-Specific and Requires Synaptic CaMKII Enrichment and Nighttime Sleep

In Drosophila, in vivo functional imaging studies revealed that associative memory formation is coupled to a cascade of neural plasticity events in distinct compartments of the mushroom body (MB). In-depth investigation of the circuit dynamics, however, will require an ex vivo model that faithfully mirrors these events to allow direct manipulations of circuit elements that are inaccessible in the intact fly. The current ex vivo models have been able to reproduce the fundamental plasticity of aversive short-term memory, a potentiation of the MB intrinsic neuron (Kenyon cells [KCs]) responses after artificial learning ex vivo However, this potentiation showed different localization and encoding properties from those reported in vivo and failed to generate the previously reported suppression plasticity in the MB output neurons (MBONs). Here, we develop an ex vivo model using the female Drosophila brain that recapitulates behaviorally evoked plasticity in the KCs and MBONs. We demonstrate that this plasticity accurately localizes to the MB α'3 compartment and is encoded by a coincidence between KC activation and dopaminergic input. The formed plasticity is input-specific, requiring pairing of the conditioned stimulus and unconditioned stimulus pathways; hence, we name it pairing-dependent plasticity. Pairing-dependent plasticity formation requires an intact CaMKII gene and is blocked by previous-night sleep deprivation but is rescued by rebound sleep. In conclusion, we show that our ex vivo preparation recapitulates behavioral and imaging results from intact animals and can provide new insights into mechanisms of memory formation at the level of molecules, circuits, and brain state.SIGNIFICANCE STATEMENT The mammalian ex vivo LTP model enabled in-depth investigation of the hippocampal memory circuit. We develop a parallel model to study the Drosophila mushroom body (MB) memory circuit. Pairing activation of the conditioned stimulus and unconditioned stimulus pathways in dissected brains induces a potentiation pairing-dependent plasticity (PDP) in the axons of α'β' Kenyon cells and a suppression PDP in the dendrites of their postsynaptic MB output neurons, localized in the MB α'3 compartment. This PDP is input-specific and requires the 3' untranslated region of CaMKII Interestingly, ex vivo PDP carries information about the animal's experience before dissection; brains from sleep-deprived animals fail to form PDP, whereas those from animals who recovered 2 h of their lost sleep form PDP.

Categorizing Visual Information in Subpopulations of Honeybee Mushroom Body Output Neurons

Multisensory integration plays a central role in perception, as all behaviors usually require the input of different sensory signals. For instance, for a foraging honeybee the association of a food source includes the combination of olfactory and visual cues to be categorized as a flower. Moreover, homing after successful foraging using celestial cues and the panoramic scenery may be dominated by visual cues. Hence, dependent on the context, one modality might be leading and influence the processing of other modalities. To unravel the complex neural mechanisms behind this process we studied honeybee mushroom body output neurons (MBON). MBONs represent the first processing level after olfactory-visual convergence in the honeybee brain. This was physiologically confirmed in our previous study by characterizing a subpopulation of multisensory MBONs. These neurons categorize incoming sensory inputs into olfactory, visual, and olfactory-visual information. However, in addition to multisensory units a prominent population of MBONs was sensitive to visual cues only. Therefore, we asked which visual features might be represented at this high-order integration level. Using extracellular, multi-unit recordings in combination with visual and olfactory stimulation, we separated MBONs with multisensory responses from purely visually driven MBONs. Further analysis revealed, for the first time, that visually driven MBONs of both groups encode detailed aspects within this individual modality, such as light intensity and light identity. Moreover, we show that these features are separated by different MBON subpopulations, for example by extracting information about brightness and wavelength. Most interestingly, the latter MBON population was tuned to separate UV-light from other light stimuli, which were only poorly differentiated from each other. A third MBON subpopulation was neither tuned to brightness nor to wavelength and encoded the general presence of light. Taken together, our results support the view that the mushroom body, a high-order sensory integration, learning and memory center in the insect brain, categorizes sensory information by separating different behaviorally relevant aspects of the multisensory scenery and that these categories are channeled into distinct MBON subpopulations.

Translaminar recurrence from layer 5 suppresses superficial cortical layers

Information flow in the sensory cortex has been described as a predominantly feedforward sequence with deep layers as the output structure. Although recurrent excitatory projections from layer 5 (L5) to superficial L2/3 have been identified by anatomical and physiological studies, their functional impact on sensory processing remains unclear. Here, we use layer-selective optogenetic manipulations in the primary auditory cortex to demonstrate that feedback inputs from L5 suppress the activity of superficial layers regardless of the arousal level, contrary to the prediction from their excitatory connectivity. This suppressive effect is predominantly mediated by translaminar circuitry through intratelencephalic neurons, with an additional contribution of subcortical projections by pyramidal tract neurons. Furthermore, L5 activation sharpened tone-evoked responses of superficial layers in both frequency and time domains, indicating its impact on cortical spectro-temporal integration. Together, our findings establish a translaminar inhibitory recurrence from deep layers that sharpens feature selectivity in superficial cortical layers.

Animal Models of Neurodegenerative Disease: Recent Advances in Fly Highlight Innovative Approaches to Drug Discovery

Neurodegenerative diseases represent a formidable challenge to global health. As advances in other areas of medicine grant healthy living into later decades of life, aging diseases such as Alzheimer's disease (AD) and other neurodegenerative disorders can diminish the quality of these additional years, owed largely to the lack of efficacious treatments and the absence of durable cures. Alzheimer's disease prevalence is predicted to more than double in the next 30 years, affecting nearly 15 million Americans, with AD-associated costs exceeding $1 billion by 2050. Delaying onset of AD and other neurodegenerative diseases is critical to improving the quality of life for patients and reducing the burden of disease on caregivers and healthcare systems. Significant progress has been made to model disease pathogenesis and identify points of therapeutic intervention. While some researchers have contributed to our understanding of the proteins and pathways that drive biological dysfunction in disease using in vitro and in vivo models, others have provided mathematical, biophysical, and computational technologies to identify potential therapeutic compounds using in silico modeling. The most exciting phase of the drug discovery process is now: by applying a target-directed approach that leverages the strengths of multiple techniques and validates lead hits using Drosophila as an animal model of disease, we are on the fast-track to identifying novel therapeutics to restore health to those impacted by neurodegenerative disease.

Optogenetic and thermogenetic manipulation of defined neural circuits and behaviors in Drosophila

Neural network dynamics underlying flexible animal behaviors remain elusive. The fruit fly Drosophila melanogaster is considered an excellent model in behavioral neuroscience because of its simple neuroanatomical architecture and the availability of various genetic methods. Moreover, Drosophila larvae's transparent body allows investigators to use optical methods on freely moving animals, broadening research directions. Activating or inhibiting well-defined events in excitable cells with a fine temporal resolution using optogenetics and thermogenetics led to the association of functions of defined neural populations with specific behavioral outputs such as the induction of associative memory. Furthermore, combining optogenetics and thermogenetics with state-of-the-art approaches, including connectome mapping and machine learning-based behavioral quantification, might provide a complete view of the experience- and time-dependent variations of behavioral responses. These methodologies allow further understanding of the functional connections between neural circuits and behaviors such as chemosensory, motivational, courtship, and feeding behaviors and sleep, learning, and memory.

Biological constraints on configural odour mixture perception

Animals, including humans, detect odours and use this information to behave efficiently in the environment. Frequently, odours consist of complex mixtures of odorants rather than single odorants, and mixtures are often perceived as configural wholes, i.e. as odour objects (e.g. food, partners). The biological rules governing this 'configural perception' (as opposed to the elemental perception of mixtures through their components) remain weakly understood. Here, we first review examples of configural mixture processing in diverse species involving species-specific biological signals. Then, we present the original hypothesis that at least certain mixtures can be processed configurally across species. Indeed, experiments conducted in human adults, newborn rabbits and, more recently, in rodents and honeybees show that these species process some mixtures in a remarkably similar fashion. Strikingly, a mixture AB (A, ethyl isobutyrate; B, ethyl maltol) induces configural processing in humans, who perceive a mixture odour quality (pineapple) distinct from the component qualities (A, strawberry; B, caramel). The same mixture is weakly configurally processed in rabbit neonates, which perceive a particular odour for the mixture in addition to the component odours. Mice and honeybees also perceive the AB mixture configurally, as they respond differently to the mixture compared with its components. Based on these results and others, including neurophysiological approaches, we propose that certain mixtures are convergently perceived across various species of vertebrates/invertebrates, possibly as a result of a similar anatomical organization of their olfactory systems and the common necessity to simplify the environment's chemical complexity in order to display adaptive behaviours.

Visual learning in a virtual reality environment upregulates immediate early gene expression in the mushroom bodies of honey bees

Free-flying bees learn efficiently to solve numerous visual tasks. Yet, the neural underpinnings of this capacity remain unexplored. We used a 3D virtual reality (VR) environment to study visual learning and determine if it leads to changes in immediate early gene (IEG) expression in specific areas of the bee brain. We focused on kakusei, Hr38 and Egr1, three IEGs that have been related to bee foraging and orientation, and compared their relative expression in the calyces of the mushroom bodies, the optic lobes and the rest of the brain after color discrimination learning. Bees learned to discriminate virtual stimuli displaying different colors and retained the information learned. Successful learners exhibited Egr1 upregulation only in the calyces of the mushroom bodies, thus uncovering a privileged involvement of these brain regions in associative color learning and the usefulness of Egr1 as a marker of neural activity induced by this phenomenon.

The anterior paired lateral neuron normalizes odour-evoked activity in the Drosophila mushroom body calyx

To identify and memorize discrete but similar environmental inputs, the brain needs to distinguish between subtle differences of activity patterns in defined neuronal populations. The Kenyon cells (KCs) of the Drosophila adult mushroom body (MB) respond sparsely to complex olfactory input, a property that is thought to support stimuli discrimination in the MB. To understand how this property emerges, we investigated the role of the inhibitory anterior paired lateral (APL) neuron in the input circuit of the MB, the calyx. Within the calyx, presynaptic boutons of projection neurons (PNs) form large synaptic microglomeruli (MGs) with dendrites of postsynaptic KCs. Combining electron microscopy (EM) data analysis and in vivo calcium imaging, we show that APL, via inhibitory and reciprocal synapses targeting both PN boutons and KC dendrites, normalizes odour-evoked representations in MGs of the calyx. APL response scales with the PN input strength and is regionalized around PN input distribution. Our data indicate that the formation of a sparse code by the KCs requires APL-driven normalization of their MG postsynaptic responses. This work provides experimental insights on how inhibition shapes sensory information representation in a higher brain centre, thereby supporting stimuli discrimination and allowing for efficient associative memory formation.

A model of lateral interactions as the origin of multiwhisker receptive fields in rat barrel cortex

While cells within barrel cortex respond primarily to deflections of their principal whisker (PW), they also exhibit responses to non-principal, or adjacent, whiskers (AWs), albeit responses with diminished amplitudes and longer latencies. The origin of multiwhisker receptive fields of barrel cells remains a point of controversy within the experimental literature, with three contending possibilities: (i) barrel cells inherit their AW responses from the AW responses of thalamocortical (TC) cells within their aligned barreloid; (ii) the axons of TC cells within a barreloid ramify to innervate multiple barrels, rather than only terminating within their aligned barrel; (iii) lateral intracortical transmission between barrels conveys AW responsivity to barrel cells. In this work, we develop a detailed, biologically plausible model of multiple barrels in order to examine possibility (iii); in order to isolate the dynamics that possibility (iii) entails, we incorporate lateral connections between barrels while assuming that TC cells respond only to their PW and that TC cell axons are confined to their home barrel. We show that our model is capable of capturing a broad swath of experimental observations on multiwhisker receptive field dynamics within barrels, and we compare and contrast the dynamics of this model with model dynamics from prior work in which employ a similar general modeling strategy to examine possibility (i).

Presynaptic stochasticity improves energy efficiency and helps alleviate the stability-plasticity dilemma

When an action potential arrives at a synapse there is a large probability that no neurotransmitter is released. Surprisingly, simple computational models suggest that these synaptic failures enable information processing at lower metabolic costs. However, these models only consider information transmission at single synapses ignoring the remainder of the neural network as well as its overall computational goal. Here, we investigate how synaptic failures affect the energy efficiency of models of entire neural networks that solve a goal-driven task. We find that presynaptic stochasticity and plasticity improve energy efficiency and show that the network allocates most energy to a sparse subset of important synapses. We demonstrate that stabilising these synapses helps to alleviate the stability-plasticity dilemma, thus connecting a presynaptic notion of importance to a computational role in lifelong learning. Overall, our findings present a set of hypotheses for how presynaptic plasticity and stochasticity contribute to sparsity, energy efficiency and improved trade-offs in the stability-plasticity dilemma.

Deep Gated Hebbian Predictive Coding Accounts for Emergence of Complex Neural Response Properties Along the Visual Cortical Hierarchy

Predictive coding provides a computational paradigm for modeling perceptual processing as the construction of representations accounting for causes of sensory inputs. Here, we developed a scalable, deep network architecture for predictive coding that is trained using a gated Hebbian learning rule and mimics the feedforward and feedback connectivity of the cortex. After training on image datasets, the models formed latent representations in higher areas that allowed reconstruction of the original images. We analyzed low- and high-level properties such as orientation selectivity, object selectivity and sparseness of neuronal populations in the model. As reported experimentally, image selectivity increased systematically across ascending areas in the model hierarchy. Depending on the strength of regularization factors, sparseness also increased from lower to higher areas. The results suggest a rationale as to why experimental results on sparseness across the cortical hierarchy have been inconsistent. Finally, representations for different object classes became more distinguishable from lower to higher areas. Thus, deep neural networks trained using a gated Hebbian formulation of predictive coding can reproduce several properties associated with neuronal responses along the visual cortical hierarchy.