Input-output Function Research Articles

Greedy Multipoint Model-Order Reduction Technique for Fast Computation of Scattering Parameters of Electromagnetic Systems

This paper attempts to develop a new automated multipoint model-order reduction (MOR) technique, based on matching moments of the system input–output function, which would be suited for fast and accurate computation of scattering parameters for electromagnetic (EM) systems over a wide frequency band. To this end, two questions are addressed. Firstly, the cost of the wideband reduced model generation is optimized by automating a greedy multipoint MOR scheme. This is achieved by introducing a new dual local-global model convergence scheme, which applies fast and reliable a posteriori error estimates to check both local model convergence, used to select the number of moments at a single expansion point, and global model convergence, used to optimally select the expansion points. Secondly, the question of optimal convergence measure is addressed by proposing an enhanced a posteriori error estimator particularly suited for scattering parameter computations for lossy EM systems. The effectiveness and efficiency of the proposed automated scheme is verified through numerical simulations using reduced-order models for examples of a bandstop dielectric resonator filter and a dielectric resonator antenna for a wide frequency band, and compared against the results obtained using the full-order model, a reduced model generated with the optimal greedy point selection algorithm, as well as the reduced-order models obtained using the reduced basis method (RBM) and the single-point second-order Arnoldi method for passive order reduction (SAPOR) method.

Passive Synaptic Normalization and Input Synchrony-Dependent Amplification of Cortical Feedback in Thalamocortical Neuron Dendrites.

Thalamocortical neurons have thousands of synaptic connections from layer VI corticothalamic neurons distributed across their dendritic trees. Although corticothalamic synapses provide significant excitatory input, it remains unknown how different spatial and temporal input patterns are integrated by thalamocortical neurons. Using dendritic recording, 2-photon glutamate uncaging, and computational modeling, we investigated how rat dorsal lateral geniculate nucleus thalamocortical neurons integrate excitatory corticothalamic feedback. We find that unitary corticothalamic inputs produce small somatic EPSPs whose amplitudes are passively normalized and virtually independent of the site of origin within the dendritic tree. Furthermore, uncaging of MNI glutamate reveals that thalamocortical neurons have postsynaptic voltage-dependent mechanisms that can amplify integrated corticothalamic input. These mechanisms, involving NMDA receptors and T-type Ca2+ channels, require temporally synchronous synaptic activation but not spatially coincident input patterns. In hyperpolarized thalamocortical neurons, T-type Ca2+ channels produce nonlinear amplification of temporally synchronous inputs, whereas asynchronous inputs are not amplified. At depolarized potentials, the input–output function for synchronous synaptic input is linear but shows enhanced gain due to activity-dependent recruitment of NMDA receptors. Computer simulations reveal that EPSP amplification by T-type Ca2+ channels and NMDA receptors occurs when synaptic inputs are either clustered onto individual dendrites or when they are distributed throughout the dendritic tree. Consequently, postsynaptic EPSP amplification mechanisms limit the “modulatory” effects of corticothalamic synaptic inputs on thalamocortical neuron membrane potential and allow these synapses to act as synchrony-dependent “drivers” of thalamocortical neuron firing. These complex thalamocortical input–output transformations significantly increase the influence of corticothalamic feedback on sensory information transfer.SIGNIFICANCE STATEMENT Neurons in first-order thalamic nuclei transmit sensory information from the periphery to the cortex. However, the numerically dominant synaptic input to thalamocortical neurons comes from the cortex, which provides a strong, activity-dependent modulatory feedback influence on information flow through the thalamus. Here, we reveal how individual quantal-sized corticothalamic EPSPs propagate within thalamocortical neuron dendrites and how different spatial and temporal input patterns are integrated by these cells. We find that thalamocortical neurons have voltage- and synchrony-dependent postsynaptic mechanisms, involving NMDA receptors and T-type Ca2+ channels that allow nonlinear amplification of integrated corticothalamic EPSPs. These mechanisms significantly increase the responsiveness of thalamocortical neurons to cortical excitatory input and broaden the “modulatory” influence exerted by corticothalamic synapses.

Integrated inference and learning of neural factors in structural support vector machines

Tackling pattern recognition problems in areas such as computer vision, bioinformatics, speech or text recognition is often done best by taking into account task-specific statistical relations between output variables. In structured prediction, this internal structure is used to predict multiple outputs simultaneously, leading to more accurate and coherent predictions. Structural support vector machines (SSVMs) are nonprobabilistic models that optimize a joint input–output function through margin-based learning. Because SSVMs generally disregard the interplay between unary and interaction factors during the training phase, final parameters are suboptimal. Moreover, its factors are often restricted to linear combinations of input features, limiting its generalization power. To improve prediction accuracy, this paper proposes: (i) joint inference and learning by integration of back-propagation and loss-augmented inference in SSVM subgradient descent; (ii) extending SSVM factors to neural networks that form highly nonlinear functions of input features. Image segmentation benchmark results demonstrate improvements over conventional SSVM training methods in terms of accuracy, highlighting the feasibility of end-to-end SSVM training with neural factors.

The Effect of Stimulus Intensity and Carrier Frequency on Auditory Middle- and Long-Latency Evoked Potentials Using a Steady-State-Response Approach.

The purpose of this study was to measure magnitude changes of auditory steady-state responses (ASSRs) and respective transient middle- and long-latency responses as a function of stimulus intensity and carrier frequency. The literature lacks clear consensus, including relationship to loudness. A cohort of 48 adults with normal hearing was examined from a companion study (Tlumak, Durrant, & Delgado, 2015) on effects of aging. ASSRs were elicited by repeated tone-burst stimuli presented at rates of 40 and 0.75 Hz at 3 frequencies and 5 levels of stimulus intensity. The design also permitted scrutiny of any gender bias to the results. Similar to derived transient response findings, ASSR magnitude (harmonic sum) systematically increased with intensity. Input-output function only at 0.75 Hz approximated a log-log linear function. However, slopes fell well below that of doubling of loudness per 10 dB SPL. Results failed to demonstrate significance as a function of carrier frequency and gender for both repetition rates. Effects of stimulus intensity, carrier frequency, and gender on ASSRs were similar to those of their transient counterparts. Findings remain disappointing for objective loudness estimation. Results suggest only a clear linkage to the long-latency response and the 0.75-Hz magnitude but require careful consideration of limitations/underlying mechanisms when measuring loudness-related effects.

Weighing method for a truck scale based on optional neural network with penalty function

This paper presents a weighing method for a truck scale based on a neural network (NN) with penalty function (PFNN). Firstly, the truck scale’s prior knowledge, i.e. the positive partial derivatives of the truck scale’s input–output function and the distributions of truck scale’s permissible weighing errors, is used to construct the constraints for optimizing the NN. Then, the penalty function is applied to construct the new NN’s performance index, and the detailed algorithm for training this NN is given. Finally, the method for assigning the value of the penalty factors is also discussed. The comparative experimental results show that the PFNN’s generalization ability is better than that of a data-induced NN (DINN) especially with a lack of training samples (the DINN is a method for training an NN only by using the data samples, not prior knowledge), and the weighing errors of the truck scale with PFNN are far less than those of DINN. In addition, the convergence of the PFNN is faster than that of the DINN.

Pathway-Specific Striatal Substrates for Habitual Behavior

The dorsolateral striatum (DLS) is implicated in habit formation. However, the DLS circuit mechanisms underlying habit remain unclear. A key role for DLS is to transform sensorimotor cortical input into firing of output neurons that project to the mutually antagonistic direct and indirect basal ganglia pathways. Here we examine whether habit alters this input-output function. By imaging cortically evoked firing in large populations of pathway-defined striatal projection neurons (SPNs), we identify features that strongly correlate with habitual behavior on a subject-by-subject basis. Habitual behavior correlated with strengthened DLS output to both pathways as well as a tendency for action-promoting direct pathway SPNs to fire before indirect pathway SPNs. In contrast, habit suppression correlated solely with a weakened direct pathway output. Surprisingly, all effects were broadly distributed in space. Together, these findings indicate that the striatum imposes broad, pathway-specific modulations of incoming activity to render learned motor behaviors habitual.

A restatement of the optimum synthesis of function generators with planar four-bar and slider-crank mechanisms examples

The problem of optimum synthesising of a mechanism to best approximate a function, while simultaneously ensuring good motion transmission characteristics is discussed. Distinction is made between the design variables that determine the shape of the input-output (I/O) function of the mechanism, and the design variables that affect the degree of overlap between this I/O function and the function to be mechanised, through scaling, mirroring and rotations in 90o increments. Examples are given of designing the planar four-bar and slider-crank linkages of a logarithmic scale, and that of a tangent-function generator. These are performed on modified mechanisms with an added degree-of-freedom, which substantially simplify the synthesis problem.

An island particle algorithm for rare event analysis

Estimating rare event probability with accuracy is of great interest for safety and reliability applications. In this paper, we focus on rare events which can be modeled by a threshold exceedance of a deterministic input–output function with random inputs. Some parameters of this function or density parameters of input random variables may be fixed by an experimenter for simplicity reasons. From a risk analysis point of view, it is not only interesting to evaluate the probability of a critical event but it is also important to determine the impact of such tuning of parameters on the realization of a critical event, because a bad estimation of these parameters can strongly modify rare event probability estimations. In the present paper, we present an example of island particle algorithm referred to as sequential Monte Carlo square (SMC2). This algorithm gives an estimate of the law of random phenomena that leads to critical events. The principles of this statistical technique are described throughout this article and its results are analysed on different realistic aerospace test cases.

The Synthesis of the Axodes of RCCC Linkages

As the coupler link of an RCCC linkage moves, its instant screw axis (ISA) sweeps a ruled surface on the fixed link; by the same token, the ISA describes on the coupler link itself a corresponding ruled surface. These two surfaces are the axodes of the linkage, which roll while sliding and maintaining line contact. The axodes not only help to visualize the motion undergone by the coupler link but also can be machined as spatial cams and replace the four-bar linkage, if the need arises. Reported in this paper is a procedure that allows the synthesis of the axodes of an RCCC linkage. The synthesis of this linkage, in turn, is based on dual algebra and the principle of transference, as applied to a spherical four-bar linkage with the same input–output function as the angular variables of the RCCC linkage. Examples of RCCC linkages are included. Moreover, to illustrate the generality of the synthesis procedure, it is also applied to a spherical linkage, namely, the Hooke joint, and to the Bennett linkage.

Transition to Chaos in Random Neuronal Networks

Firing patterns in the central nervous system often exhibit strong temporal irregularity and heterogeneity in their time averaged response properties. Previous studies suggested that these properties are outcome of an intrinsic chaotic dynamics. Indeed, simplified rate-based large neuronal networks with random synaptic connections are known to exhibit sharp transition from fixed point to chaotic dynamics when the synaptic gain is increased. However, the existence of a similar transition in neuronal circuit models with more realistic architectures and firing dynamics has not been established. In this work we investigate rate based dynamics of neuronal circuits composed of several subpopulations and random connectivity. Nonzero connections are either positive-for excitatory neurons, or negative for inhibitory ones, while single neuron output is strictly positive; in line with known constraints in many biological systems. Using Dynamic Mean Field Theory, we find the phase diagram depicting the regimes of stable fixed point, unstable dynamic and chaotic rate fluctuations. We characterize the properties of systems near the chaotic transition and show that dilute excitatory-inhibitory architectures exhibit the same onset to chaos as a network with Gaussian connectivity. Interestingly, the critical properties near transition depend on the shape of the single- neuron input-output transfer function near firing threshold. Finally, we investigate network models with spiking dynamics. When synaptic time constants are slow relative to the mean inverse firing rates, the network undergoes a sharp transition from fast spiking fluctuations and static firing rates to a state with slow chaotic rate fluctuations. When the synaptic time constants are finite, the transition becomes smooth and obeys scaling properties, similar to crossover phenomena in statistical mechanics

Nonlinear Error Compensation for Load Cells Based on the Optimal Neural Network With an Augmented Lagrange Multiplier

This paper presents a new approach for compensating nonlinear errors of a load cell, based on the optimal neural network with an augmented Lagrange multiplier (ALMNN). The load cell has serious nonlinear errors, which lower the accuracy of the weighing results. In this proposed approach, first, we construct the constraints for training the neural network (NN) using the load cell’s prior knowledge, i.e., the monotonic increasing property of a load cell’s input-output function. Second, we create the augmented Lagrange function with a multiplier and a penalty factor to serve as an NN’s performance index, and then deduce the ALMNN’s detailed training algorithm. The penalty factor plays an important role in ALMNN, thus the reasonable range of numerical value is discussed in this paper. Meanwhile, the weighing principle of a strain gauge load cell is introduced and its nonlinear error model is established. Finally, the experimental results demonstrate the effectiveness of ALMNN. By comparing with the conventional data induction method (DINN, i.e., the training of NN not with the prior knowledge but only the data samples), ALMNN significantly improves the NN’s generalization ability, especially in the case of lacking training samples. In addition, the applicability of ALMNN is verified by three additional experiments.

High frequency bone conduction auditory evoked potentials in the guinea pig: Assessing cochlear injury after ossicular chain manipulation

Permanent high frequency (>4 kHz) sensorineural hearing loss following middle ear surgery occurs in up to 25% of patients. The aetiology of this loss is poorly understood and may involve transmission of supra-physiological forces down the ossicular chain to the cochlea. Investigating the mechanisms of this injury using animal models is challenging, as evaluating cochlear function with evoked potentials is confounded when ossicular manipulation disrupts the normal air conduction (AC) pathway. Bone conduction (BC) using clinical bone vibrators in small animals is limited by poor transducer output at high frequencies sensitive to trauma. The objectives of the present study were firstly to evaluate a novel high frequency bone conduction transducer with evoked auditory potentials in a guinea pig model, and secondly to use this model to investigate the impact of middle ear surgical manipulation on cochlear function. We modified a magnetostrictive device as a high frequency BC transducer and evaluated its performance by comparison with a calibrated AC transducer at frequencies up to 32 kHz using the auditory brainstem response (ABR), compound action potential (CAP) and summating potential (SP). To mimic a middle ear traumatising stimulus, a rotating bur was brought in to contact with the incudomalleal complex and the effect on evoked cochlear potentials was observed. BC-evoked potentials followed the same input–output function pattern as AC potentials for all ABR frequencies. Deterioration in CAP and SP thresholds was observed after ossicular manipulation. It is possible to use high frequency BC to evoke responses from the injury sensitive basal region of the cochlea and so not rely on AC with the potential confounder of conductive hearing loss. Ongoing research explores how these findings evolve over time, and ways in which injury may be reduced and the cochlea protected during middle ear surgery.This article is part of a Special Issue entitled <IEB Kyoto>.

Gain control through divisive inhibition prevents abrupt transition to chaos in a neural mass model.

Experimental results suggest that there are two distinct mechanisms of inhibition in cortical neuronal networks: subtractive and divisive inhibition. They modulate the input-output function of their target neurons either by increasing the input that is needed to reach maximum output or by reducing the gain and the value of maximum output itself, respectively. However, the role of these mechanisms on the dynamics of the network is poorly understood. We introduce a novel population model and numerically investigate the influence of divisive inhibition on network dynamics. Specifically, we focus on the transitions from a state of regular oscillations to a state of chaotic dynamics via period-doubling bifurcations. The model with divisive inhibition exhibits a universal transition rate to chaos (Feigenbaum behavior). In contrast, in an equivalent model without divisive inhibition, transition rates to chaos are not bounded by the universal constant (non-Feigenbaum behavior). This non-Feigenbaum behavior, when only subtractive inhibition is present, is linked to the interaction of bifurcation curves in the parameter space. Indeed, searching the parameter space showed that such interactions are impossible when divisive inhibition is included. Therefore, divisive inhibition prevents non-Feigenbaum behavior and, consequently, any abrupt transition to chaos. The results suggest that the divisive inhibition in neuronal networks could play a crucial role in keeping the states of order and chaos well separated and in preventing the onset of pathological neural dynamics.

The domain of neuronal firing on a plane of input current and conductance.

The activation of neurotransmitter receptors increases the current flow and membrane conductance and thus controls the firing rate of a neuron. In the present work, we justified the two-dimensional representation of a neuronal input by voltage-independent current and conductance and obtained experimentally and numerically a complete input-output (I/O) function. The dependence of the steady-state firing rate on the input current and conductance was studied as a two-parameter I/O function. We employed the dynamic patch clamp technique in slices to get this dependence for the whole domain of two input signals that evoke stationary spike trains in a single neuron (Ω-domain). As found, the Ω-domain is finite and an additional conductance decreases the range of spike-evoking currents. The I/O function has been reproduced in a Hodgkin-Huxley-like model. Among the simulated effects of different factors on the I/O function, including passive and active membrane properties, external conditions and input signal properties, the most interesting were: the shift of the right boundary of the Ω-domain (corresponding to the exCitation block) leftwards due to the decrease of the maximal potassium conductance; and the reduction of the Ω-domain by the decrease of the maximal sodium concentration. As found in experiments and simulations, the Ω-domain is reduced by the decrease of extracellular sodium concentration, by cooling, and by adding slow potassium currents providing interspike interval adaptation; the Ω-domain height is increased by adding color noise. Our modeling data provided a generalization of I/O dependencies that is consistent with previous studies and our experiments. Our results suggest that both current flow and membrane conductance should be taken into account when determining neuronal firing activity.

Computational Design of Stephenson II Six-Bar Function Generators for 11 Accuracy Points

This paper presents a design methodology for Stephenson II six-bar function generators that coordinate 11 input and output angles. A complex number formulation of the loop equations yields 70 quadratic equations in 70 unknowns, which is reduced to a system of ten eighth degree polynomial equations of total degree 810=1.07×109. These equations have a monomial structure which yields a multihomogeneous degree of 264,241,152. A sequence of polynomial homotopies was used to solve these equations and obtain 1,521,037 nonsingular solutions. Contained in these solutions are linkage design candidates which are evaluated to identify cognates, and then analyzed to determine their input–output angles in each assembly. The result is a set of feasible linkage designs that reach the required accuracy points in a single assembly. As an example, three Stephenson II function generators are designed, which provide the input–output functions for the hip, knee, and ankle of a humanoid walking gait.

GABAA receptor-mediated feedforward and feedback inhibition differentially modulate the gain and the neural code transformation in hippocampal CA1 pyramidal cells

Diverse variety of hippocampal interneurons exists in the CA1 area, which provides either feedforward (FF) or feedback (FB) inhibition to CA1 pyramidal cell (PC). However, how the two different inhibitory network architectures modulate the computational mode of CA1 PC is unknown. By investigating the CA3 PC rate-driven input–output function of CA1 PC using in vitro electrophysiology, in vitro-simulation of inhibitory network, and in silico computational modeling, we demonstrated for the first time that GABAA receptor-mediated FF and FB inhibition differentially modulate the gain, the spike precision, the neural code transformation and the information capacity of CA1 PC. Recruitment of FF inhibition buffered the CA1 PC spikes to theta-frequency regardless of the input frequency, abolishing the gain and making CA1 PC insensitive to its inputs. Instead, temporal variability of the CA1 PC spikes was increased, promoting the rate-to-temporal code transformation to enhance the information capacity of CA1 PC. In contrast, the recruitment of FB inhibition sub-linearly transformed the input rate to spike output rate with high gain and low spike temporal variability, promoting the rate-to-rate code transformation. These results suggest that GABAA receptor-mediated FF and FB inhibitory circuits could serve as network mechanisms for differentially modulating the gain of CA1 PC, allowing CA1 PC to switch between different computational modes using rate and temporal codes ad hoc. Such switch will allow CA1 PC to efficiently respond to spatio-temporally dynamic inputs and expand its computational capacity during different behavioral and neuromodulatory states in vivo.

Inhibition shapes acoustic responsiveness in spherical bushy cells.

Signal processing in the auditory brainstem is based on an interaction of neuronal excitation and inhibition. To date, we have incomplete knowledge of how the dynamic interplay of both contributes to the processing power and temporal characteristics of signal coding. The spherical bushy cells (SBCs) of the anteroventral cochlear nucleus (AVCN) receive their primary excitatory input through auditory nerve fibers via large, axosomatic synaptic terminals called the endbulbs of Held and by additional, acoustically driven inhibitory inputs. SBCs provide the input to downstream nuclei of the brainstem sound source localization circuitry, such as the medial and lateral superior olive, which rely on temporal precise inputs. In this study, we used juxtacellular recordings in anesthetized Mongolian gerbils to assess the effect of acoustically evoked inhibition on the SBCs input-output function and on temporal precision of SBC spiking. Acoustically evoked inhibition proved to be strong enough to suppress action potentials (APs) of SBCs in a stimulus-dependent manner. Inhibition shows slow onset and offset dynamics and increasing strength at higher sound intensities. In addition, inhibition decreases the rising slope of the EPSP and prolongs the EPSP-to-AP transition time. Both effects can be mimicked by iontophoretic application of glycine. Inhibition also improves phase locking of SBC APs to low-frequency tones by acting as a gain control to suppress poorly timed EPSPs from generating postsynaptic APs to maintain precise SBC spiking across sound intensities. The present data suggest that inhibition substantially contributes to the processing power of second-order neurons in the ascending auditory system.

Modelling the effects of cell-to-cell variability on the output of interconnected gene networks in bacterial populations.

BackgroundThe interconnection of quantitatively characterized biological devices may lead to composite systems with apparently unpredictable behaviour. Context-dependent variability of biological parts has been investigated in several studies, measuring its entity and identifying the factors contributing to variability. Such studies rely on the experimental analysis of model systems, by quantifying reporter genes via population or single-cell approaches. However, cell-to-cell variability is not commonly included in predictability analyses, thus relying on predictive models trained and tested on central tendency values. This work aims to study in silico the effects of cell-to-cell variability on the population-averaged output of interconnected biological circuits.MethodsThe steady-state deterministic transfer function of individual devices was described by Hill equations and lognormal synthetic noise was applied to their output. Two- and three-module networks were studied, where individual devices implemented inducible/repressible functions. The single-cell output of such networks was simulated as a function of noise entity; their population-averaged output was computed and used to investigate the expected variability in transfer function identification. The study was extended by testing different noise models, module logic, intrinsic/extrinsic noise proportions and network configurations.ResultsFirst, the transfer function of an individual module was identified from simulated data of a two-module network. The estimated parameter variability among different noise entities was limited (14%), while a larger difference was observed (up to 62%) when estimated and true parameters were compared. Thus, low-variability parameter estimates can be obtained for different noise entities, although deviating from the true parameters, whose measurement requires noise knowledge. Second, the black-box input-output function of a two/three-module network was predicted from the knowledge of the transfer function of individual modules, identified in the presence of noise. Estimates variability was low (16%); however, differences up to 68% were observed by simulating a typical experimental study where the predictions obtained above were compared to network outputs generated in the presence of noise. Network predictions can, thus, deviate from real outputs when modules are characterized and re-used in different noise contexts.ConclusionsThe adopted approach can support predictability studies in synthetic biology by distinguishing between actual unpredictability and contribution of noise and by guiding researchers in the design of suitable experimental measurement for gene networks.

Activity-Dependent Transmission and Integration Control the Timescales of Auditory Processing at an Inhibitory Synapse

To capture the context of sensory information, neural networks must process input signals across multiple timescales. In the auditory system, a prominent change in temporal processing takes place at an inhibitory GABAergic synapse in the dorsal nucleus of the lateral lemniscus (DNLL). At this synapse, inhibition outlasts the stimulus by tens of milliseconds, such that it suppresses responses to lagging sounds, and is therefore implicated in echo suppression. Here, we untangle the cellular basis of this inhibition. We demonstrate with invivo whole-cell patch-clamp recordings in Mongolian gerbils that the duration of inhibition increases with sound intensity. Activity-dependent spillover and asynchronous release translate the high presynaptic firing rates found invivo into a prolonged synaptic output in acute slice recordings. A key mechanism controlling the inhibitory time course is the passive integration of the hyperpolarizing inhibitory conductance. This prolongation depends on the synaptic conductance amplitude. Computational modeling shows that this prolongation is a general mechanism and relies on a non-linear effect caused by synaptic conductance saturation when approaching the GABA reversal potential. The resulting hyperpolarization generates an efficient activity-dependent suppression of action potentials without affecting the threshold or gain of the input-output function. Taken together, the GABAergic inhibition in the DNLL is adjusted to the physiologically relevant duration by passive integration of inhibition with activity-dependent synaptic kinetics. This change in processing timescale combined with the reciprocal connectivity between the DNLLs implements a mechanism to suppress the distracting localization cues of echoes and helps to localize the initial sound source reliably.

The adhesion GPCR latrophilin/CIRL shapes mechanosensation.

G-protein-coupled receptors (GPCRs) are typically regarded as chemosensors that control cellular states in response to soluble extracellular cues. However, the modality of stimuli recognized through adhesion GPCR (aGPCR), the second largest class of the GPCR superfamily, is unresolved. Our study characterizes the Drosophila aGPCR Latrophilin/dCirl, a prototype member of this enigmatic receptor class. We show that dCirl shapes the perception of tactile, proprioceptive, and auditory stimuli through chordotonal neurons, the principal mechanosensors of Drosophila. dCirl sensitizes these neurons for the detection of mechanical stimulation by amplifying their input-output function. Our results indicate that aGPCR may generally process and modulate the perception of mechanical signals, linking these important stimuli to the sensory canon of the GPCR superfamily.