How neurons may compute: the case of insect sexual pheromone discrimination

Rapid processing of chemosensor transients in a neuromorphic implementation of the insect macroglomerular complex

We present a biologically-constrained neuromorphic spiking model of the insect antennal lobe macroglomerular complex that encodes concentration ratios of chemical components existing within a blend, implemented using a set of programmable logic neuronal modeling cores. Depending upon the level of inhibition and symmetry in its inhibitory connections, the model exhibits two dynamical regimes: fixed point attractor (winner-takes-all type), and limit cycle attractor (winnerless competition type) dynamics. We show that, when driven by chemosensor input in real-time, the dynamical trajectories of the model's projection neuron population activity accurately encode the concentration ratios of binary odor mixtures in both dynamical regimes. By deploying spike timing-dependent plasticity in a subset of the synapses in the model, we demonstrate that a Hebbian-like associative learning rule is able to organize weights into a stable configuration after exposure to a randomized training set comprising a variety of input ratios. Examining the resulting local interneuron weights in the model shows that each inhibitory neuron competes to represent possible ratios across the population, forming a ratiometric representation via mutual inhibition. After training the resulting dynamical trajectories of the projection neuron population activity show amplification and better separation in their response to inputs of different ratios. Finally, we demonstrate that by using limit cycle attractor dynamics, it is possible to recover and classify blend ratio information from the early transient phases of chemosensor responses in real-time more rapidly and accurately compared to a nearest-neighbor classifier applied to the normalized chemosensor data. Our results demonstrate the potential of biologically-constrained neuromorphic spiking models in achieving rapid and efficient classification of early phase chemosensor array transients with execution times well beyond biological timescales.

Interaction of cellular and network mechanisms for efficient pheromone coding in moths

Sensory systems, both in the living and in machines, have to be optimized with respect to their environmental conditions. The pheromone subsystem of the olfactory system of moths is a particularly well-defined example in which rapid variations of odor content in turbulent plumes require fast, concentration-invariant neural representations. It is not clear how cellular and network mechanisms in the moth antennal lobe contribute to coding efficiency. Using computational modeling, we show that intrinsic potassium currents (I(A) and I(SK)) in projection neurons may combine with extrinsic inhibition from local interneurons to implement a dual latency code for both pheromone identity and intensity. The mean latency reflects stimulus intensity, whereas latency differences carry concentration-invariant information about stimulus identity. In accordance with physiological results, the projection neurons exhibit a multiphasic response of inhibition-excitation-inhibition. Together with synaptic inhibition, intrinsic currents I(A) and I(SK) account for the first and second inhibitory phases and contribute to a rapid encoding of pheromone information. The first inhibition plays the role of a reset to limit variability in the time to first spike. The second inhibition prevents responses of excessive duration to allow tracking of intermittent stimuli.

Competition-based model of pheromone component ratio detection in the moth

For some moth species, especially those closely interrelated and sympatric, recognizing a specific pheromone component concentration ratio is essential for males to successfully locate conspecific females. We propose and determine the properties of a minimalist competition-based feed-forward neuronal model capable of detecting a certain ratio of pheromone components independently of overall concentration. This model represents an elementary recognition unit for the ratio of binary mixtures which we propose is entirely contained in the macroglomerular complex (MGC) of the male moth. A set of such units, along with projection neurons (PNs), can provide the input to higher brain centres. We found that (1) accuracy is mainly achieved by maintaining a certain ratio of connection strengths between olfactory receptor neurons (ORN) and local neurons (LN), much less by properties of the interconnections between the competing LNs proper. An exception to this rule is that it is beneficial if connections between generalist LNs (i.e. excited by either pheromone component) and specialist LNs (i.e. excited by one component only) have the same strength as the reciprocal specialist to generalist connections. (2) successful ratio recognition is achieved using latency-to-first-spike in the LN populations which, in contrast to expectations with a population rate code, leads to a broadening of responses for higher overall concentrations consistent with experimental observations. (3) when longer durations of the competition between LNs were observed it did not lead to higher recognition accuracy.

Blocking sensory inputs to identified antennal glomeruli selectively modifies odorant perception in Drosophila

Neural coding of sensory input is a major unsolved issue in neuroscience. Current experimental methods rely on neural activity recording or visualization following sensory stimulation. Most of them, however, do not include behavioral correlates on the actual perception by the animal. We present a novel approach to address olfaction and coding in adult Drosophila. Sensory input was selectively blocked in two subsets of sensory neurons that project to different, albeit overlapping, groups of central targets, by means of tetanus toxin expressed under the control of the yeast transcription factor Gal4. Glomeruli DL1, DL2, VM1, and VM4 were tested following stimulation with benzaldehyde, ethyl acetate, propionic acid, butanol, or acetone at various concentrations. The behavioral response was found to be modified in an odorant-specific and a concentration-dependent manner. Sensory input to DL2 and, to a minor extent, VM1 and/or VM4, appear to be required for benzaldehyde perception, while acetone is processed through DL1. None of these glomeruli, however, seem necessary for butanol perception. In addition, sexual differences were observed for some stimuli. These results demonstrate the behavioral relevance of odor representation as maps of glomerular activity generated in the antennal lobes following specific sensory input. The strategy used here should be useful to characterize olfactory coding, as new and selective Gal4 lines become available.

Genes, neurons and codes: remarks on biological communication

I examine critically the application of information-theoretic ideas to biological communication during embryonic development and in the functioning central nervous system (CNS). I show that intercellular communication relies mostly on simple signals whose role is to effect a selection among predetermined cellular states. Hence, a crucial role is played by cellular memory, which stabilizes such states. Memory in cells is partly located in the nuclear DNA; no less important however is (phenotypic) memory lying in the cell's organelles and compartments. Because of combinatorial effects in gene expression patterns, cell memory is an enormously powerful mechanism, which also underlies plasticity, and thus constitutes the factor unifying genetic determination, plasticity and learning. Communication in the CNS is analyzed in some detail: here, cellular memory is embodied in anatomy (i.e., cell shape) and neurochemistry. These are the major, relatively "static" factors affecting the routing of neural impulses in the adult CNS. In addition, however, faster channeling is also required: action potentials must be directed to their targets along a few of exponentially many paths, and this dynamic routing is crucial for proper operation. I suggest that collective oscillatory modes may play a role in solving this addressing problem, in the same way that the clock signal gates the operation of man-made computers.

Morphometric modeling of olfactory circuits in the insect antennal lobe: I. Simulations of spiking local interneurons

Inhibitory local interneurons (LNs) play a critical role in shaping the output of olfactory glomeruli in both the olfactory bulb of vertebrates and the antennal lobe of insects and other invertebrates. In order to examine how the complex geometry of LNs may affect signaling in the antennal lobe, we constructed detailed multi-compartmental models of single LNs from the sphinx moth, Manduca sexta, using morphometric data from confocal-microscopic images. Simulations clearly revealed a directionality in LNs that impeded the propagation of injected currents from the sub-micron-diameter glomerular dendrites toward the much larger-diameter integrating segment (IS) in the coarse neuropil. Furthermore, the addition of randomly-firing synapses distributed across the LN dendrites (simulating the noisy baseline activity of afferent input recorded from LNs in the odor-free state) led to a significant depolarization of the LN. Thus the background activity typically recorded from LNs in vivo could influence synaptic integration and spike transformation in LNs through voltage-dependent mechanisms. Other model manipulations showed that active currents inserted into the IS can help synchronize the activation of inhibitory synapses in glomeruli across the antennal lobe. These data, therefore, support experimental findings suggesting that spiking inhibitory LNs can operate as multifunctional units under different ambient odor conditions. At low odor intensities, (i.e. subthreshold for IS spiking), they participate in local, mostly intra-glomerular processing. When activated by elevated odor concentrations, however, the same neurons will fire overshooting action potentials, resulting in the spread of inhibition more globally across the antennal lobe. Modulation of the passive and active properties of LNs may, therefore, be a deciding factor in defining the multi-glomerular representations of odors in the brain.

Coincident stimulation with pheromone components improves temporal pattern resolution in central olfactory neurons

Male moths must detect and resolve temporal discontinuities in the sex pheromonal odor signal emitted by a conspecific female moth to orient to and locate the odor source. We asked how sensory information about two key components of the pheromone influences the ability of certain sexually dimorphic projection (output) neurons in the primary olfactory center of the male moth's brain to encode the frequency and duration of discrete pulses of pheromone blends. Most of the male-specific projection neurons examined gave mixed postsynaptic responses, consisting of an early suppressive phase followed by activation of firing, to stimulation of the ipsilateral antenna with a blend of the two behaviorally essential pheromone components. Of 39 neurons tested, 33 were excited by the principal (most abundant) pheromone component but inhibited by another, less abundant but nevertheless essential component of the blend. We tested the ability of each neuron to encode intermittent pheromonal stimuli by delivering trains of 50-ms pulses of the two-component blend at progressively higher rates from 1 to 10 per second. There was a strong correlation between 1) the amplitude of the early inhibitory post-synaptic potential evoked by the second pheromone component and 2) the maximal rate of odor pulses that neuron could resolve (r = 0.92). Projection neurons receiving stronger inhibitory input encoded the temporal pattern of the stimulus with higher fidelity. With the principal, excitatory component of the pheromone alone as the stimulus, the dynamic range for encoding stimulus intermittency was reduced in nearly 60% of the neurons tested. The greatest reductions were observed in those neurons that could be shown to receive the strongest inhibitory input from the second behaviorally essential component of the blend. We also tested the ability of these neurons to encode stimulus duration. Again there was a strong correlation between the strength of the inhibitory input to a neuron mediated by the second pheromone component and that neuron's ability to encode stimulus duration. Neurons that were strongly inhibited by the second component could accurately encode pulses of the blend from 50 to 500 ms in duration (r = 0.94), but that ability was reduced in neurons receiving little or no inhibitory input (r = 0.23). This study confirms that certain olfactory projection neurons respond optimally to a particular odor blend rather than to the individual components of the blend. The key components activate opposing synaptic inputs that enable this subset of central neurons to copy the duration and frequency of intermittent odor pulses that are a fundamental feature of airborne olfactory stimuli.

Investigation of the role of interneurons and their modulation by centrifugal fibers in a neural model of the olfactory bulb

Olfactory bulb processing results from the interaction of relay neurons with two main categories of interneurons which mediate inhibition in two distinct layers: periglomerular cells and granule cells. We present here a neural model of the mammalian olfactory bulb which allows to separately investigate the functional consequences of the two types of interneurons onto the relay neurons responsiveness to odors. The model, although built with simplified representations of neural elements generates various aspects of neural dynamics from the cellular to the populational level. We propose that the combined action of centrifugal control at two different layers of processing is complementary: reduction of the number of active relay neurons responding to a given odorant through increased activity of periglomerular cells, and an increase of response intensity of active mitral cells through decrease of granule cell inhibition.