Predictable and unpredictable coding of odorant mixtures by olfactory receptor neurons (Prof. Jean-Pierre Rospars)

The response of an olfactory receptor neuron (ORN) to an odourant mixture is not a simple function of the responses to the individual components of the mixture, a phenomenon called mixture interaction. First, we will present a biochemical model of pure competitive interaction of odourants with olfactory receptors. In this model the ORN response to the mixture is predictable from the ORN responses to the components. Second, based on neurophysiological experiments in the rat, we will show that not all mixtures obey the competitive model. We will discuss the possible mechanisms leading to these unpredictable responses and stress their importance in the coding of natural odours

 

CMOS technology for highly integrated bioelectronic and chemo/biosensor microsystems (Prof. Andreas Hierlemann)

Microfabrication techniques and, in particular, CMOS technology have been widely used to devise chemo/biosensors as well as bioelectronic and sensor microsystems in a generic approach. Examples of micromachined bio/chemosensors will be shown.  Then, the electrical interfacing of CMOS microelectronics with biological entities or electrogenic cells, i.e., cells that react upon electrical stimulation and, in turn, produce electrical signals (heart cells or neurons) will be shown and explained.

CMOS-based, fully integrated microelectrode arrays for bidirectional communication (stimulation and recording) with electrogenic cells will be presented. These devices are capable of monitoring relevant electrophysiological responses of cells to electrical stimuli or to pharmacological agents with prospective applications in the field of bio-inspired information processing or pharmascreening.

 

Functional modularization in the odor maps of the olfactory bulb (Kensaku Mori)

Because individual glomeruli in the mammalian olfactory bulb represent a single odorant receptor, the glomerular sheet forms odorant receptor maps (or odor maps). I will summarize the emerging view of the zonal, domain and cluster organization of the odor maps and their functional significance in terms of behavioral responses.

 

Olfactory Coding in Insects (prof. Giovanni Galizia)

We analyze how insects percieve and process olfactory information. The first part will be devoted to the properties of biological olfactory receptors, i.e. their sensitivity, selectivity, and temporal response properties. The second part will look at how neural networks in the insect brain process these signals in order to maximize information, in particular odor identification.

 

A system-wide model of the olfactory pathway for chemosensor arrays (prof. Ricardo Gutierrez-Osuna)

 In this talk, I will describe a computational model for chemical sensor arrays inspired by information processing in the olfactory system.  First, I will present a model of sensory convergence that leads to spatial representations consistent with those observed in the olfactory bulb.  Next, I will describe models of lateral inhibition in the olfactory bulb that provide concentration normalization and contrast enhancement of odor patterns.  Finally, I will propose a model of bulb-cortex interactions that can be used to perform odor segmentation and background suppression.  Our models are validated on experimental data from temperature-modulated metal-oxide sensor, optical microbead arrays, and infrared absorption spectroscopy.

 

Learning pattern classification from the olfactory system of insects (prof. Thomas Nowotny)

Animals have an outstanding ability to perceive and recognize odors in a complex space of chemical stimulants. In my presentation I will review our recent work on the olfactory system of insects and its implications for the construction of bio-inspired classification systems. As a real-world example I will discuss the classification of handwritten digits from the MNIST database and point out the many links to original ideas of Rosenblatt (3 layer Rosenblatt Perceptron) and Cover / Cortes and Vapnik (non-linear expansion in Support Vector Machines). The resulting general classification system is highly suitable for massive parallel implementation in custom chips and therefore, contrary to Rosenblatt's original concerns, very practical for technical applications.

 

The computational architecture of biological olfaction (Thomas A. Cleland)

Cellular and neural network properties in the olfactory epithelium and olfactory bulb help resolve many difficult problems inherent to chemical detection and identification.  Some of these biophysical properties provide novel insights and mechanisms for artificial chemosensory systems to emulate, whereas others simply reflect the peculiarities of the biological substrate and have little to contribute to artificial systems design.  I will present a theoretical and computational overview of early olfactory processing, emphasizing particular features relevant to the design and construction of artificial systems.

 

Olfaction Targeted (Peter Mombaerts)

The main olfactory system of the mouse is a mosaic of 2000 populations of olfactory sensory neurons. Each population expresses one allele of one of the 1000 intact odorant receptor genes. Odorant receptors determine both the odorant response profile of the neuron and the projection of its axon. My laboratory focuses on genetic approaches to odorant receptor gene choice and to the axonal wiring problem.

 

NEUROCHEM: Biologically Inspired Computation for Chemical Sensing (Santiago Marco)

NEUROCHEM is an Fp7 FET Bio-ICT Convergence STREP project coordinated by UB and involving eight additional european partners. The main project objective is to  develop novel computing paradigms and biomimetic artefacts for chemical sensing taking inspiration from the biological olfactory pathway. This project proposes to build computational models of its main building blocks: olfactory receptor layer, olfactory bulb, and olfactory cortex. Additionally, the project will integrate a large scale sensor array in resistive polymer technology.

 

Is concentration represented by means of a labeled line code in early olfactory processing? (prof. Anders Lansner)

Population codes and labeled line codes are commonly observed in sensory and motor representations in the central nervous system. These are typically seen as extending the dynamic range of the population beyond that of the individual receptor neuron. The concentration-frequency response properties of olfactory receptor neurons in the olfactory epithelium that express the same olfactory receptor appear to produce a population code of odor intensity. We examine by means of a computational model the hypothesis that the olfactory bulb converts this population code into a labeled line code for odor intensity.

 

A possible platform for artficial olfaction? (Ingemar Lundstrom)

In this talk we discuss how arbitrarily shaped spots of different metallo porphyrins can be used to form an experimental system, which surprisingly well mimics some of the features of the mammalian olfactory system. These features are obtained both through the layout of the sample itself and the use of an imaging optical method to elaborate the response caused by different odours. Each pixel in the image is treated as an olfactory neuron and each spot as a glomerulus. It was found that spots of the seven different porphyrins used were best described by twelve glomeruli. The improvement in signal to noise ratio going from the neuron to the glomeruli level was observed as expected. Interestingly even an image with rather low resolution contains thousands of “neurons” (pixels). Furthermore the geometry used for the exposure to odours provide spatial differences between different glomeruli, giving rise to spatio-temporal features resembling those found in the olfactory system. Since the experimental system needs no patterning or delicate fabrication step and is evaluated with a computer screen and a web camera it could be an interesting platform for artificial olfaction.

 

The lipocalin family, odorant binding proteins and mouse urinary proteins: potential biomimetic sensing systems (prof. Krishna C. Persaud)

The lipocalins are a family of functionally diverse, small proteins that comprise 160–180 amino acid residues. They have important biological functions from bacteria to humans. The β-barrel structural element of the lipocalins represents a rigid folding unit. The backbone conformation of the β-barrel is highly conserved throughout the lipocalins. This β-barrel structure can support loops with highly variable lengths, sequences and conformations at its open end. This is analogous to the mode in which antibodies present their six hypervariable loops (complementarity-determining regions) on top of a structurally conserved framework. However, compared with antibodies, lipocalins exhibit several biotechnological advantages because they are smaller in size, are composed of a single polypeptide chain and they exhibit a simpler set of four hypervariable loops that can be more easily manipulated at the genetic level. This  architecture appears to be well-suited to the implementation of novel binding activities via combinatorial protein design. It was shown that using combinatorial techniques that “anticalins” could be produced that had different binding activities to various ligands (Beste et al. 1999, Skerra 2001, Schlehubera S & Skerra 2005).

One lipocalin is termed "odorant binding protein" and as such can be viewed as an "extracellular sensor-protein" (Pelosi 1982,1996,1998,2001; Vosshall & Stensmyr 2005; Cavaggioni A & Mucignat-Caretta C 2000). Other members of this class include MUP (major urinary proteins) and α(2U) globulins (Beynon & Hurst 2004, Cavaggioni & Mucignat-Caretta 2000). These "sensor-proteins" are extracellular and have a specific chemical binding properties. Lipocalins have potential to be used both as "sensor-proteins" by themselves in cell-free systems.This presentation reports initial investigations into using this family of proteins for sensing volatiles in an artificial sensing platform.

 

A model of stimulus-specific neural assemblies in the insect antennal lobe (Dominique Martinez)

We investigate how olfactory stimuli trigger neural assemblies in the first relay of the insect olfactory system i.e., the antennal lobe. Using computational modeling we study the role of the GABAergic network in producing stimulus-specific synchrony.

 

The computational architecture of biological olfaction (Thomas A. Cleland) 

Cellular and neural network properties in the olfactory epithelium and olfactory bulb help resolve many difficult problems inherent to chemical detection and identification.  Some of these biophysical properties provide novel insights and mechanisms for artificial chemosensory systems to emulate, whereas others simply reflect the peculiarities of the biological substrate and have little to contribute to artificial systems design.  I will present a theoretical and computational overview of early olfactory processing, emphasizing particular features relevant to the design and construction of artificial systems. 

 

Space and time in the nose, an artificial olfaction mucosa (Tim Pearce)

The chromatographic properties of the olfactory mucosa have long been considered to contribute to chemosensory perception in mammals, presumably through segregation of odour components when presented in combination. I will describe a new biomimetic sensing technology for complex odour detection, termed the artificial olfactory mucosa, which explicitly makes use of this segregation principle, and exhibits behaviour reminiscent of the olfactory epithelium. Experimental data from this device demonstrates temporal segregation of odorants due to selective phase partitioning during odour delivery, in turn giving rise to complex spatio-temporal dynamics in the responses of the chemosensor array population, which will be shown to lead to enhanced complex odour discrimination. An analytical model and new measure of spatio-temporal information will be introduced that quantifies the contribution of both space and time to the discrimination performance of the system. Finally, I will consider the challenge of extracting stimulus-specific information from these new chemosensing devices, which requires specialised time-dependent signal processing, information measures and classification techniques.

 

A neuromorphic approach towards artificial olfaction in robots (Dr Sergio Bermudez)

Recently, the advances in the engineering of artificial noses have provided us with a vast number of potential applications ranging from environmental monitoring, odor detection/discrimination to demining. However, these technologies are still to brittle to be deployed at a large scale in the real world. In contrast, biological systems outperform by far the capabilities of any technological solution on this domain. We therefore aim at understanding the biological solutions to the real world chemical sensing problem with the goal to construct more robust technologies.

Usually, olfaction is seen as a processing module independent from the other senses, and separate from behavior. I will show by both analyzing biological examples and our own work, that aims at constructions a fully autonomous neuromorphic artificial insect, that this is not necessarily the case but rather olfaction depends on multimodal integration processes and specific behaviors. Our neuromorphic approach is based on the neural subtrate and behavior of odor processing in moths. In particular, I will show different steps towards the construction of an artificial moth, and how these applied to a number of real world applications, such as environmental mapping and odor source localization by teams of mobile and flying robots.