Our lab studies olfactory information processing. We are primarily interested in solving two fundamental questions:

  1. How is odor information represented in the brain of the awake, behaving mouse?

  2. How is the information that is relevant to animal behavior extracted by the brain? In short, we want to understand what the mouse’s nose tells its brain. 


This animation shows how information about smells is processed by the nervous system. Molecules travel through nose and activate olfactory receptor neurons (ORNs) on the surface of the epithelium. ORNs express one and only one olfactory receptor gene from the large family of genes (human: ~350, mouse: ~1200). The axons of ORNs, which express the same receptor gene, converge into one or few small areas in the olfactory bulb, called glomeruli. Olfactory bulb is the first processing center in the brain of the information about odorants. The principle olfactory bulb neurons, mitral/tufted cells, receive their inputs from the glomeruli and send their outputs to the olfactory cortices. Their activity is modulated by a vast network of inhibitory interneurons.


Mammalian olfaction has unique structural and functional characteristics as a system for studying sensory processing. Each channel of sensory information has a specific genetic tag. Compared to the visual or auditory systems, the neuronal organization of the olfactory system is relatively simple: receptors are only "two synapses away" from the cortex, and the flow of information to the cortex bypasses the thalamus. Direct corticobulbar projections, which are analogous to corticothalamic projections in other sensory systems, are presumed to be responsible for feedback control of the olfactory processing. Overall, the availability of modern genetic tools, the relative simplicity of the neuronal circuitry, and the extremely high relevance of smells to animal behavior makes olfaction a great model system for studying the principles of sensory information processing.

Neural coding

To understand the sensory processing, we need to understand how information is represented by neuronal activity at different levels (1,2), and which features of neuronal activity is read by the brain and leads to behavior (3). In addition, the feedforward flow of information from sensory system to motor output is modulated by feedbacks at multiple levels (4,5..), including direct effects of behavior on extracting of sensory information (6).


Research in the lab is directed towards understanding different arrows at this diagram: what computations are responsible for processing sensory information, and how these computations are implemented in neural networks. 

Our lab's early, foundational work focused on unraveling the temporal aspects of olfactory information processing using psychophysics, electrophysiology, and optogenetics. Contrary to the common belief, we demonstrated that olfactory processing is temporally very precise at both the levels of the neural code (Shusterman-2011) and of behavior


(Smear-2011, -2013), and also that it can be very fast: mice can make olfactory decisions  in less than 100 ms (Resulaj-2015). 

Previous work

Odor identity coding


Humans can easily identify visual objects independent of their position and lighting, words in speech independently of their volume and pitch, and smells independently of odor concentration or the presence of background odors. Understanding such invariant object recognition is a fundamental problem of sensory neuroscience. Olfaction, being arguably the most genetically tractable and anatomically compact sensory system, is an ideal system in which to approach this basic question. Despite the accumulated knowledge on molecular, anatomical, and physiological properties in olfaction, the fundamental basis of how odorant identity is encoded in patterns of neural activity is still an unsolved problem.


To understand the principles of concentration-invariant odor identity coding we utilize two complementary approaches: 

  1. In behavioral experiments we train a mouse to discriminate between odorants independently of their concentrations. We then use optogenetic stimulation to perturb neural processing at different times during stimulus acquisition, in order to identify the temporal interval relevant for this behavioral task.

  2. In a parallel effort, we record the activity of multiple MT cells during this task, and search for concentration invariant signature of MT cell odor responses


see also recent presentation ->

Current projects

The role of single channels in the highly distributed olfactory processing

Focusing on the timing aspects of sensory processing allowed us to temporally localize the features of the olfactory neuronal code that are relevant to the animal's behavior. Alternatively, capitalizing on the genetic organization of the mouse's olfactory system, one can localize neuronal code features in “space”, that is, in the space of the 1000+ genetically specific channels converging onto spatially segregated glomeruli.

The olfactory code is highly spatially distributed, in the sense that each odor is represented by activity pattern of many glomeruli. However, using optogenetic stimulation of individual glomeruli, we demonstrated that surprisingly, the activation of even a single glomerulus is detectable by the animal, with both the total activation level and the timing of the glomerulus activation serving as cues that can be behaviorally perceived (Smear-2013). Moreover, in a parallel line of research, the Bozza group established that a single glomerulus is sufficient to evoke an innate behavior (Dewan-2013). These results prompt us to pursue the following fundamental question: how is the signal from a single glomerulus conveyed to the cortex?


To address this problem, we developed a new method for optogenetic targeting of mitral/tufted (MT) cells, the cells that receive excitatory input from a specific glomerulus. This allows us to achieve two goals:

  1. By recording in an awake animal the responses of these MT cells to a battery of odorants with a known affinity to a corresponding olfactory receptor, we probe the transformation of information at the single-glomerulus level.

  2. Using precise stimulus and behavioral control we experimentally manipulate the relevance of the information transmitted via these cells in order to determine the features of the MT code relevant to behavior.

The most powerful way to test a neural coding model is to validate that the hypothesized code can be read by higher brain areas where it leads to an observable behavioral output. To this end, we are developing experimental strategies to establish causal links between the neuronal code and animal behavior. Following our earlier work with single glomerulus stimulation (Smear-2013) we are developing a general approach to test the readability of the olfactory code using an optogenetic pattern stimulation system in behaving mice. In this paradigm, mice are trained to recognize a specific spatial-temporal pattern of glomerulus activation. Through systematic perturbation of this pattern during measurements of the behavioral response to these perturbations, we investigate the role of different features of the code for animal behavior.


The next step after dissecting the olfactory code at the glomerulus/receptor level is testing of neural coding models at the individual neuron level. This requires developing a different technology capable of individual neuron stimulation. In a BRAIN Initiative funded collaborative project with Prof. Shy Shoham (Technion, Israel) we are developing a system for 2-photon holographic optogenetic stimulation of multiple neurons deep in the brain of the behaving animal.


Behavioral relevance of the neuronal code features