The neural basis of odour concentration invariance

  • Neuroscientists Doug Storace, Larry Cohen and Yunsook Choi, from Yale School of Medicine and the Korea Institute of Science and Technology, aim to improve our understanding of olfactory processing and odour recognition. In particular, the team have studied the neural basis of ‘concentration invariance’, whereby the concentration of an odour does not impact our identification of that specific odour. Input and output signals of the olfactory bulb were measured over a range of concentrations. Results showed that despite significant changes in input signals, output signals remain relatively stable, suggesting that the olfactory bulb has an important role in generating concentration invariance of odour recognition.

    Our sense of smell is the most powerful sense that we possess. In fact, humans can detect around 10,000 different scents! Odours are extremely evocative, triggering memories and emotions in humans. Additionally, from a more primal point of view, odour recognition is crucial for survival. Odours are used to recognise individuals, find mates and food and even to warn us of potential danger. Interestingly, humans and other animals can identify a specific odorant as the same over a range of odorant concentrations. This phenomenon is known as ‘concentration invariance’. In other words, the concentration of the odour does not alter our recognition of the odour. The olfactory sensory pathway is extremely complex, involving 13 different brain regions and millions of neurons and receptors. The intricate processes involved in odour recognition have made it difficult to study concentration invariance in the past. However, neuroscientists Douglas Storace, Larry Cohen and Yunsook Choi have focussed their research efforts on tackling this challenge and aim to improve our knowledge of the neural basis underpinning concentration invariance.

    A histological section shows anatomical targeting of Cal-590 dextran to the olfactory receptor nerve terminal input in the glomeruli (left) and genetic targeting of ArcLight to the mitral and tufted cell output (middle) in the same section along with a merged image containing DAPI (right).

    The olfactory sensory pathway
    To investigate odour concentration invariance, it is vital that we understand the olfactory sensory pathway. When an odour is present, the odour molecules enter the nasal cavity and bind to epithelial olfactory receptor cells. This binding of the chemical signal to the receptor activates the olfactory receptor cells, causing them to send information to the olfactory bulb, located in the forebrain in vertebrates. Within the olfactory bulb are glomeruli, spherical structures in which synapses (or neural connections) form between the axons of olfactory receptor cells and the dendrites of mitral cells and tufted cells. These cells then relay the information to higher olfactory centres where it is further processed, resulting in three main responses:

    1. we consciously acknowledge there is an odour present;
    2. the odour can trigger our memory; and
    3. the odour can evoke an emotional response.

    Storace, Cohen, and Choi asked the question: ‘In what part of the olfactory sensory pathway is concentration invariance computed?‘ Studies have speculated tht the olfactory bulb has a role in odour concentration invariance. Therefore, the team compared activity signals of the inputs and outputs of the olfactory bulb over a range of different odour concentrations. In this study the ‘input’ is the signals from the olfactory receptor cells and the ‘output’ is the activity of the mitral and tufted cells.

    Optical measurements from input and output sensors in one glomerulus in response to ethyl tiglate presented across ~2 log units of odorant concentration (0.12–11% of saturated vapor). Input and output measurements were performed sequentially using Fura dextran and ArcLight, using excitation wavelengths of 380 nm and 480 nm, respectively. The optical traces are low-pass filtered at 1 Hz and are the average of 3–7 individual trials aligned to the first sniff following odor onset. The odor and respiration traces are from one of the single trials. The data in b and c are from different preparations. onl, olfactory nerve layer; gl, glomerular layer; epl, external plexiform layer; mcl, mitral cell layer, ORN, olfactory receptor neuron; M/T, mitral and tufted cells.

    The experimental procedure
    The team developed an innovative method for determining the input/output activity of individual glomeruli within the olfactory bulb. To distinguish between input and output signals, the team used different sensors for the two types of neural activity, enabling independent detection and measurement of the two different signals. To anatomically detect olfactory receptor cells (input), the team used a nasal infusion with an organic calcium sensitive dye. Calcium was used as a target because it is an essential element in neural signal transduction. A more complex process was used to identify mitral and tufted cells (output). Storace, Cohen, and Choi used transgenic mice that expressed specific genetically encoded voltage or calcium indicators only within the mitral and tufted cells.

    Comparing activity maps
    The team developed activity maps in order to compare input and output signals. In the study, signal measurements were made from thirteen glomeruli at four different odour concentrations. Interestingly, the results indicated that the output maps for each of the four odour concentrations are much more similar to each other than are the input maps. This suggests that the olfactory bulb has an important role in generating the perception of odour concentration invariance.

    The team compared activity signals between the inputs and outputs of the olfactory bulb over a range of different odour concentrations.

    In these experiments, different sensors were used in order to distinguish between input and output signals in individual mice. However, the use of these two different sensors could potentially affect the outcome of the study. Therefore, the team performed similar experiments but used several different sensors i.e. calcium dyes or genetically encoded indicators. Again, the output activity maps were significantly more correlated with each other than the input maps.

    Experimental approach: (Left) Olfactory receptor neuron nerve terminal input was labelled via a nasal infusion of a calcium sensitive dextran dye. Cre recombinase expressing mitral and tufted cell output was targeted using cre-dependent viral vectors that expressed genetically encoded voltage or calcium indicators (GEVI or GECI). (Right) By using activity sensors with substantially different excitation or emission spectra, input vs. output can be measured independently from the same glomerulus by changing the excitation or emission wavelengths.

    Importance of the study
    In the past, researchers believed that odour identity is influenced by the combination and quantity of odorant receptors activated by the odorant. However, Storace, Cohen, and Choi have shown that odour identity is actually more likely to be determined by the glomerular output of the olfactory bulb, rather than the input. A process within the olfactory bulb essentially stabilises the effect of odour concentration so that the output represents odour identity rather than a combination of quality and quantity. More research is needed to understand exactly how this process occurs. It is interesting to note that although output maps are very similar to each other, they are not identical. However, this may not be a major issue – perhaps other brain regions eliminate these relatively small differences, further along the olfactory sensory pathway.

    Although this study has provided evidence that suggests the olfactory bulb significantly impacts the perception of concentration invariance, the team acknowledge that this conclusion can only be confirmed if they could suppress the olfactory bulb calculation that generates a concentration invariant output and show that this disrupts olfactory perception. Unfortunately, it is not presently known how this suppression might be carried out. The input and output signals of the olfactory bulb are interconnected with over 20 different interneuron cell types in the glomerular layer which modulate output. For example, these interneurons could increase or decrease responses to stimuli depending on the strength of the stimuli i.e. boosting responses to weak stimuli and reducing responses to strong stimuli via electrical synapses.

    The mean spatial correlation of each frame subtraction map with the map from the other three concentrations. The activity maps of the output (right) are more similar to each other than are the activity maps of the input (left). This summary includes 14 measurements from 13 preparations (responses to two different odorants were measured in one preparation).

    How this process occurs is a question yet to be answered. Processing in the olfactory bulb must convert relatively weak inputs into signals that can be used by higher centres for odour recognition. But, how are small input signals converted into larger output signals? Exactly how they do so is unknown and an area of future research.

    Output maps for each of the four odour concentrations are much more similar to each other than are the input maps.

    Storace, Cohen, and Choi have performed novel measurements that provide evidence that odour concentration invariance is computed by the olfactory bulb. Different odour concentrations greatly affect input signals. However, these signals are processed within the olfactory bulb to produce relatively stable output signals that are used to identify the odour. The novel methods used by the team for this study could also be used in future experiments to determine whether the olfactory bulb affects odour perceptions other than concentration invariance and to look at input-output conversion in other brain regions.

    Are there other areas in the brain where input-output conversion is key?
    The input-output transformation is key to understanding in every brain region. The olfactory bulb and the retina are examples where determining the transformation can be done simply and comprehensively.

    The finding that the olfactory bulb contributes to the perception of concentration invariance raises the possibility that it also does computations for other olfactory perceptions. Simultaneously?
    Very interesting!


    • Storace, D.A. and Cohen, L.B. (2017) Measuring the olfactory bulb input-output transformation reveals a contribution to the perception of odorant concentration invariance. Nature communications, 8(1), p.81.
    • Axel, R. (1995) The molecular logic of smell. Scientific American, 273(4), pp.154-159.
    • Chapman, R. and Bernays, E. (2008). Olfactory Bulb Anatomy, Encyclopedia Britannica. Available at:
  • Research Objectives
    We can recognise a smell, even when the concentration of it changes. Storace and Cohen’s paper investigates which part of the brain is responsible for generating this ‘concentration invariance’.


    • US National Institutes of Health
    • Korea Institute of Science and Technology

    The three neuroscientists are excited about learning how the brain accomplishes odour perceptions including concentration invariance, odour accommodation, and odour blocking. They suspect that the initial processing stage, the olfactory bulb, may have a role in all three and may carry out all three computations simultaneously. Oh my! It is possible because the bulb does have many different interneuron types.

    Douglas Storace, Larry Cohen, and Yunsook Choi
    Dept of Physiology: Yale, New Haven, CT, USA
    Center for Functional Connectomics, KIST, Seoul, Korea

    T: +1 (203) 785 4047

  • The neural basis of odour concentration invariance


Leave A Reply

%d bloggers like this: