["The Journal of Physiology, EarlyView. ", "\nAbstract figure legend The olfactory bulb input–output transformation was imaged using dual‐colour two‐photon imaging. (Top) Cartoon and histological examples of labelling the olfactory bulb input and output using spectrally distinct optical sensors. (Middle) Modelling and experimental work reveal that input neurons that respond with primarily monotonic concentration‐response relationships are transformed into a mix of monotonic and non‐monotonic responses. (Bottom) Olfactory bulb output state space for a population of three glomeruli with exclusively monotonic responses (black) and with both monotonic and non‐monotonic responses (red). Each trajectory represents a cloud of points that reflects variations in the fluctuations of molecular components present in natural odour stimuli. Monotonic responses alone cluster together, whereas multiple response types broaden coverage of mitral and tufted cell (MTC) state space. Greater coverage of state space makes it easier to discriminate one odour from another, even though the dimension of olfactory bulb state space is greater than 3.\n\n\n\n\n\n\n\n\n\nAbstract\nAnimals can recognize an odour as the same odour across a range of concentrations and discriminate between odours in complex environments. Processing within the mouse olfactory bulb (OB) may be involved, yet the underlying mechanisms remain unclear. Each olfactory receptor neuron (ORN) type maps to the OB in olfactory receptor‐specific channels called glomeruli, where they connect with the dendrites of mitral and tufted cells (MTCs), which project their axons to the rest of the brain. Modelling this transformation yielded predictions about how two kinds of inhibition, local intraglomerular and lateral interglomerular processing, shape MTC output as a function of concentration changes. We confirmed these predictions using in vivo single and dual‐colour two‐photon Ca2+ imaging from the ORNs and MTCs innervating the same glomeruli in the awake mouse OB in response to odours presented across a wide concentration range. We identified a transformation where concentration increases transformed ORN inputs with exclusively monotonically increasing responses into MTC outputs that non‐monotonically increased then decreased, or vice versa. This transformation was odour‐specific, consistent with rising levels of inhibition that scale with excitatory input and predicted by certain ORN characteristics. Therefore non‐monotonic concentration‐response relationships in MTCs are common and expected given how each glomerulus is shaped by feed‐forward excitation, local and lateral inhibition. We propose that this transformation facilitates odour discrimination and the ability to achieve concentration‐invariant odour perception by broadening MTC odour state space.\n\n\n\n\n\n\n\n\n\nKey points\n\nThe role of the olfactory bulb in transforming sensory information remains poorly understood. In the bulb different olfactory receptor neuron (ORN) types map to olfactory receptor‐specific channels called glomeruli, where they interact with the dendrites of mitral and tufted cells (MTCs), which project to the olfactory cortex.\nWe developed a mathematical model detailing how different ORN inputs are transformed by intraglomerular and lateral interglomerular inhibition as a function of odour concentration.\nSingle‐ and dual‐colour in vivo two‐photon Ca2+ imaging of glomerular inputs and outputs across a wide concentration range confirmed some of the predictions.\nIncreasing odour concentration transformed ORN inputs with monotonically increasing responses into non‐monotonic MTC outputs, which responded with increases then decreases, or vice versa. This transformation is heterogeneous, odour‐specific and predicted by certain ORN characteristics.\nThis network transformation broadens the MTC odour state space, providing a mechanistic basis for achieving concentration‐invariant odour perception and fine odour discrimination.\n\n\n"]