NEURAL ENCODING OF HIGH-FREQUENCY ODOUR STRUCTURE ACROSS OLFACTORY AREAS

Natural odour plumes possess complex spatiotemporal structures that carry critical information about the environment, yet how these dynamic temporal features are represented across the olfactory system is poorly understood. To address this, we recorded large-scale neuronal activity using Neuropixels probes in the olfactory bulb (OB), anterior olfactory nucleus (AON), and piriform cortex (PCx) of awake, head-fixed mice. We delivered a panel of systematically varying odours modulated at 50Hz, aligned to the onset of inhalation, to investigate the representation of sub-sniff temporal structure. We found that key aspects of the temporal structure, such as onset, intermittency and total odour, could be reliably decoded from neural activity, with linear classifiers performing above chance level across all three regions. However, computational analyses revealed distinct processing strategies: Principal Component Analysis (PCA) showed lower dimensionality in the OB compared to cortical regions, and Representational Similarity Analysis (RSA) indicated that OB representations were preserved more strongly in the AON than in the PCx. Furthermore, while Generalized Linear Models (GLMs) successfully predicted firing rates based on stimulus features, the specific feature weights varied substantially across regions. These findings demonstrate that temporal odour features are not uniformly inherited but are differentially transformed and selectively emphasised across olfactory circuits, suggesting that the AON and PCx have different computational roles in encoding dynamic sensory environments.