(A) To assess if the learning-dependent improvement in the decoder performance requires an involvement of the piriform cortex, we infused muscimol (500 nl, 2 mM) in the ipsilateral, anterior piriform cortex during imaging, while trained mice performed the mixture task. A cannula was implanted at an angle, in a manner similar to Otazu et al., 2015, in order to infuse muscimol. (B) When the mice were trained to perform the task well, the imaging sessions commenced. Muscimol and control sessions alternated, where coordinates for the imaged fields of view were recorded and could be revisited, so that imaging could be obtained from the same location under the control and muscimol conditions. (C) Example matched fields of view for control session (left) and muscimol session (right). Arrowheads point to the same TCs imaged on different days. (D) Example confocal image of a coronal section of the brain, showing a track left by an implanted cannula (white dotted line) in the left piriform cortex. Blue, DAPI fluorescence; red, fluorescence from DiI, which was injected before perfusion. (E) Behavioural performance for control and muscimol sessions. Sessions with the common fields of view are paired. p = 0.069, paired t-test (n = 6 sessions, three mice). (F) Support vector machine (SVM) performance in the ability to discriminate S+ odours (ethyl butyrate [EB] present) vs. S- odours (EB absent) for control (black) and muscimol (green) datasets. Mean ± SEM shown. (G) Same as (F), except that the training set comprised responses from single-odour trials, and the test set comprised mixture responses.