The world often presents us with sequences of multisensory stimuli unfolding over time, like a car appearing after we hear it approaching. In the cerebral cortex, individual neurons integrate information across senses and time in order to identify these objects. Meanwhile, neuromodulatory molecules shape neural computations, adapting them to context and experience. It is from this interplay that the spatio-temporal patterns of neural activity containing the code for how we perceive the world originate. However, these activity patterns have long been experimentally inaccessible, and they have never been investigated concurrently to neuromodulation. With this project, I will reveal the causal role of functionally-identified ensembles of neurons in the perception and learning of audio-tactile sequences. I will combine, for the first time, two-photon (2p) imaging and holographic-optogenetics with fluorescent neuromodulation sensors. My experiments will be performed in the posterior parietal cortex (PPC) of the mouse, a multimodal area acting as sensory history buffer. I will focus on noradrenaline (NA), a modulator so far involved in sharpening processing of weak unisensory features. I will 1) test which PPC neurons are sufficient, when activated, to bias perception of multisensory sequences during a behavioural task; 2) simultaneously, I will test whether and how PPC levels of NA change when altering sequence discriminability. My experiments will be supported by a solid theoretical framework based on Information Theory. In conclusion, I will use a unique combination of advanced molecular, optical, and computational tools to pursue a novel, causal and integrative approach to cortical function: one where functional neuronal subgroups and neuromodulation are investigated together, at high-resolution, to explain the multi-layered biological basis of the fascinating interaction between our brains and the world.