Mechanisms of top-down modulation in sensory perception and their relation to the underlying connectivity are not completely understood. Here we present a computational model of two interconnected cortical areas and use it to study the impact of inter-areal connectivity on sensory processing. The model integrates a multitude of data from rodent primary somatosensory cortex and reproduces biological features across multiple scales: from tens of ion channel models and electrical types, to thousands of morphologically detailed neurons, to millions of stochastic synapses mediating local and long-range networks. Notably, long-range connectivity in the model incorporates target lamination patterns associated with feedforward or feedback pathways in the literature. First, we defined model cortical areas X and Y by extracting reciprocally connected subvolumes from a previous larger model. Then, we provided sensory inputs to area X by activating its thalamic afferents. We calibrated the levels of noisy background inputs and the number of recruited fibers to obtain PSTH peak latencies comparable to the literature. Afterwards, we analyzed feedback-mediated secondary responses in area A and performed connectivity manipulations to better understand their origin. Finally, we studied sensory discrimination and integration. We detected functional cell assemblies after repeated presentation of four different stimulation patterns, finding stimulus specificity in early but not in late assemblies. We also analyzed the responses to two stimuli presented with variable time delay, finding that feedback-mediated secondary responses can be signatures of spatio-temporal summation. This work represents a first approximation to the study of cortico-cortical interactions in biophysically-detailed computational models of cortex.