Anticipating potential threats in the environment and responding with adaptive behavior is essential for animal survival. Therefore, strong mechanisms of aversively-motivated associative learning, called fear learning, have evolved in the animal kingdom. In the mouse model, auditory-cued fear learning can be used to study how the association between a neutral sensory cue (conditioned stimulus; CS), and an aversive stimulus (unconditioned stimuli; US) leads to plasticity in neural networks. The neuronal mechanisms of fear learning are complex, involving several brain areas, and within each area, there is likely a coordinated activity and plasticity of excitatory neurons and several types of inhibitory interneurons. In my PhD thesis, I have leveraged the availability of Cre mouse lines for the main cortical interneuron types and principal neurons, together with miniature microscope Ca2+ imaging, to systematically record the activity of three interneuron classes and two types of principal neurons during a three-day fear learning protocol. I focused on the posterior insular cortex (pInsCx), a multisensory area processing auditory, somatosensory, nociceptive, and interoceptive information and regulating aversive behavior and fear balance. Regarding the responses to mildly aversive foot shocks (US) during fear learning training, I found a sparse representation (~ 14%) of a short-latency US-onset response in CaMKII+ principal neurons; interestingly, the US-onset representation was higher (~ 50%) in the three classes of GABAergic interneurons (PV -, SOM -, and VIP - interneurons). In all neuron classes, I found sub-populations coding for foot shock termination, suggesting that the pInsCx is relevant for pain relief. Variable fractions of all neuron classes were also modulated by the behavioral states of freezing onset, or termination of freezing. Regarding the plasticity of the CS representation in the pInsCx, each neuron class contained sub-populations that newly developed responses to the CS ("CS-learners"), but also those that lost CS-responses after fear learning ("CS-unlearners"). Importantly, VIP - and PV - interneurons showed a relatively high number of CS-unlearners, thus they could mediate a loss-of-inhibition of principal neurons during fear memory recall. On the other hand, principal neurons and SOM - interneurons showed a larger percentage of "CS-learners"; nevertheless, the plasticity types varied according to whether phasic or tonic CS responses were analyzed. I next investigated the connectivity of VIP - and SOM - interneurons using optogenetically-assisted circuit mapping, and found that VIP - interneurons inhibit SOM -interneurons, as in other cortical areas. Unexpectedly, VIP - interneurons also inhibited principal neurons with a similar strength, providing a possible route for a loss-of-inhibition of principal neurons after fear learning. Finally, I employed optogenetic silencing to investigate whether the activity of SOM - or VIP - interneurons during the training session is necessary for fear memory formation, but effects on fear memory retrieval could not be demonstrated. These data suggest that there are different microcircuits in the pInsCx recruiting distinct neuron types for the coding of specific sensory information and behavior states during fear learning. Together, the findings in this thesis provide new insights into the complexity of neuronal signaling and plasticity during fear learning.