AUDITORY CORTEX POPULATION DYNAMICS MEDIATING PERCEPTION OF AMBIGUOUS THREAT

Naturally occurring threats are often both complex and ambiguous. To understand how the brain perceives threats in such naturalistic contexts, we introduced controlled ambiguity into an auditory threat perception task while recording neuronal activity in the auditory cortex of mice. We identified a dynamically emerging threat perception moment (TPM) that fluctuated across trials and animals, preceded by a phase of ambiguity reflecting the uncertain nature of potentially threatening cues. To detect the TPM, we used facial expression, which we were able to associate with distinct cognitive processes, as a temporally precise behavioral readout. Additional measures including pupillometry, body posture, and movement were incorporated and used as inputs to a support vector machine classifier (SVM). By correlating these behavioral markers with two-photon calcium imaging in secondary auditory cortex (AuV) we revealed that while individual neurons responded to sounds across a range of intensities, population responses increased after the TPM, indicating learning-dependent plasticity. Dimensionality reduction using principal component analysis showed improved discrimination along with an expansion of population trajectories in the ambiguity period that is maintained after threat is perceived. Consistently, SVM classification of neuronal activity revealed a transition to a threat-related population state during ambiguous stimuli, in agreement with behavioral decoding. Furthermore, population vectors progressively converged onto the threat-related population geometry as ambiguity resolved. Finally, targeting amygdala-projecting neurons in AuV revealed a population highly enriched in ambiguity encoders, suggesting that this specific projection is involved in uncertainty representation. These data therefore delineate how cortical processing subserves the perception of ambiguous threat.