INTEGRATIVE ANALYSIS OF FEAR LEARNING DYNAMICS: COMBINING HUMAN FMRI AND MOUSE MINISCOPE DATA

Fear learning is a central mechanism in adaptive and pathological responses to threat, yet it is typically investigated with different methods in animals and humans, complicating translational inference. We develop a cross-species framework, based on representational similarity analysis (RSA), to directly compare the activation patterns of specific brain areas using calcium imaging in mice and fMRI in humans during fear learning. Miniscope calcium recordings from the prelimbic cortex (PL) were acquired while the mice performed fear acquisition and extinction sessions. For the human data, we use BOLD signal data from fMRI in the homologous dorsal anterior cingulate cortex (dACC). To make the two measures comparable, we temporally filter and downsample the calcium traces in order to match the BOLD signal timescale and allow the construction of representational dissimilarity matrices (RDMs) analogous in both datasets. RSA help us quantify the similarity of representational geometry across species and sex, testing whether fear-related conditions are organized along shared dimensions in mouse and human circuits. Complementary multivariate approaches (e.g., independent component analysis and connectivity-based metrics inspired by calcium–fMRI coupling work) will characterize latent networks and their temporal dynamics, providing convergent evidence for cross-species alignment. By emphasizing representational structure rather than exact spatial correspondence, this framework aims to establish a translational bridge between cell-specific calcium activity and human BOLD patterns, clarifying how conserved network geometries support fear learning and how they are modulated by sex and potential interventions.