Uncovering motifs of concurrent signaling across multiple neuronal populations

Modern recording techniques now allow us to record from distinct neuronal populations in different brain networks. However, especially as we consider multiple (more than two) populations, new conceptual and statistical frameworks are needed to characterize the multi-dimensional, concurrent flow of signals among these populations. Here, we develop a dimensionality reduction framework that determines (1) the subset of populations described by each latent dimension, (2) the direction of signal flow among those populations, and (3) how those signals evolve over time within and across experimental trials. We illustrate these features in simulation, and further validate the method by applying it to previously studied recordings from neuronal populations in macaque visual areas V1 and V2. Then we study interactions across select laminar compartments of areas V1, V2, and V3d, recorded simultaneously with multiple Neuropixels probes. Our approach uncovered signatures of selective communication across these three areas that related to their retinotopic alignment. This work advances the study of concurrent signaling across multiple neuronal populations. Introduction Cortical circuits functionally involve feedforward, feedback, and horizontal interactions between many neuronal populations that span distinct areas and layers. Recording techniques now allow us to record from many neurons across these populations [1-3] (Fig. 1a ). To capitalize on the scientific opportunities presented by these recordings, however, new conceptual and statistical frameworks are needed, particularly as we consider communication across multiple (more than two) populations. Characterizing interactions between just two populations is a challenging high-dimensional problem. Dimensionality reduction techniques have therefore been increasingly used for this purpose [4] [5] [6] [7] . Methods like canonical correlation analysis (CCA) [8] and its probabilistic variants [9] , in particular, identify a low-dimensional set of latent variables that parsimoniously describe the interactions between two populations. This cross-population shared-latent model has inspired several extensions targeted toward neural recordings [10] [11] [12] [13] [14] . Communication between two populations, however, occurs bidirectionally and likely concurrently, and disentangling this concurrent communication is a substantial challenge in neuroscience. A