Rapid fluctuations in functional connectivity of cortical networks encode spontaneous

Abstract

Experimental work across a variety of species has demonstrated that spontaneously generated behaviors are robustly coupled to variation in neural activity within the cerebral cortex. Indeed, functional magnetic resonance imaging (fMRI) data suggest that functional connectivity in cortical networks varies across distinct behavioral states, providing for the dynamic reorganization of patterned activity. However, these studies generally lack the temporal resolution to establish links between cortical signals and the continuously varying fluctuations in spontaneous behavior typically observed in awake animals. Here, we took advantage of recent developments in wide-field, mesoscopic calcium imaging to monitor neural activity across the neocortex of awake mice. Applying a novel approach to quantifying time-varying functional connectivity, we show that spontaneous behaviors are more accurately represented by fast changes in the correlational structure versus the magnitude of large-scale network activity. Moreover, dynamic functional connectivity reveals subnetworks that are not predicted by traditional anatomical atlas-based parcellation of the cortex. These results provide insight into how behavioral information is represented across the mammalian neocortex and demonstrate a new analytical framework for investigating time-varying functional connectivity in neural networks.

Not so spontaneous: Multi-dimensional representations of behaviors and context in sensory areas

“…In the developing awake rat, an increase in movement-associated firing rates in the visual cortex occurs just before eye opening (between P10 and P13) (Murata and Colonnese, 2018;Figure 1Bii). This link between SA and behavior becomes even stronger in the adult mouse where spontaneous neural activity is predicted with high accuracy from multiple dimensions of facial movements (Figure 1Biii; Benisty et al, 2021;Musall et al, 2019;Salkoff et al, 2020;Stringer et al, 2019b). Thus, SA in developing animals demonstrates an increasing correlation with behavior in agreement with observations in the mammalian adults (see section behavioral representations in spontaneous activity patterns).…”

“…Several SA patterns captured in large-scale recordings correspond to various behaviors such as running, whisking, sniffing, grooming and freezing in mice (Benisty et al, 2021;Clancy et al, 2019;Gr€ undemann et al, 2019;Lanore et al, 2021;Stringer et al, 2019b), leg movements and locomotion in flies (Fujiwara et al, 2017;Strother et al, 2018;Zolin et al, 2021), and tail flicks in zebrafish (Pietri et al, 2017;Romano et al, 2015). Due to the high-dimensional nature of SA, many different behaviors can be represented in the same circuit, and indeed, multiple dimensions of behavior were detected in SA across Box 1.…”

“…OPEN ACCESS brain areas in mice (Benisty et al, 2021;Gr€ undemann et al, 2019;Lanore et al, 2021;Stringer et al, 2019b). These behaviorally correlated patterns continue during presentations of sensory stimuli (Stringer et al, 2019b) and during decision-making tasks (Engelhard et al, 2019;Musall et al, 2019;Salkoff et al, 2020;Steinmetz et al, 2019).…”

Variability in sampling of cortex-wide neural dynamics explains individual differences in functional connectivity and behavioral phenotype

“…In other words, animals with more atypical expression of motifs also had more atypical behavior. Deeper analyses confirmed that animals’ neurotype was broadly associated with individual behavioral tests; but as expected given the dominance of motor representations in cortical activity 26,27 , neurotype was most strongly related to tests of motor phenotype (Extended Data Fig. 6).…”

Decomposed Linear Dynamical Systems (dLDS) for learning the latent components of neural dynamics

“…Traditional linear methods, such as PCA, ICA, and their variants [57,56], have recently given way to more flexible descriptions of dimensionality reduction. Example nonlinear methods include local embeddings [3,49] and variational auto-encoders [27], which rely on the manifold hypothesis [56,43,22,16,7,39]. Assuming that instantaneous neural activity patterns lie on a low-dimensional manifold removes the assumption of linearity in the data, and thus enables the identification of correlated activity that corresponds to a much lower-dimensional geometric structure.…”