The secret life of predictive brains: what’s spontaneous activity for?

Abstract

Brains at rest generate dynamical activity that is highly structured in space and time. We suggest that spontaneous activity, as in rest or dreaming, underlies top-down dynamics of generative models. During active tasks, generative models provide top-down predictive signals for perception, cognition, and action. When the brain is at rest and stimuli are weak or absent, top-down dynamics optimize the generative models for future interactions by maximizing the entropy of explanations and minimizing model complexity. Spontaneous fluctuations of correlated activity within and across brain regions may reflect transitions between "generic priors" of the generative model: low dimensional latent variables and connectivity patterns of the most common perceptual, motor, cognitive, and interoceptive states. Even at rest, brains are proactive and predictive.

A visual representation of the hand in the resting somatomotor regions of the human brain

“…In summary, the stability of the spontaneous activity suggests that this set of neural signals is a possible candidate to preserve long-term models and priors of common behaviors and natural stimuli 3,4 . These prior representations are the result of statistical learning mechanisms that store the co-occurrence statistics of hand visibility and usage, instrumental to the exploration and manipulation of the surroundings.…”

“…More specifically, during offline periods, the brain forms generic priors or low-dimensional representations, as categories or synergies, rather than individual instances or movements, that summarize the relative abundance of visual stimuli, objects, or motor patterns in the natural environment. Interestingly, this reduced subspace of summary representations is formed along a hierarchy that has the sensorimotor cortices at the lower level 4 . Consistently, our results show that at rest the somatomotor regions maintain a multivoxel pattern of activity that resembles that evoked by the presentation of the natural hands compared to control stimuli (e.g., food).…”

Symptom-severity-related brain connectivity alterations in functional movement disorders

“…Learned beliefs about the world and about oneself act as top-down predictions explaining sensory signals that pass prediction errors up the neuronal hierarchy ( Friston, 2010 ). It has also been proposed that the top-down dynamics of generative models detached from sensory or task-specific signals is closely related to the spontaneous activity in brain networks during resting state ( Pezzulo et al, 2021 ). Importantly, the DMN is thought to lie at the top of this processing hierarchy involved in generating and retrieving the most complex and context-dependent schemas of various aspects of the self and the external environment ( Friston, 2010 , Igelstrom and Graziano, 2017 , Yeshurun et al, 2021 ).…”

The evolution of brain architectures for predictive coding and active inference

“…Second, it is possible to define a priori which problems can or cannot be addressed by using different kinds of generative models. This implies that the generative model of each particular animal is tightly constrained by the statistics of their ecological niches and the control demands of their bodies [22,23].…”