Uncovering hidden brain state dynamics that regulate performance and decision-making during cognition

Jalil Taghia,Tianwen Chen,Srikanth Ryali,Jonathan Nicholas,John Kochalka,Weidong Cai,Vinod Menon

doi:10.1038/s41467-018-04723-6

Abstract

Human cognition is influenced not only by external task demands but also latent mental processes and brain states that change over time. Here, we use novel Bayesian switching dynamical systems algorithm to identify hidden brain states and determine that these states are only weakly aligned with external task conditions. We compute state transition probabilities and demonstrate how dynamic transitions between hidden states allow flexible reconfiguration of functional brain circuits. Crucially, we identify latent transient brain states and dynamic functional circuits that are optimal for cognition and show that failure to engage these states in a timely manner is associated with poorer task performance and weaker decision-making dynamics. We replicate findings in a large sample (N = 122) and reveal a robust link between cognition and flexible latent brain state dynamics. Our study demonstrates the power of switching dynamical systems models for investigating hidden dynamic brain states and functional interactions underlying human cognition.

Highlights

Human cognition is influenced by external task demands and latent mental processes and brain states that change over time
Uncovering hidden brain states and their dynamic spatio-temporal evolution in relation to cognitive task demands remains an important unresolved problem in human cognitive neuroscience research[7]. Progress in tackling this challenge has been limited due to a lack of appropriate computational tools for quantitative characterization of hidden brain states and their dynamic functional properties such as state transition probabilities and time-varying functional connectivity. To address this challenge here we develop and apply novel unsupervised learning procedures based on Bayesian switching linear dynamical systems (BSDS) to identify latent brain states and characterize their dynamic spatiotemporal properties
Using BSDS we uncovered multiple critical spatiotemporal properties of latent brain states, including (i) occupancy rates and mean lifetimes of task dominant and non-dominant brain states, (ii) transition states and their relation to flexible task switching, (iii) probability of transitions across states, (iv) dynamic functional networks associated with distinct brain states, and (v) brain states that best predict cognitive performance and decision-making dynamics

Summary

Introduction

Human cognition is influenced by external task demands and latent mental processes and brain states that change over time. Uncovering hidden brain states and their dynamic spatio-temporal evolution in relation to cognitive task demands remains an important unresolved problem in human cognitive neuroscience research[7] Progress in tackling this challenge has been limited due to a lack of appropriate computational tools for quantitative characterization of hidden brain states and their dynamic functional properties such as state transition probabilities and time-varying functional connectivity. BSDS applies a hidden Markov model (HMM) to latent space variables of the observed data, resulting in a parsimonious model of generators underlying the observed data—this is contrast to previous approaches that have applied HMMs directly to observed MEG27 and resting-state fMRI28,29 data These and other features (Methods) allow BSDS to uncover latent brain states, their temporal evolution, volatility, and persistence over time, probability of transition to other brain states, and non-optimal brain state transitions that impair performance. The temporal evolution of brain states and the covariance structures of each state can be used to extract moment-by-moment connectivity patterns and dynamic functional networks associated with each brain state

Methods

Results

Conclusion