Guiding Questions

Every day, our brains cycle through different states of awareness: from the rich conscious experiences during wakefulness to dreamless sleep to the bizarre experiences during dreaming sleep. The neural basis of conscious awareness has intrigued scientists and laymen alike for centuries, and has been intensely studied in the past few decades. Yet, it remains one of the lasting mysteries in neuroscience. Because distorted awareness of both external and internal events characterize many neuropsychiatric illnesses, research on this topic has potentially wide clinical and societal implications.

Perceptual Awareness

Sensory inputs may be processed by the brain unconsciously (as in “subliminal priming”), or consciously—giving rise to our conscious awareness of surrounding environment (termed “perceptual awareness”).  We seek to elucidate the fundamental differences between conscious and unconscious processing in the brain (e.g., PCB 2017), and how brain activity gives rise to various contents of perceptual awareness (e.g., Nat. Commun. 2021). These are questions that are uniquely addressable in humans given the availability of subjective report. In our pursuit of neural mechanisms underlying perceptual awareness, we believe that simply mapping “where” and “when” is insufficient (TiCS 2018). By combining multimodal neuroimaging in humans with computational modeling and brain stimulation (e.g., J Neurosci 2015), we aim to illuminate the spatiotemporal evolution of neural processes giving rise to perceptual awareness, uncover their underlying principles, and explain the rich and varied phenomenology we experience in our daily life.

Priors in Perception

Preexisting brain states have enormous influences on conscious perception. Depending on the preexisting brain state at the time of stimulus arrival, a physically identical stimulus may be consciously perceived or not, a visual object may be consciously recognized or not, and we may perceive something that is not out there. Preexisting brain states include both anatomical connections shaped by past experiences and the moment-to-moment fluctuations in spontaneous brain activity that may reflect cognitive or non-cognitive processes. Using a powerful one-shot perceptual learning paradigm, we have been dissecting the neural network and dynamical mechanisms that allow past experiences to powerfully shape our perception (eLife 2018, eLife 2019, J. Neurosci. 2021, PNAS 2021). Using threshold-level perception (PCB 2017, Nat. Commun. 2019, Nat. Commun. 2021) and bistable perception (PNAS 2013) paradigms, we have been probing the influences of spontaneous brain activity and top-down inferences on conscious perception. For a recent review, see Annu. Rev. Neurosci. 2020.

Forward Prediction

The brain is constantly making predictions about its sensory input: about the cause of current sensory input (“Is that a drift wood or a crocodile?”) and about the state of future sensory input (“Is it coming at me?”; termed “forward prediction”). Making correct forward predictions about environmental input is crucial to survival. Through a novel computational framework, psychophysics, and neuroimaging, we have been addressing how perception subserves forward prediction and its underlying neural computational mechanisms. Our results so far suggest that when processing naturalistic dynamical sensory input, human brain activity contains a weighted integration of past sensory input that directly forms the neural basis of forward prediction (J. Neurosci. 2018), and that this integration is limited by an informational bottleneck (Nat. Commun. 2021). For relevant reviews, see J. Neurosci. 2018 and TiCS 2014. 

Methodologies

We use multimodal human brain imaging, including 3T and 7T fMRI, non-invasive and invasive electrophysiology (e.g., EEG, MEG, ECoG), combined with brain stimulation (e.g., tDCS, TMS) and computational modeling to study the neural bases of human perception.