Predictive coding (PC) is a neurocognitive concept, according to which, the brain processes not the whole external information, but only residual errors emerging from comparisons of the incoming information flow to an individual inner model of the world, minimizing thus the free energy or brain entropy. Many electroencephalogram (EEG) studies using EEG event- related potentials (ERPs) and/or event-related oscillations (EROs) have demonstrated that both the earliest (gamma) and the later (alpha and theta) brain oscillations can predict the fashion of error, mismatch and emotional processing unconsciously. Similarly, several other empirical data have shown that specific changes in slow (less than 0.05 Hz) fluctuations in brain activation have determined unconsciously lapses in attention, decision making and error occurrence, with these slow brain fluctuations being found different and affecting behavior dissimilarly, depending on presence or absence of psychiatric disorders.
Collectively, the above observations point to a manifestation of different aspects of PC in relation to the type of cognitive processing during normal and psychopathological conditions. Importantly, they question (1) to what extent neurobehavioral functions depend on our free will or are determined by individual characteristics of the inner model of the world, and (2) what is the role for PC in psychopathological conditions. Addressing these questions could be approached by investigating ERPs and EROs in response to sensory stimuli across different conscious states, in which top-down executive-control mechanisms are reduced, totally inhibited, or qualitatively altered. Such states are best represented by different stages of sleep, anesthesia, coma, and brain and mental diseases.
It still remains poorly understood whether and how brain states differing dramatically in top-down and bottom-up mechanisms, neurochemistry, synaptic connectivity, and neuroelectric signaling such as sleep stages affect deviance detection (mismatch). If inherent, free of executive-control prediction, mismatch signals can be generated, yet the extent to which the dynamic Bayesian predictions are moderated by consciousness remains unknown. For instance, cortico-subcortical loops involved in emotion generation are more active during rapid eye movement (REM) sleep compared with wake and non-REM sleep, whereas executive-control and primary-sensory cortices are suppressed. In non-REM sleep, the activity of all brain areas relevant for information processing during wake is mostly suppressed. Interestingly, although inconsistent, there is some evidence that in non-REM and REM sleep, an analog of the mismatch emerges despite the gross suppression of both executive top-down processing and external input transmission. Furthermore, some studies have indicated that the mismatch analog during non-REM and REM sleep does not undergo extinction in the course of stimulation. Rather, each stimulus is processed as a novelty.
Do these observations reflect inherent virtual reality of the world during sleep dreaming states, which may incorporate previously encoded memories into a broader vital context comprising residuals of hypotheses testing, emotions, basic needs, and individual genetic traits, optimizing thus the PC for the succeeding wakefulness? If so, could we manage our behavior by navigating sleep dreaming states? Targeting these questions by investigating brain oscillations during sleep in response to stimulation could better explain various cognitive functions of sleep.
Predictive coding (PC) is a neurocognitive concept, according to which, the brain processes not the whole external information, but only residual errors emerging from comparisons of the incoming information flow to an individual inner model of the world, minimizing thus the free energy or brain entropy. Many electroencephalogram (EEG) studies using EEG event- related potentials (ERPs) and/or event-related oscillations (EROs) have demonstrated that both the earliest (gamma) and the later (alpha and theta) brain oscillations can predict the fashion of error, mismatch and emotional processing unconsciously. Similarly, several other empirical data have shown that specific changes in slow (less than 0.05 Hz) fluctuations in brain activation have determined unconsciously lapses in attention, decision making and error occurrence, with these slow brain fluctuations being found different and affecting behavior dissimilarly, depending on presence or absence of psychiatric disorders.
Collectively, the above observations point to a manifestation of different aspects of PC in relation to the type of cognitive processing during normal and psychopathological conditions. Importantly, they question (1) to what extent neurobehavioral functions depend on our free will or are determined by individual characteristics of the inner model of the world, and (2) what is the role for PC in psychopathological conditions. Addressing these questions could be approached by investigating ERPs and EROs in response to sensory stimuli across different conscious states, in which top-down executive-control mechanisms are reduced, totally inhibited, or qualitatively altered. Such states are best represented by different stages of sleep, anesthesia, coma, and brain and mental diseases.
It still remains poorly understood whether and how brain states differing dramatically in top-down and bottom-up mechanisms, neurochemistry, synaptic connectivity, and neuroelectric signaling such as sleep stages affect deviance detection (mismatch). If inherent, free of executive-control prediction, mismatch signals can be generated, yet the extent to which the dynamic Bayesian predictions are moderated by consciousness remains unknown. For instance, cortico-subcortical loops involved in emotion generation are more active during rapid eye movement (REM) sleep compared with wake and non-REM sleep, whereas executive-control and primary-sensory cortices are suppressed. In non-REM sleep, the activity of all brain areas relevant for information processing during wake is mostly suppressed. Interestingly, although inconsistent, there is some evidence that in non-REM and REM sleep, an analog of the mismatch emerges despite the gross suppression of both executive top-down processing and external input transmission. Furthermore, some studies have indicated that the mismatch analog during non-REM and REM sleep does not undergo extinction in the course of stimulation. Rather, each stimulus is processed as a novelty.
Do these observations reflect inherent virtual reality of the world during sleep dreaming states, which may incorporate previously encoded memories into a broader vital context comprising residuals of hypotheses testing, emotions, basic needs, and individual genetic traits, optimizing thus the PC for the succeeding wakefulness? If so, could we manage our behavior by navigating sleep dreaming states? Targeting these questions by investigating brain oscillations during sleep in response to stimulation could better explain various cognitive functions of sleep.