This ERC project consists of three subprojects, which are all completed (we are at the end of the ERC project now).
Theme 1
Goal: validate and refine existing influential theories of consciousness by creating four stages of processing (shallow/deep unconscious, preconscious, conscious) and inspect their underlying neural mechanisms. We have been able to successfully combine manipulations that impair attention (attentional blink) with manipulations that impair perception (masking), and we even matched objective performance between those consciousness manipulations, controlling for common confounds in the field with differences in task performance between conditions. Thereby we isolated four critical stages of neural information processing important for consciousness, measured with electroencephalography (EEG). We designed a novel stimulus set that allowed us to isolate neural processing mediated by feedforward, lateral, and feedback connections. Isolating these different neural processes, was one of the main aims of the entire ERC project. Performing classification analyses on the EEG data for the different stages of information processing (shallow/deep unconscious/preconscious/conscious) allowed us to prove existing theoretical models. This project revealed that the four stage model, which is at the heart of global neuronal workspace theory and local recurrence theory, is correct. Specifically we showed that attentional impairments of conscious access leave local recurrent interactions in visual cortex largely intact, whereas perceptual manipulations do not (they only leave feedforward processing intact). We also demonstrate that how you measure consciousness matters, and is crucial and not theory neutral. In an fMRI study using multivariate decoding techniques we showed that object specific activations in inferior temporal cortex (processing face/house stimuli) is observed when stimuli are masked (objective invisibility, participants never see the stimulus). However, using subjective visibility measures in combination with post hoc sorting of trials, revealed even stronger neural processing of subjectively invisible stimuli. Therefore, the choice how to determine invisibility (objectively versus subjectively) strongly influences the conclusions one may draw about the potential neural correlates of consciousness.
Theme 2
Goal: test two main hypotheses, namely 1) that NMDA receptors play a crucial role in recurrent processing, the dynamic information exchange between brain regions, thought to give rise to consciousness, and 2) the hypothesis that rapid fluctuations in spontaneous network activity (modulating arousal levels), which are controlled by noradrenaline and acetylcholine neuromodulatory systems, determine the likelihood of sensory evoked recurrent processing, and hence consciousness, to occur. Generally, the two main hypotheses were confirmed. We showed that memantine (NMDA antagonist) selectively affects recurrent processing and not feedforward processing, thereby providing causal support of the hypothesis that NMDA mediated recurrent processing is crucial for perceptual inference. Regarding the second hypothesis, we showed that noradrenergic enhancement increased perceptual experience of stimuli in various task settings. Increased conscious perception was accompanied by increased evidence accumulation signals in neural activity measured over parietal cortex. We showed that sensory encoding of stimuli is not changed by noradrenergic enhancement. Although perceptual experience is improved by noradrenergic enhancement, the metacognitive assessment of task performance (introspective insight) is largely unaffected or even impoverished. This suggests that general theories of brain functions that hypothesize that the upregulation of arousal using neuromodulation may increase the efficacy of bottom-up neural processes, but attenuates top-down processes, is likely to be correct. Further we show that the relationship between pupil-linked arousal fluctuations and perceptual performance is inverted U-shaped, with optimal performance at mid-levels of arousal and worse performance at low and high ends of the arousal spectrum.
Theme 3
Goal: testing the hypothesis that recurrent processing provides the possibility for prolonged and flexible information processing, by linking it to cognitive and perceptual functions. Overall, we confirmed the hypotheses that preconscious processing, associated with local recurrent interactions in sensory cortex, is not sufficient for conflict detection, a high-level cognitive control function mediated by the prefrontal cortex. Therefore this study linked neural processes not only to perceptual experience but also to cognition. We also showed that NMDA mediated recurrent processing (feedback) is mainly important for perceptual integration, combining local elements of a single visual stimulus, but less so with perceptual integration across time (integrating maintaining several delay in working memory and comparing them).
