Our research is funded to great extent by an ERC Starting
Predispositions of conscious perception (Windows to consciousness) Our team is investigating which brain states predispose conscious perception within and across multiple modalities. Traditionally, research regarding perception has focused on the neural activity following the presentation of a sensory stimulus. For instance, to investigate the processing of auditory sounds, a conventional analysis would consist of the averaging of electrophysiological signals across trials, which reveals neural activity caused by the auditory sound (so called 'time-locked'), while minimising signals that are not time-locked. In this conventional approach, neural activity prior to stimulus onset is regarded as background noise without any relevance to stimulus processing. Studies from the last decade, however, have shown that such pre-stimulus neural activity (also called brain states) predispose perception, i.e. whether an upcoming weak sensory stimulus (e.g. a very faint auditory sound) is consciously perceived or not.
These brain states are the focus of our research. Recently, we have published a framework regarding conscious perception, called the 'Windows to Consciousness' framework (Win2Con). In this framework, we argue that conscious perception depends firstly on the pre-stimulus cortical excitability or sensory regions, and secondly, on functional connectivity between these sensory regions and higher-order areas. For instance, it is more likely to perceive a weak auditory sound, if the auditory cortex is closer to firing threshold, and if it is already ‘communicating with’ higher-order regions in the frontal and parietal cortex necessary for conscious perception. We specifically aim to address several questions: a) How can brain states (particularly functional connectivity) predisposing conscious perception be described? b) How does pre-stimulus functional connectivity between sensory regions and higher-order areas influence perception? Is there a neural signature underlying conscious perception across all modalities, or is brain state dependent perception modality-specific?
To tackle these questions, we use magneto- and electroencephalography (M-/EEG) and analyse pre- and post-stimulus oscillatory neural activity. For the analysis of functional connectivity, we use a graph theoretical approach (also used in the analysis of e.g. social networks) and dynamical causal modelling (DCM) to infer effective connectivity of brain networks. To address the question whether there are brain states underlying conscious perception common for all modalities (audition, vision, tactile perception), we use advanced decoding tools from machine learning framework such a multivariate pattern analysis (MVPA). By decoding brain activity patterns which could predict trial-by-trial variation of the perceptual subjective report from one sensory modality to another, we aim to reveal brain states supporting conscious perception independent from sensory modality.
Neurophysiological processes during transcranial electrical stimulation
Oscillations are suggested to be the brains’ tool to communicate within and across regions and to code from bits of information up to full objects. Recently, these oscillations have been target of studies using non-invasive brain stimulation, to inject external rhythms into the brain to modulate internal rhythms and to test if a subsequent modulation of cognition is possible. One method is transcranial alternating current stimulation (tACS), which is thought to force intrinsic oscillations to adapt to an injected frequency (often called entrainment). This way, different parameters of brain oscillations (e.g., amplitude, phase, frequency) become the independent variable and behavioural measures the dependent variable, which in turn allows for causal interpretations.
Using the novel combination of MEG and tACS (Fig. a), we have demonstrated that meaningful modulations of brain oscillations during tACS can be recovered (Fig. b). This enables future tACS experiments to deliver direct physiological insights to further the understanding of the contribution of brain oscillations to cognition and behaviour. Within the Win2Con project, this opens up the possibility to shape pre-stimulus neural activity and probe behavioural consequences of the stimulation.
Analysis of ongoing brain state networks using graph theory
The brain itself is a complex dynamic network of functionally interconnected neurons, therefore graph theoretical measures can be calculated to describe its architecture. We investigate ongoing brain state activity using graph theory offering rich descriptive measures of global and local network characteristics such as clustering coefficients, density, path length, smallworldedness, etc. These measures give the most interesting insights into the functional network properties that go beyond classical connectivity. Most commonly, graph theory is applied to resting state data collapsing long time periods; also stimulation-free meditation data (see figure 1) can be anaylzed accordingly. Recently, analyzing steady-state data we calculate graph theoretical measures in a time-frequency resolved manner averaged over short but many sensory stimulation episodes (see figure 2), enabling us to track temporally fine grained network changes.
In close future we will apply this kind of analysis on MEG resting state data acquired before and after experimental MEG measurements of healthy participants (also repeated measurements of single individuals) as well as on high-density EEG resting state data of epilepsy patients prior to surgery. Among others, we will pursue the following questions:
- do resting state networks change before and after an experimental stimulation
- do resting state networks stay robust across time within subjects
- what are the predictions of resection in epilepsy patients given their resting state networks
Neural correlates of tinnitus
Which neurophysiological processes cause the abnormal perception of a phantom sound, experienced by approximately 5-10% of the population. Our group was the first to identify changes in the
MEG resting-state patterns of individuals with chronic tinnitus, in particular reduced alpha activity in temporal regions. Conceptually we have linked this phenomenon to notions associating alpha
to reduced inhibition (according to the inhibition frameworks forwarded e.g. by groups such as Klimesch or Jensen) as well as altered network-level integration. Our current efforts focus on graph
theoretical investigations of tinnitus-related brain activity as well as probing online neurostimulation effects in individuals with tinnitus.