More on Networks

Brain Networks

As a cognitive neuroscientist, I am mainly interested in brain networks, that is, how different regions in the brain interact. Hereby, we can distinguish different kinds of connectivity: the structural, functional and effective connectivity. The first describes brain networks based on their actual excisting fibre tracts between brain areas. The second includes all networks that are connect regions by some functional properties. It could be for example, that different brain areas show similar activation patterns for the same event (i.e. correlations). And the third connectivity type is directed, in the sense that it plays a role which region exerts its causal influence over the other. Effective connectivity is exaclty that what DCM wants to measure. Looking at certain brain regions, it wants to know how much they influence eachother. In other words, it tries to find the weights of the brain network of interest.

fMRI Networks

One way of retrieving these kind of networks is with using fMRI data. FMRI yields brain imaging data that can be used to define regions and extract their brain signal. This signal does not reflect the real neural activity but a BOLD response, which is the related blood-oxygenated level dependent reaction to neural activation. A benefit of the BOLD response is that it can be measured by fMRI scanners, making it possible to tract changes in that signal during an experiment. Due to the sensitivity of the scanner to movement and other sort of artifacts, it is most common to limit the whole experiment to simply presenting pictures to the participants. These stimuli are then considered as visual input to certain brain regions that get activated accordingly. Subsequently, they will now transfer their received information to other regions of the brain, constituting the brain network. Depending on the input, the network changes differently. Thus we can try to understand how certain stimuli change certain connections while others do not. Unfortunatly, we do not have access to the real changes occuring in these networks, as we only get the indirect BOLD measures. Thus we try to “infer” (predict) these changes depending on our fMRI data.