"Recent developments of high-field functional magnetic resonance imaging (fMRI) have advanced the level of functional detail to sub-millimetre spatial resolution. This is of critical importance because in the mammalian cortex, small functional cortical patches appear to constitute fundamental units of brain function. These functional units are often organized as ""cortical columns"" that contain clusters of neurons with similar functional preferences. The present project will investigate what ""features"" are coded by functional “columnar-level” units in the visual cortex, how represented entities can be decoded from distributed activity patterns, and how modelled intra- and inter-areal connections between feature representations enable neuro-cognitive computations. The research of this project strives towards a new level of insight in the functional organization of the human brain: Instead of describing observed fMRI activity at the level of specialized brain areas, the focus will be shifted towards the content coded within brain regions. The project investigates columnar-level coding in three cross-fertilising sub-projects. In the first sub-project, sophisticated experimental designs, ultra high-field fMRI and advanced data analyses will be combined to unravel columnar-level feature representations and the entities represented by distributed patterns at different levels of the visual hierarchy. In the second sub-project, a large-scale neural network model will be developed with the major goal to integrate measured columnar-level representations in a new theory of invariant object recognition and visual attention. In the third sub-project, high-resolution Brain Computer Interfaces (hr-BCIs) will be created that are based on information extracted from columnar-level representations. The hr-BCIs will implement highly content-specific neurofeedback tools for therapeutic treatment, and advanced communication devices for patients with severe motor impairments."