What is the self? This major philosophical question has been considered too complex to be studied in neuroscience. However, during the last 20 years, researchers from different disciplines within the cognitive neurosciences have started investigating the neural correlates of self-consciousness, using a range of behavioural paradigms and functional imaging techniques. Where as many researchers have studied highorder representational aspects of self-consciousness, such as language (i.e. the use of the self-referential pronoun "I" or "Me"), converging evidence suggests that the most basic sense of self, e.g. the sensation that we possess a stable and coherent representation of our body, that we identify with our body and that we perceive the self as being located within the bodily boundaries (e.g. bodily self-consciousness), is due to integration of bodily signals from different exteroceptive sensory modalities (vision, touch, proprioception). Others have highlighted the importance signals related to internal signals (homeostasis, interoception) for self-consciousness. Using lesion analysis, we could demonstrate a causal link between damage to the right temporo-parietal junction and an altered state of bodily self-consciousness, e.g out-of-body experiences. This was confirmed using functional magnetic resonance imaging, multisensory exteroceptive conflicts and virtual reality: if healthy participants perceived themselves as "disembodied", activity in the same cortical region (i.e. temporo-parietal junction) was systematically modulated. Moreover, we were able to demonstrate in a second line of experiments that interoceptive signals can be used to systematically alter bodily self-consciousness and somatosensory processing, extending previous paradigms that focused exclusively on exteroceptive signals. This was strengthened by our finding that the left posterior insula, repeatedly linked to the representation and integration of internal and external bodily signals, is implicated in another state of altered bodily self-consciousness, e.g. heautoscopy. Our results demonstrate that bodily self-consciousness is based on a neural network that represents and integrates both external and internal signals from the body.