Somewhere along the course of evolution, and at some time in any one of us on the way from zygote to adult, some forms of detection became beliefs, and some tropisms turned into deliberate desires. Two transitions are involved: from functional responses to intentional ones, and from non-conscious processes to conscious ones that presuppose language and are powered by neocortical resources. Unconscious and functional mental processes remain and constitute an "intuitive" system that collaborates uneasily with the conscious intentionality of the "analytic" system. Emotions bridge these divides: in particular, specific feelings affect inference, cognition and metacognition. In what follows, after a brief reminder of the crucial role of emotions to rational thought and action in general, I first look at how fear affects belief. I then narrow my focus to some examples of what I shall refer to as epistemic feelings. These include specialized variants of fear and greed; and feelings of doubt, certainty, knowing and familiarity. I shall also describe some surprising recent findings about the influence of oxytocin on trust and about the direct influence of social conformity on perception and belief.

Interdisciplinary approaches to emotions hold a tension which we know already from other discourses such as the debate about free will or the problem of phenomenal qualities - the tension between the perspective of the observer and the personal point of view (i.e., the participant perspective). While we know, from our own experience, some portion of the richness which emotional experiences are capable of affording, we can partake in experiences and feelings of others through narratives. By contrast, insights into the neuronal mechanisms forming the basis of emotional processes do not deliver insight into the corresponding feelings. So, one might think: the more accurate neuroscientific descriptions of the basis of emotions are, the more we lose track of what we esteem most in our emotional life - the multi-layered and fine-grained nature of experience itself. But it need not be like this.

Emotions are an essential part of human life, but clearly they are more than we subjectively experience. Neuroscience has tried to explore and disentangle the underlying neural correlates of emotions. A brief overview of the historical, theoretical, and conceptual background aids the understanding of contemporary neuroscientific findings. From the early pioneering research of James Papez and Paul McLean, the review swiftly moves on to contemporary neuroscience research. A summary of the findings from several meta-analyses outlines the complex nature of emotional wiring in the brain. Current findings on the anatomy of emotions are reviewed focussing on the following structures: the amygdala, the insula, the nucleus accumbens, the prefrontal cortex (including the orbitofrontal cortex), and the anterior cingulate cortex. Findings on the various different functions of these structures, also including their involvement in psychopathology, are presented before concluding on the future prospects of neuroscience research.

Human emotion, considered in its structure and functionality, can serve as a means of communication, interaction and cooperation in robots with humans. Emotions have to be analyzed as both intrasubjective and intersubjective processes in order to construct androids as human-like robots. Based on a cognitive modeling of emotions and the description of a "language of mind" we developed a methodological tool for the representation of situations and the elicitation conditions of emotions. The reconstruction of human emotions along these lines will open new perspectives for the construction of android emotions which more than just imitate the surface structures of stable pictures of human emotions. Already single emotions (and the more sequences of different emotions or emotions mutually reacting on the other's emotion) are time-dependent series of mental events which are, as processes, embedded in changing situations. Robot emotions have to be appropriate to this generation of human emotions in order to achieve a parallelized functionality which can give rise to the possibility of interaction, communication and cooperation between androids and humans. So the task will be twofold: (i) construction of an android mental system for representing situations and coping with them by emotions, and (ii) testing these new abilities in model worlds that are manageable and calculable in order to solve the hard problem of situation representation for androids.