Addressing the generalization problem in neural models of fear

Northeastern University, Boston MA

Investigators

Abstract

Emotions are one of the most important aspects of human cognition. Whether it’s the thrill of sky diving, the amusement of a comedy show, or the serene tranquility of meditation, emotions are often intrinsically desirable states. In the extreme, they can also be undesirable or even debilitating, such as in cases of uncontrollable fear or anxiety. Emotions also play a role in perhaps every aspect of human psychology by influencing people’s memories, perceptions, thoughts, behaviors, and decisions. The search for an understanding of emotions has stimulated research on the representation of emotions in the brain. This research has lead to considerable insight into emotions, but also has lead to some apparently conflicting results: different brain regions were activated for the same emotional state depending on the context, and how scientists studied it. Take fear, for example. Since fear is an ancient emotion, common to animals as well as humans, it was originally studied in animal models. Thus, at first, most of what was understood about the neuroscience of fear was based on studies in non-human animals, particularly rodents, who were placed in threatening circumstances while scientists recorded autonomic responses (e.g. heart rate, sweating) and defensive behaviors (e.g. freezing or escape). This work uncovered a network of activated brain regions centered on a part of the brain (the amygdala), and was initially called a “fear circuit”. Based on these results from animal research, this circuit was also proposed to underlie human subjective experiences or feelings of fear. To discover if this was the case, non-invasive neuroimaging studies using fMRI (functional magnetic resonance imaging) were used to examine brain activity during fear in humans. Of great advantage to this research was that the human subjects could simply tell scientists whether they in fact felt fear in response to certain stimuli. However, the results from these human neuroimaging studies were quite different than the results from animal research. In humans, the brain regions predicting and experiencing fear were widespread and highly distributed throughout the brain. Strikingly, the amygdala and other areas in the “fear circuit” were not activated in these studies. These apparently conflicting results stimulated many new questions. If the brain regions underlying fear are different in humans and animals, and depend on the tasks, measurements, and methods that are used, how can a unified scientific account of fear be developed that generalizes across the particularities of any specific study to the diverse variety of fears in everyday life? One possible explanation of the divergent results could be that the early neuroimaging studies in humans were not precise enough in delineating activation of key brain areas that are important in mediating fear. Components of the “fear circuit” including the amygdala, hypothalamus, and periaqueductal gray are small brain nuclei and each are composed of several subregions. It is at the level of these small brain nuclei that circuits for fear and defensive behaviors are organized. In this project, researchers use a more powerful fMRI scanner, with greater spatial resolution that previous neuroimaging studies in order to examine these brain regions and the pattern of fear-related brain activation in greater detail. Participants are induced to feel fear in many distinct situations using both videos and threat of pain (but no actual pain). High resolution recordings of brain activity, subjective self-reported fear, and autonomic activity are all obtained simultaneously. This rich dataset provides a basis for studying how subjective fear and autonomic activity both relate to activity in all parts of the brain, including the subregions of key subcortical structures in the “fear circuit”. This project also includes analyses to identify brain states where subjective fear and measures of autonomic activity (such as heart rate, galvanic skin response) coincide. These correlations reveal whether autonomic measures can be used as indices for human feelings of fear, and thus provide information about whether functionally homologous findings from studies in non-human animals may generalize to subjective experiences of fear in humans. This may reconcile the results from animal and human research and lead to a deeper understanding of the neural mechanisms of fear, with implications for treatments to alleviate fear and anxiety disorders. Complementing the research project, this project also includes an extensive set of broader impacts including a summer workshop for local high school students from underserved communities on the neuroscience of fear and emotion, and ways of overcoming fear and anxiety. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.