UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Understanding the geometry of emotion Lindsay, Adrian John

Abstract

A large body of experimental evidence across humans, primates, and rodents implicates the anterior cingulate cortex (ACC) as a component of emotional cognition. Studies of neural correlates have shown a wide variety of information being represented, including behaviour and emotional state, and previous work suggests that ACC constructs combined, complex representations of context from this variety of information, which is passed on to downstream areas concerned with emotional or autonomic state. Previous analyses in our lab have demonstrated that emotional state plays a role in this overall context, and that other representations such as behaviour are likely modulated by this emotional context. To tease apart these representations, describe their properties, and investigate how the contextualisation of representations by emotional state is functionally achieved we built and adapted a set of machine learning tools and analyses and applied it to our experimental paradigm of conditioned emotionally relevant outcomes, the ‘3-valence task’. We first built and applied HUB-DT, and unsupervised behavioural discovery tool, and generated a complete description of behaviour during our experimental task. Then, we utilised these complete sets of behavioural labels in conjunction with ACC recordings to disambiguate behaviour and emotional state in ACC, and quantify how these representations were related. Finally, we built and applied a novel neural network analysis approach, called the Neural Trajectory Point Net (NTPN) to extract the geometry of representations, their features and differences, and to gain an understanding of how their representational geometry is informed by their functional relationship. The results suggest that context representations in ACC are organized primarily by emotional state. Representations of behaviour, and likely other factors, are geometrically stable but are modulated or transformed by the influence of emotional state to form a combined context signal. These combined, transformed representations likely form the output of ACC to downstream emotional and autonomic brain areas, and support its proposed role in modulating the emotional response to context and informing the process of emotionally motivated decision making.

Item Citations and Data

Rights

Attribution-NonCommercial-NoDerivatives 4.0 International