Unsupervised classification reveals consistency and degeneracy in neural network patterns of emotion

Using deep generative models for simultaneous representational and predictive modeling of brain and behavior: A graded unsupervised-to-supervised modeling framework

This paper uses a generative neural network architecture combining unsupervised (generative) and supervised (discriminative) models with a model comparison strategy to evaluate assumptions about the mappings between brain states and behavior. Most modeling in cognitive neuroscience publications assume a one-to-one brain-behavior relationship that is linear, but never test these assumptions or the consequences of violating them. We systematically varied these assumptions using simulations of four ground-truth brain-behavior mappings that involve progressively more complex relationships, ranging from one-to-one linear mappings to many-to-one nonlinear mappings. We then applied our Variational AutoEncoder-Classifier framework to the simulations to show how it accurately captured diverse brain-behavior mappings,provided evidence regarding which assumptions are supported by the data, and illustrated the problems that arise when assumptions are violated. This integrated approach offers a reliable foundation for cognitive neuroscience to effectively model complex neural and behavioral processes, allowing more justified conclusions about the nature of brain-behavior mappings.

Naturalistic multimodal emotion data with deep learning can advance the theoretical understanding of emotion

What are emotions? Despite being a century-old question, emotion scientists have yet to agree on what emotions exactly are. Emotions are diversely conceptualised as innate responses (evolutionary view), mental constructs (constructivist view), cognitive evaluations (appraisal view), or self-organising states (dynamical systems view). This enduring fragmentation likely stems from the limitations of traditional research methods, which often adopt narrow methodological approaches. Methods from artificial intelligence (AI), particularly those leveraging big data and deep learning, offer promising approaches for overcoming these limitations. By integrating data from multimodal markers of emotion, including subjective experiences, contextual factors, brain-bodily physiological signals and expressive behaviours, deep learning algorithms can uncover and map their complex relationships within multidimensional spaces. This multimodal emotion framework has the potential to provide novel, nuanced insights into long-standing questions, such as whether emotion categories are innate or learned and whether emotions exhibit coherence or degeneracy, thereby refining emotion theories. Significant challenges remain, particularly in obtaining comprehensive naturalistic multimodal emotion data, highlighting the need for advances in synchronous measurement of naturalistic multimodal emotion.

Neural Predictors of Fear Depend on the Situation

The extent to which neural representations of fear experience depend on or generalize across the situational context has remained unclear. We systematically manipulated variation within and across three distinct fear-evocative situations including fear of heights, spiders, and social threats. Participants (n = 21; 10 females and 11 males) viewed ∼20 s clips depicting spiders, heights, or social encounters and rated fear after each video. Searchlight multivoxel pattern analysis was used to identify whether and which brain regions carry information that predicts fear experience and the degree to which the fear-predictive neural codes in these areas depend on or generalize across the situations. The overwhelming majority of brain regions carrying information about fear did so in a situation-dependent manner. These findings suggest that local neural representations of fear experience are unlikely to involve a singular pattern but rather a collection of multiple heterogeneous brain states.

The degenerate coding of psychometric profiles through functional connectivity archetypes

Introduction

Degeneracy in the brain-behavior code refers to the brain's ability to utilize different neural configurations to support similar functions, reflecting its adaptability and robustness. This study aims to explore degeneracy by investigating the non-linear associations between psychometric profiles and resting-state functional connectivity (RSFC).

Methods

The study analyzed RSFC data from 500 subjects to uncover the underlying neural configurations associated with various psychometric outcomes. Self-organized maps (SOM), a type of unsupervised machine learning algorithm, were employed to cluster the RSFC data. And identify distinct archetypal connectivity profiles characterized by unique within- and between-network connectivity patterns.

Results

The clustering analysis using SOM revealed several distinct archetypal connectivity profiles within the RSFC data. Each archetype exhibited unique connectivity patterns that correlated with various cognitive, physical, and socioemotional outcomes. Notably, the interaction between different SOM dimensions was significantly associated with specific psychometric profiles.

Discussion

This study underscores the complexity of brain-behavior interactions and the brain's capacity for degeneracy, where different neural configurations can lead to similar behavioral outcomes. These findings highlight the existence of multiple brain architectures capable of producing similar behavioral outcomes, illustrating the concept of neural degeneracy, and advance our understanding of neural degeneracy and its implications for cognitive and emotional health.

Complex Emotion Dynamics Contribute to the Prediction of Depression: A Machine Learning and Time Series Feature Extraction Approach

Emotion dynamics have demonstrated mixed ability to predict depressive symptoms and outperform traditional metrics like the mean and standard deviation of emotion reports. Here, we expand the types of emotion dynamic features used in prior work and apply a machine learning algorithm to predict depression symptoms. We obtained seven ecological momentary assessment (EMA) studies from previous work on depression and emotion dynamics (N = 890). These studies measured self-reported sadness, positive affect, and negative affect 5 to 10 times per day for 7 to 21 days (schedule varied across studies). These data were fed through a feature extraction routine to generate hundreds of emotion dynamic features. A gradient boosting machine (GBM) using all available emotion dynamics features was the best of all models assessed. This model's out-of-sample prediction (R 2 pred) for depression severity ranged from .20 to .44 depending on EMA interpolation method and samples included in the analysis. It also explained significantly more variance than a benchmark model of individuals' mean emotion ratings over the assessment period, R 2 pred = .089. Comprehensive feature mining of emotion dynamics obtained during EMA may be necessary to identify processes that predict depression symptoms beyond mean emotion ratings.

Fear-related psychophysiological patterns are situation and individual dependent: A Bayesian model comparison approach

Is there a universal mapping of physiology to emotion, or do these mappings vary substantially by person or situation? Psychologists, philosophers, and neuroscientists have debated this question for decades. Most previous studies have focused on differentiating emotions on the basis of accompanying autonomic responses using analytical approaches that often assume within-category homogeneity. In the present study, we took an alternative approach to this question. We determined the extent to which the relationship between subjective experience and autonomic reactivity generalizes across, or depends upon, the individual and situation for instances of a single emotion category, specifically, fear. Electrodermal activity and cardiac activity-two autonomic measures that are often assumed to show robust relationships with instances of fear-were recorded while participants reported fear experience in response to dozens of fear-evoking videos related to three distinct situations: spiders, heights, and social encounters. We formally translated assumptions from diverse theoretical models into a common framework for model comparison analyses. Results exceedingly favored a model that assumed situation-dependency in the relationship between fear experience and autonomic reactivity, with subject variance also significant but constrained by situation. Models that assumed generalization across situations and/or individuals performed much worse by comparison. These results call into question the assumption of generalizability of autonomic-subjective mappings across instances of fear, as required in translational research from nonhuman animals to humans, and advance a situated approach to understanding the autonomic correlates of fear experience. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Multimodal, Idiographic Ambulatory Sensing Will Transform our Understanding of Emotion

Emotions are inherently complex - situated inside the brain while being influenced by conditions inside the body and outside in the world - resulting in substantial variation in experience. Most studies, however, are not designed to sufficiently sample this variation. In this paper, we discuss what could be discovered if emotion were systematically studied within persons 'in the wild', using biologically-triggered experience sampling: a multimodal and deeply idiographic approach to ambulatory sensing that links body and mind across contexts and over time. We outline the rationale for this approach, discuss challenges to its implementation and widespread adoption, and set out opportunities for innovation afforded by emerging technologies. Implementing these innovations will enrich method and theory at the frontier of affective science, propelling the contextually situated study of emotion into the future.

Anxious individuals shift emotion control from lateral frontal pole to dorsolateral prefrontal cortex

Anxious individuals consistently fail in controlling emotional behavior, leading to excessive avoidance, a trait that prevents learning through exposure. Although the origin of this failure is unclear, one candidate system involves control of emotional actions, coordinated through lateral frontopolar cortex (FPl) via amygdala and sensorimotor connections. Using structural, functional, and neurochemical evidence, we show how FPl-based emotional action control fails in highly-anxious individuals. Their FPl is overexcitable, as indexed by GABA/glutamate ratio at rest, and receives stronger amygdalofugal projections than non-anxious male participants. Yet, high-anxious individuals fail to recruit FPl during emotional action control, relying instead on dorsolateral and medial prefrontal areas. This functional anatomical shift is proportional to FPl excitability and amygdalofugal projections strength. The findings characterize circuit-level vulnerabilities in anxious individuals, showing that even mild emotional challenges can saturate FPl neural range, leading to a neural bottleneck in the control of emotional action tendencies.

Coherence and divergence in autonomic-subjective affective space

A central tenet of many theories of emotion is that emotional states are accompanied by distinct patterns of autonomic activity. However, experimental studies of coherence between subjective and autonomic responses during emotional states provide little evidence of coherence. Crucially, previous studies investigating coherence have either adopted univariate approaches or made limited use of multivariate analytic approaches by investigating subjective and autonomic responses separately. The current study addressed this question using a multivariate dimensional approach to build a common autonomic-subjective affective space incorporating subjective responses and three different autonomic signals (heart rate, skin conductance response, and pupil diameter), measured during an emotion-inducing task, in 51 participants. Results showed that autonomic and subjective responses could be adequately described in a two-dimensional affective space. The first dimension included contributions from subjective and autonomic responses, indicating coherence, while contributions to the second dimension were almost exclusively of autonomic covariance. Thus, while there was a degree of coherence between autonomic and subjective emotional responses, there was substantial structure in autonomic responses that did not covary with subjective emotional experience. This study, therefore, contributes new insights into the relationship between subjective and autonomic emotional responses, and provides a framework for future multimodal emotion research, enabling both hypothesis- and data-driven testing.

Improving the study of brain-behavior relationships by revisiting basic assumptions

Neuroimaging research has been at the forefront of concerns regarding the failure of experimental findings to replicate. In the study of brain-behavior relationships, past failures to find replicable and robust effects have been attributed to methodological shortcomings. Methodological rigor is important, but there are other overlooked possibilities: most published studies share three foundational assumptions, often implicitly, that may be faulty. In this paper, we consider the empirical evidence from human brain imaging and the study of non-human animals that calls each foundational assumption into question. We then consider the opportunities for a robust science of brain-behavior relationships that await if scientists ground their research efforts in revised assumptions supported by current empirical evidence.