The P3b is a well-established event-related potential (ERP) linked to higher-order cognitive processes such as attention allocation, information processing, and working memory. Research indicates that reductions in P3b amplitude may serve as a neurophysiological marker for substance use disorder (SUD). This study aims to explore the relationship between P3b amplitude and the variety of substances used across an individual’s lifetime among young adults. Specifically, it investigates whether participants who report using a greater number of different substances (i.e., broader history of multi-substance use) display P3b amplitude reduction (P3b-AR) in response to rare, task-relevant stimuli during a rotated heads oddball task. Survey data on lifetime substance use will be analyzed to construct relevant variables and establish comparison groups for EEG analysis. We hypothesize that reduced P3b amplitude will be associated with broader lifetime substance use, indexing elevated risk for addiction. Understanding this relationship may offer valuable insight for early detection and implementing targeted interventions for vulnerable populations, particularly among college-aged young adults.

Data augmentation is a critical technique in Deep Learning (DL) for improving model generalization. However, the combined application of multiple augmentation methods remains underexplored in the context of electroencephalogram (EEG) data. This study investigates how various permutations of Gaussian noise, Time Transformation, and synthetic EEG signal generation using Generative Adversarial Networks (GANs) influence EEG model performance. Using the publicly available PhysioNet EEG Motor Movement/Imagery Dataset (EEGMMIDB), which includes 64-channel recordings across three classes from 109 subjects, the EEG signals were preprocessed and subjected to 16 different permutations of the listed augmentation techniques. These augmented datasets were then used to train a neural network classifier, allowing for comparative analysis of model performance across augmentation strategies. This work will show the impact that stacking augmentation methods have on the accuracy and reliability of classifiers trained on EEG data.

Attention is a finite cognitive resource, particularly for college students who may be frequently overloaded by competing demands and digital distractions. Since Kaplan proposed Attention Restoration Theory (ART), researchers continue to explore how exposure to natural environments could improve and restore directed attention. More recently, studies have used electroencephalography (EEG) to measure frontal alpha-band EEG – a marker of wakeful relaxation and attentional restoration – to examine how viewing natural stimuli affects this activity. However, individuals may differ in their responsiveness to nature-based stimuli processing due to stable personality traits. One such trait may be Absorption, which is the tendency to become deeply immersed in sensory or imaginative experiences. The current study evaluates whether trait Absorption moderates the effect of image type (nature vs. non-nature) on frontal alpha activity during passive viewing. I hypothesize that: 1) frontal alpha power will be greater when participants view images of nature compared to non-nature images, 2) trait Absorption will be significantly associated with overall frontal alpha power, and 3) Absorption will moderate the relationship between image type and frontal alpha activity.