Spencer K. Lynn

In theories of embodied cognition, concept integration, or blending , integrates features from different source concepts to construct a new, context-sensitive meaning and is considered a core mechanism of human conceptual flexibility. Blending, while a cognitive act, is reflected in language, including that produced by large language models (LLMs). The seeming naturalness of blends produced by language models raises a question about underlying mechanisms of blending in brains versus models, explored in this conceptual analysis. For humans, blending is related to the embodied nature of cognition: we are constrained to perceive the environment in terms of affordances that may achieve our goals. Blending combines perceived affordances into a new concept that can satisfy the goal. I develop a predictive processing account of human generative-causal blends in which goals are parent nodes in a generative hierarchy, predicting affordances that influence perception and categorization. The account is illustrated with a toy example implemented as a vector symbolic architecture in a Python script. In contrast, language models are optimized for next-token prediction and produce distributionally novel blends constrained by proxy-goals that are derived from alignment and prompt framing within the context window. I discuss distinctions bearing on conceptual flexibility, agent autonomy, and their implications for agent design.

Objective. Depression is a prevalent mental health disorder that significantly impacts well-being and quality of life. This study investigates the relationship between depression and cardiovascular function, exploring time-series features derived from electrocardiogram (ECG) and photoplethysmogram (PPG) data as potential biomarkers for depression prescreening. Approach. As part of a comprehensive psycholinguistic experiment, we collected data from 60 individuals, including both healthy participants and those with varying levels of depression, assessed using the Beck Depression Inventory-II (BDI-II) and the Patient Health Questionnaire-9 (PHQ-9). Bimodal features derived from both ECG and PPG data were used to develop machine learning models for depression risk classification, employing classifiers such as random forest, XGBoost, logistic regression, and support vector machines (SVMs). Additionally, regression models were built to predict depression severity based on ECG- and PPG-derived biomarkers. Main results. Key findings indicate that short-term variability (SD1) features in the ECG RR interval, peripheral systolic and diastolic phases from the PPG, and pulse duration significantly differ between healthy individuals and those at risk of depression. SVM achieved the best classification performance, with an area under the ROC curve of 0.83 ± 0.11 for BDI-II-based classification and 0.78 ± 0.11 for PHQ-9-based classification. SHapley Additive exPlanations analysis consistently identified systolic-SD1 and RR-SD1 as key predictors. Regression analysis further supported the role of cardiovascular features in assessing depression severity, yielding a mean absolute error of 10.18 for BDI-II and 5.27 for PHQ-9 score regression. Significance. This study demonstrates the feasibility of using wearable ECG and PPG technologies for depression prescreening. The findings suggest that cardiac activity-based biomarkers can contribute to the development of cost-effective, objective, and non-invasive tools for mental health assessment, complementing traditional diagnostic methods.